MAS-Based Interaction Simulation within Asymmetric ...

Research ArticleMAS-Based Interaction Simulation within AsymmetricInformation on Emergency Management of UrbanRainstorm Disaster

Qing Yang1 Jinmei Wang2 Xingxing Liu 1 and Jiajia Xia2

1School of Safety Science and Emergency Management Wuhan University of Technology Wuhan 430070 Hubei China2School of Management Wuhan University of Technology Wuhan 430070 Hubei China

Correspondence should be addressed to Xingxing Liu liuxingxingwhuteducn

Received 25 July 2020 Revised 17 September 2020 Accepted 12 October 2020 Published 26 October 2020

Academic Editor Peter Bian

Copyright copy 2020 Qing Yang et al -is is an open access article distributed under the Creative Commons Attribution Licensewhich permits unrestricted use distribution and reproduction in any medium provided the original work is properly cited

-e frequent occurrence of urban waterlogging constantly affects resident living and urban construction Improved adaptiveprevention and control strategies are highly requested due to huge economic losses and casualties caused by flood andwaterlogging in China -e urban waterlogging may evolve into a serious emergency generally characterized by high complexityuncertainty and time pressure Coupled with the asymmetric information waterlogging often exacerbates the impact of urbanrainstorm disasters -rough the multi-agent system simulation with given geographic information government and residentsinteract under dynamic risk distribution in rainstorm disaster -e results show that the proactive attitude of residents and thegovernment towards disaster relief could have a promoting effect for both thereby increasing the disaster relief efficiencyObviously rapid accurate information collection and analysis facilitate disaster relief to a large extent Meanwhile appropriatesupply rather than excessive supply may mobilize residentsrsquo self-help and balance replenishment of relief supplies

1 Introduction

Increasingly frequent and severe rainstorms have causedenormous damage and casualties all around the world [1]Disasters are large intractable problems that test the ability ofnations to protect their populations and infrastructure ef-fectively and recover quickly [2] In addition teamwork ofresidents helps much on disaster relief [3 4] Urban rain-storm usually causes serious waterlogging due to heavyprecipitation [5ndash7] which is one of the most commonnatural disaster phenomena in many big cities in ChinaRainstorm waterlogging induced by torrential rain or ty-phoon has become one of the most frequent and seriousnatural hazards in many big cities around the world es-pecially in the coastal cities of the developing countries thatare the centres of human inhabit and social-economic de-velopment [8]

Waterlogging is a kind of seeper phenomenon due toheavy precipitation or continuous precipitation beyond the

urban drainage capacity which can usually be attributed torainstorms [5ndash7] -ere are many other causes of urbanwaterlogging frequent extreme weather events change inunderlying surface condition artificial occupancy of theriver section reduction in water storage capacity urban heatand rain island effect drainage engineering construction laglow terrain and lag-management [9 10] With globalwarming the emergence of heavy rainfall is becoming morefrequent Accompanied with the accelerating urbanizationmany big cities (such as Beijing Wuhan GuangzhouNanjing Hangzhou Chengdu and so forth) in China havesuffered severe urban waterlogging due to heavy rainfallWaterlogging occurred in the frequency and scale has ex-panded for each year which exposed the vulnerability ofmajor cities in defence of meteorological disasters Rain-storm disaster often brings much adverse impact on healthtraffic communication and peoplersquos property [11ndash14]

On June 23 2011 a rainstorm hit Beijing and causedserious waterlogging and traffic congestion Many people

HindawiComplexityVolume 2020 Article ID 1759370 13 pageshttpsdoiorg10115520201759370

were stuck in the streets and could not get home until late atnight On July 21 2012 a more powerful rainstorm struckBeijing Serious waterlogging occurred in some low-lyingroad sections and some people even drowned in their cars[15] During the disaster 79 people lost their lives about 19million people suffered and the total economic damageswere almost 10 billion RMB [16] -us a series of problemscaused by urban storm waterlogging has become the re-search focus and more attention is paid to it [17]

Disaster management has everlasting attention [18] Inorder to reduce the damage caused by rainstorm disastersensure traffic safety and control emergency cost extensiverainstorm disaster models have been developed to assessdisaster risk and predict the impact of rainstorm disasters aswell [19ndash21]Wang et al [22] used the hiddenMarkovmodel(HMM) method to conduct a dynamic risk assessment ofnatural disasters in Dalian China Javier Salmeron et al [23]developed a two-stage stochastic optimization model toguide the relief assets allocation Mahootchi and Golmo-hammadi [24] extended the mathematical two-stage sto-chastic optimization model to provide appropriate responseand better service Sawada and Takasaki [25] established amicroeconomic analysis framework between disasters andpoverty revealing that mutual aid is critical to effectivepoverty alleviation ldquoLife firstrdquo is often seen as the principalrule in the disaster rescue so as in the rainstorm relief Whilehydrological analysis and facility planning occupied majorresearch position in urban rainstorm evolution like SWMM(Storm Water Management Model) [17] MOUSE (Modelfor Urban Sewers) [26] etc various mutual influencingfactors underlie the complex system of urban drainage [27]Actually owing to the increase of complexity of urbandevelopment the effectiveness and efficiency of modelling issubject to the dynamic parameter optimization [28]

-ere are many tools for simulating [29] human andsocial behaviours in emergencies -e use of these tools canhelp determine the locations of temporary shelter [30] andpredict disaster losses [31] which can improve the efficiencyof disaster response Search and rescue are the major parts ofdisaster response [32] in which agent-based simulation hasbeen proved as a useful method [33] Tang Jet al [34] op-timized search and rescue using auction-based task alloca-tion scheme through agent-based simulation Choi et al [35]developed a flexible and interoperable distributed simulationenvironment for comprehensive disaster response man-agement Alsubaie et al [36] constructed a disaster responseplan based on supervisory control and data acquisition(SCADA) systems Modelling by multi-agent system (MAS)has been a common tendency for the operation of social-economic system under changing natural conditions [37]like evacuation route choice of pedestrians in an urbanrainstorm [38 39]

Effective communication and coordination are crucialaspects of emergency management [40] However infor-mation asymmetry often manifests when one party has moreor better information than the other Agents with moreinformation will have higher expected payoff than those withless [41] Besides agentsrsquo decisions will have potential dis-tortion due to information asymmetry [42]-e difference of

decision among agents therefore has a positive correlationwith information asymmetry [43]

2 Research Method

21 MAS Modelling Agent-based simulation seems to be akind of dynamic networks of interacting agents [44] Byspecifying the rules between individuals computer simu-lations can be used to demonstrate complex behaviourpatterns from other simple models [45] Agent-based sim-ulation is a relatively new approach to modelling systemscomposed of autonomous interacting agents [46] Eachindividual is able to make decisions based on circumstancesand rules [47] Repetitive interactions are hard to implementwith mathematical methods which in contrast is a pecu-liarity of agent-based simulation Agent-based simulation iswidely used in many fields to deal with problems covered incomplex systems [48] like resource allocation [49] crowdevacuation [50] decision-making systems [51] and coop-erative game [52] -e application of agent-based simulationin disaster management can improve search and rescueefficiency [53]

Since the 1990s agent-modelling platforms such asSTARLOGO SWARM NETLOGO REPAST (RecursivePorous Agent Simulation Toolkit) etc have been developedREPAST has outstanding scalability for its JAVA pro-gramming [54] and has already been used in modellingsocial behaviour [55 56] and city management with geo-spatial design [57] REPAST can be redesigned for diversepractical needs Bridging the gap between agent-basedsimulation and actuality is a challenge for MAS modellingwhich is often conducted by parallel execution and adaptivecontrol between artificial and real world [58] through theartificial societies-computational experiments-parallel exe-cution (ACP) platform [59] to interact with the real socialsystem and provide reliable support for themanagement anddecision-making of real social scenes -erefore the evo-lution of the situation co-evolution and closed-loopfeedback are realized

For deepening our previous study on human reaction indisaster we followed the previous MAS modelling onrainstorm relief [38 39] Yang et al analysed the efficiency ofdisaster relief by considering the crowd psychology of res-idents and diverse density distribution of relief settlementHowever the study focuses on the outcome generated fromdiverse pedestrian categories and distribution of shelterswhich does not involve the mutual reaction from diversesubjects

Actually information asymmetry is inevitable in rain-storm the interaction between the government and resi-dents underlies the disaster relief efficiency Exploring theimpact on the disaster relief from different initiatives andinteractions helps implement disaster situation analysis anddisaster relief decision-making reasonably

22 Agent Attributes -e agent is a physical or abstractentity and the basic unit in multi-agent systems (MAS) [60]From a practical modelling standpoint the agent is required

2 Complexity

to possess specific characteristics Here are three basicagents resident government and waterlogging-e settingsof these agents attributes are as follows

221 Residents It is assumed that every resident agent hasthe attributes of direction energy information decisionand reaction from government Energy represents the basicphysiological requirements for residents to survive Whenresidents perform series of activities such as walkingwading etc energy will be reduced Meanwhile residentscan also increase their energy such as government rescueand materials which will increase directly the energy of theresidents saved and the information obtained by the resi-dents will also be converted into energy Residents mustmaintain energy greater than zero during the disasteravoidance process which is boundary for living status -eprobability of residentsrsquo autonomous disaster avoidance isrelated to average expected payoff If disaster avoidance isautonomous residents may increase the probability of beingrescued because they have obtained more disaster infor-mation but they also consume more energycontemporaneously

222 Government In China disaster relief organizationsare a kind of organizations strictly supervised by govern-ment or are simply referred to as government It is assumedevery government agent has attributes of disaster analysisbenefit and decision Government arranges the disasterrelief actions with limited relief supplies according to theinformation of rescue from residents Government dis-patches relief resource and rescue team to set up shelters andspread aid response

(1) Residents and Government Are Rational Men in DisasterResistance Every involved participant is a kind of rationalman who is self-interested and pursues maximum payoff[61] When encountering rainstorm risk residents couldmake a strategic decision to move out for disaster avoidanceautonomously or stay in original place according to indi-vidual experience environmental conditions and disastersituation [62] -e government could arrange the disasterrelief actions or not based on time sequence and reliefsupplies [12 63] -e government has a fixed consumptionof disaster relief supplies for single rescued resident

(2) Residents Have Energy Consumption and Recovery Whenrainstorm occurs the life value [64] of residents will beaffected to various extents [2] Regardless of whether resi-dents behave or not they have basic consumption of energyto sustain life and self-adjusted recovery factor each cycletime If residents get rescued they could get life recoveryfrom relief supplies then energy could be added to a certainrecovery level

(3) Residents Gain Information In a rainstorm residentsneed to obtain relevant information in time to assistthemselves in making decision of disaster avoidance In-formation could represent feasible evacuation route disaster

avoidance approach medical care food and so on Hereinformation is simplified as the directions to temporaryshelter or rescuers If there does not exist temporary shelterinformation is empty Moreover residents have a certainprobability of capturing information and the amount ofinformation obtained by the resident is transformed into atype of energy variable for convenience -e informationresidents own has a diverse value

(4) Direction and Decision of Residents Residents use theavailable information and governmentrsquos reaction to makethe decision strategies to move out for disaster avoidanceautonomously or stay in original place and determine thedirection on the route to rescue

(5) Government Analyses the Disaster -e government is themain body of disaster relief to some extent Not only shouldthey dispatch rescue teams in time and set up disaster reliefshelters but also they must properly analyse the entirerainstorm situation to make more accurate decisions andincrease the probability of successful rescue -e more di-saster information the government release the more in-formation asymmetry can be reduced between agents[65 66]

(6) Benefit of Government Once residents move out fordisaster avoidance autonomously they become disaster-resistant subjects rather than bearers [67 68] Residentsparticipating in disaster relief autonomously will help reducethe losses [69] and the government will also gain corre-sponding benefit from residentsrsquo disaster avoidance au-tonomously [70ndash72]

-e key parameters about the residents and governmentagents are summarized as Table 1

223 Waterlogging In the rainstorm there may be manywaterlogging places in urban area -e setting of water-logging is generally obtained in the simulation based on theneighbourhood analysis in spatial analysis When the sur-rounding terrain of a certain area is higher than that area andthe underground pipe network has insufficient drainageduring heavy rain scenarios waterlogging point will occurAs the rainfall process continues the waterlogging point willgradually spread -ese waterlogging points may lead torisks including traffic congestion landslides leakage ofelectricity or spread of harmful substance Residents maylose much energy since waterlogging points may hindernormal order of life in particular they have to get acrossthem To simulate the scenarios which may happen inrainstorm random test is used to reproduce as manyrainstorm disasters as possible

23 Agent Interaction Rules

231 Interaction Rules between Government and ResidentsResidents are sensitive to habitats where they live whichunderlies evacuation routes -e rescue organization willdeploy disaster relief activities and temporary shelters based

Complexity 3

on disaster situation meteorological data and geographicinformation In general residents can stay in original placewaiting for rescue or search for temporary shelters based ontheir own circumstances and decision-making modes Di-saster relief organizations arrange the rescue activities andsupplies distribution according to distress signals Affectedby factors such as limited relief supplies [73] and rescue timesequence the disaster relief organizations cannot respond toevery affected resident simultaneously in time How toachieve a greater evacuation rate in short time is extremelycritical Cooperative behaviour between residents and di-saster relief organizations within information asymmetrycould have a vital impact on disaster relief In fact thedecisions of both will affect their benefits and loss whichforms an interaction within asymmetric disaster informa-tion between government and residents in rainstorm di-saster management

-e simulation is terminated when all residents moveout for disaster avoidance autonomously or relief suppliesare exhausted Starting points of residents and shelter pointsare not unique since diverse situations of disaster need to besimulated A common disaster relief process is shown inFigure 1 When residents arrive at a shelter or their energy isless than zero their activities are terminated

232 Disaster Relief Rules of Government In the rainstormdisaster the government will set shelters and dispatch rescuegroups according to the calling for help from residents -egovernment will distribute the relief materials needed by theresidents to various rescue groups Meanwhile residents cansearch for the rescue shelters or stay at original place waitingfor help according to their own plight and rescue signalsfrom government Limited by resources for disaster reliefand space-time barrier it cannot be guaranteed that allresidents get satisfied rescue Efficient distributing controland cooperative work really underlie effective disaster re-sistance Furthermore the residentsrsquo resistance for disastercan facilitate disaster relief [74]

Generally rescue teams are dispatched to cover everydisaster area as much as possible When residents encountera rescue team they can get relief resources to recharge theirenergy and information of shelters and risk distribution ofdisaster simultaneously Rescue teams canmake a fixed place

as a shelter to implement allocation of relief for neighbourresidents

In each round of model evolution there are differentrates of government response combined with the previousagents attributes and interaction rules the evolution formulais shown in equation (1)

PGt+1 P

Gt +

PGt lowast 1 minus P

Gt1113872 1113873 E

PGminus E

G1113874 1113875

EPG

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (1)

EPG

1 minus PGt1113872 1113873 P

Rt λA minus αS1113872 1113873 + E

G (2)

PG represents the possibility of rescue groups distributedon the map to start rescue PR represents the possibility ofresidents moving out for disaster avoidance autonomouslyPG

t and PRt represent the probability at certain time t EPG

represents the relief performance of each rescue group inequation (2) E

PG represents average performance EG

represents the disaster relief effectiveness of government-e probability of disaster relief in the next stage of the

government is determined by the probability of disasterrelief in the previous stage and relief performance On theone hand the relief performance of each rescue team isdetermined by the governmentrsquos disaster relief efficiencyand on the other hand it is related to the governmentrsquosoverall disaster relief payoff and loss

233 Disaster Avoidance Rules of Residents If residentsmove out for disaster avoidance autonomously governmentcan get positive feedback from residents like reduction ofrelief demand timely information collection and efficientresource allocation In the simulation process the disasterrelief information researched by government can also helpresidents avoid disasters

During the simulation process the number of partici-pants play a vital role in analysing the law of evolution Inorder to depict positive attitude and activeness of all subjectsinvolved in disaster resistance Entire Reaction (ER) ratiocould be a simple and useful index to evaluate the effects ofdifferent strategies where the calculation formula is shownas follows

ER N minus n

N (3)

N denotes the total of residents n refers to the number ofresidents who stay at original place without receiving gov-ernment assistance Otherwise either residents or govern-ment takes actions for disaster avoidance autonomously andrescue respectively In particular the residents who moveout for disaster avoidance autonomously without receivinggovernment assistance are also attributed to reaction processsince the rescue for mobile individual is an intermediatestatus Residents in search of getting rescue could bringthemselves risk but also opportunity

In each round of model evolution there are differentrates of residents who make decision of disaster avoidanceautonomously combined with the previous agentsrsquo

Table 1 Parameters of resident and government

Agent Parameter Symbol

Resident

Life value (energy) LInformation value IBasic consumption L

Probability of capturing information εRecovery factor βRainstorm risk α

Government

Government benefit QDisaster analysis A

Disaster analysis coefficient λRelief supplies for single resident S

4 Complexity

attributes and interaction rules the evolution formula isshown in equation (4)

PRt+1 P

Rt +

PRt lowast 1 minus P

Rt1113872 1113873 E

PRminus E

R1113872 1113873

EPR

⎡⎢⎢⎣ ⎤⎥⎥⎦ (4)

EPR

1 minus PRt1113872 1113873 αP

Gt minus αβP

Gt minus α1113872 1113873L

+ PGt minus P

Gt ε + ε1113872 1113873I + E

R

(5)

EPR represents the performance of each resident whomoves out for disaster avoidance autonomously in equation(5) E

PR represents average performance ER represents the

disaster relief and anti-disaster effectiveness of residents-e avoiding disaster probability of residents in the next

stage is determined by the avoiding disaster probability ofthe previous stage and avoiding disaster performance -eperformance of residents choosing avoid disaster auto-matically is limited by the residentsrsquo avoiding disaster ef-fectiveness on the one hand and related to the residentsrsquoenergy gains and losses and information acquisition on theother

24 Simulation Area Considering great similarities amongregional plans in Wuhan a typical community (a certainarea) in Wuhan (N 29deg58prime～ 31deg22prime and E 113deg41prime～115deg05prime)Hubei province in China is constructed to provide an il-lustrative disaster relief process Wuhan has over 10 millionpeople in 8569 km2 the biggest city in the central district ofChina Wuhan is also an international city with rapid ur-banization Subtropical monsoon brings much rain toWuhan especially from June to August -e average annual

rainfall in Wuhan was approximately 1200mm from 2017 to2019 -e rainstorm in July 2016 had caused 14 deaths and agreat direct economic loss of 4 billion RMB

-ere are different houses as the starting positions ofresidents and disaster relief points set by the government-e roads are intricate and complicated to set up differentwaterlogging points -e simulation map is shown inFigure 2

-e green circles represent residents who walk along theroad shown by polylines When walking residents maycome across a lot of waterlogging points displayed as redforks (shown in Figure 3 one running screenshot of sim-ulation) Residents may spend some energy and time tocomplete these actions Residents aim at reducing the lossrate of energy and reaching the shelters

Considering the density of houses and spatial distance 4residentsrsquo start points 3 start points of rescue teams and 3shelters are set in different locations Discrete and randominitial settings may evolve into many emerging scenariosStrict settings from practical situation may restrict thepossible evolution processing which is against ACP -epopulation of residents is 50 for reducing computing load

3 Results and Discussion

Comparative analysis of possible situations could help un-derstand the origins Display using geographic informationsystem (GIS) is meaningful for locating disaster of emer-gency rescue -rough agent-modelling simulation theevolutions of diverse situation can be discovered and vi-sualized Table 2 shows the initial value setting of eachparameter

Residentaction

Initialization

Move out

Arrive

Rescued

Wading

Energyrecovery L gt 0

Failure

Yes

Yes No

YesNo

Yes

No

No

No

Terminated

Yes

Governmentaction

Rescue

Figure 1 Common disaster relief process

Complexity 5

-e setting of each parameter is based on the generalsituation and it will change with the interaction processSome parameters do not depend on the initial value (such asrecovery factor) and some parameters are only used formeasurement (such as energy)

From 150 experiments the average energy consumptionof disaster-affected agents is shown in Figure 4 -e averageenergy consumption roughly follows the normal distribu-tion -e mode of average energy consumption is within

1100ndash1300 In the evolution process the rainstorm riskcoefficient α (the ratio of the energy consumption to theenergy) reflecting the main decision interval of a rationalperson is [037 043]

31 8e Best Action of Residents and Government withinAsymmetric Information According to the statistics theenergy consumption conforms to the normal distribu-tion in the evacuation process -e basic life value(energy) of each agent is 1200 close to the mode of thesimulation experiment results Nearly half of the resi-dentsrsquo energy consumption is more than 1200 in theevacuation process which means it is necessary for thegovernment to carry out rescue like providing energy tothe residents

While there are also many residents who consume lessenergy in the evacuation process for various reasons such asbeing closer to the rescue point knowing correct evacuationdirection encountering rescue team and so on these sce-narios do not mean no energy supplement for thoseresidents

Shelter ResidentsRescue teamsHouseRoad

Figure 2 Initial display of rainstorm disaster simulation using Repast

Waterlogging point

Figure 3 Running display of rainstorm disaster simulation using Repast

Table 2 Parameter setting list

Parameter Description Value (initialization)L Life value (energy) 3000I Information value 1500L Basic consumption 1200ε Information delivery 03β Recovery factor 05Q Government benefit 500A Disaster analysis 800λ Disaster analysis coefficient 06

S Relief supplies for a singleresident 120

6 Complexity

In the process of disaster relief information asymmetrybetween the two agents makes it difficult for each other tomake effective and helpful decisions Residents need todecide whether to move out for disaster avoidance auton-omously or stay in original place to wait for rescue based ontheir own disaster information -e government needs tomake decisions based on own disaster relief resources anddisaster analysis results and then determine the strategy ofdispatching rescue teams and the establishment of tempo-rary shelters -e average remaining energy of residents isused to depict efficiency of disaster relief of differentprobabilities

Figure 5 shows the average remaining energy of residentsof different rescue probabilities of the government Obvi-ously when PG 07 a maximum value is obtained then itstarts to decrease So it is inferred that when the govern-mentrsquos disaster relief probability is around 07 greater di-saster relief efficiency will be achieved

Figure 6 that has two maxima shows the final averageremaining energy of different disaster avoidance probabil-ities of residents -e first maximum occurs at PR 022although the residents can obtain the maximum residualenergy the probability of the residents staying in originalplace is very high which may increase the difficulty of thegovernmentrsquos disaster relief process -e second maximumis obtained at PR 054 Although the remaining energy issmaller than the counterpart at PR 022 it is still greaterthan the initial energy of 3000 It convinces us about the factthat when the residentsrsquo disaster avoidance probability isaround 054 the maximum disaster relief efficiency can beobtained for both agents

32 Entire Reaction Analysis In the simulation process thechoices of residents and government influence each otherEntire Reaction (ER) describes the positive attitude andenthusiasm of all agents involved in disaster relief analysingthe ER of different strategies to reflect the overall disasterrelief efficiency Figure 7 reveals the positivity evolution ofdisaster relief and avoidance of government and residentsand evolution with Entire Reaction

Figure 7(a) shows PG tends to approach 100 and PR

tends to approach 08 when starting from PG PR 03Governmentrsquos strategy becomes stable on 1 at half of thesimulation ER climbs to high level at an early stage shown inFigure 7(b)

As can be observed in Figure 7(a) with the increase ofthe governmentrsquos disaster relief probability the residentsrsquodisaster avoidance probability decreases firstly and thenincreases indicating that the governmentrsquos low responsespeed has not stimulated residentsrsquo automatic disasteravoidance When the governmentrsquos disaster relief efficiencyreaches 07 the residentsrsquo disaster avoidance probability isstable at 08 which is consistent with the previous analysis ofthe governmentrsquos best disaster relief efficiency of 07

-e steady increase of ER from Figure 7(b) means thatwith the interaction of both agents residents and govern-ment will make decisions to avoid disasters and relief di-saster respectively which will increase disaster reliefefficiency

Figure 8 reveals the proportion that residents move outfor disaster avoidance autonomously with PG 05 -evalue of PR fluctuates around the initial value PR 05 -ecorresponding ER waves between 06 and 08

When the governmentrsquos disaster relief efficiency is fixedat a low level the residentsrsquo disaster avoidance probabilitywill also be very low and the ER will also be at a very lowlevel which shows that when one sidersquos disaster relief at-titude is negative it cannot stimulate the other sidersquos efficientavoiding disaster autonomously

For instance in a rainstorm disaster the governmentand residents must rely on themselves independently be-cause of asymmetric information When residents move outfor disaster avoidance autonomously the government mightaccept them during the disaster relief process which willincrease the efficiency of the governmentrsquos disaster relief andconvey positive feedback to increase residentsrsquo confidence indisaster resistance

33 8e Impact of Government Disaster AnalysisInformation is a very important recourse in the process ofdisaster relief and emergency management -e amount of

700 800 900 1000 1100 1200 1300 1400 1500 16000

5

10

15

20

25

30

35

40

Am

ount

Average energy consumption

Figure 4 Statistics on average energy consumption

Complexity 7

information directly affects the decisions of both agents soall involved agents must collect as much information aspossible to help them make the most appropriate decisionResidents are often weak agents in the process of avoidingdisaster and collect existing information without muchcapability to research useful information individuallyGenerally government can do information concentrationand dissemination with complex disaster analysis which

could guide disaster relief activities Figure 9 reveals thepositivity evolution of disaster relief and avoidance ofgovernment and residents without disaster analysis and withdisaster analysis

Figure 9(a) depicts the strategies evolution when gov-ernment arranges rescue regardless of analysing disasterGovernment and residents both are apt to decline the activestrategies On the contrary Figure 9(b) shows that if

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t

Resident agentGovernment agent

17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 25 26 27 28 29 300

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)

Figure 7 Evolution with initial PR 03 and PG 03 (a) Evolution of PG and PR (b) Entire Reaction

200022002400260028003000320034003600

012 020 032 040 052 060 070 080 090

Ave

rage

rem

aini

ng en

ergy

PG

Figure 5 Average remaining energy with different relief probabilities of government

2400

2600

2800

3000

3200

3400

3600

012 022 034 042 048 054 062 072 084

Ave

rage

rem

aini

ng en

ergy

PR

Figure 6 Average remaining energy of different disaster avoidance probabilities of residents

8 Complexity

government cares about analysing disaster in the process ofdisaster relief both government and residents tend toperform actively

34 Relief Supplies for Single Resident Analysis Because ofhigh uncertainty and time urgency governments usuallystockpile a certain amount of relief supplies in advance ofpotential disasters Hence governments face inventory riskand stock-out risk in the relief supply management -estored relief supplies by governments are wasted if no di-saster happens If the quantity of stored relief supplies is notsufficient stock-out risk will occur [75] It is of significancefor government to determine the relief supplies for a single

rescued resident By analysing the relief supplies of a singleresident when they are rescued it is possible to explore theactions of both agents and help the government make betterdecisions To explore governmentrsquos and residentsrsquo positivityof disaster relief and avoidance about different amount ofrelief supplies in urban rainstorms the variations in reliefsupplies in low level (Figure 10(a)) medium level(Figure 10(b)) and high level (Figure 10(c)) amounts arestudied through simulation Figure 10 reveals the positivityevolution of disaster relief and avoidance of government andresidents with different levels of relief supplies

Comparing Figure 10(a) with Figure 10(b) the reactionsof government and residents do not change a lot when S risesfrom 120 to 240 It cannot be neglected that there is a slight

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

t


20

01

02

03

04

05

06

Ratio

07

08

09

1

(a)


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17t

180

01

02

03

04

05

06

Ratio

07

08

09

1

(b)

Figure 9 Evolution with binary choice of analysing disaster (a) Regardless of analysing disaster (b) Disaster analysis matters

01 2 3 4 5 6 7 8 9 10 11

t


12

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11t

120

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)

Figure 8 Evolution with fixed PG 05 (a) Evolution of PG and PR (b) Entire Reaction

Complexity 9

decrease when S increases However residents and gov-ernment would reduce reactions if S is adequate like thesimulated S 480 shown in Figure 10(c) which denotes thatsufficient relief supplies may make all involved agents moretolerant of disaster

Specifically the greater the S the greater the pressure onthe government to undertake disaster relief and the moreenergy supplements residents receive when rescued whichwill lead to both agents being inactive in disaster relief Toconclude it is difficult for the government to bear huge reliefsupplies for a single resident and residents expect to wait inoriginal place for getting enough supplies

4 Conclusions

-e rainstorm disaster has caused great losses and harm tohuman beings June-September each year is the flood seasonin China nearly one-third of the cities are directly affectedby the rainstorms and floods -e cities need to be resiliently

transformed Flood prevention and drainage have becomean important part of urban modernization China has alsogiven a lot of investment to improve urban waterloggingproblems and maintain peoplersquos normal living orderHowever in the context of ldquosmall government big societyrdquothe overall security of the country requires the participationof all members of the society

Exploring residents evacuation discipline and effectiverescue is an important part of it-e asymmetry informationbetween the government and residents is the most difficultfactor In this paper the MASmethod is used to simulate thegovernmentrsquos and residentsrsquo strategic choices after therainstorm disaster in the city -e main conclusions are asfollows

(1) In the process of disaster relief it is of great sig-nificance to determine the best decision strategy ofresidents and government Considering the researchbackground it can be concluded that the best

Resident agent

Government agent

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

Resident agent

Government agent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 250

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3

t4

01

02

03

04

05

06

Ratio

07

08

09

1

(c)

Figure 10 -e evolution with the different S (a) Evolution with S 120 (b) Evolution with S 240 (c) Evolution with S 480

10 Complexity

probability for residents to avoid disaster automat-ically is 054 and the best probability for the gov-ernment to relief disaster is 07

(2) -e interaction between residents and the govern-ment can improve the efficiency of disaster reliefMaintaining a positive attitude towards disasterrelief between government and residents will stim-ulate both agents to act actively and enhance EntireReaction

(3) In the process of disaster relief the governmentshould automatically analyse the disaster situationcontrol the overall situation of the disaster macro-scopically increase decision-making informationand make more accurate strategies

(4) Appropriate relief supplies amount should be ap-plied for a single resident when they are rescued It isnecessary to mobilize the initiative of residents andreduce the pressure of the government on disasterrelief

Data Availability

-e data used to support the findings of this study areavailable from the corresponding author upon request

Conflicts of Interest

-e authors declare that they have no known competingfinancial interests or personal relationships that could haveappeared to influence the work reported in this paper

Acknowledgments

-e authors would like to thankWei Zhou TianyuWan andXu Sun for their helpful suggestions and technology support-is research was supported in part by National NaturalScience Foundation of China (Grant no 71603197)

References

[1] T Wang J B Liu and G Li ldquoA real options-based decision-making model for infrastructure investment to preventrainstorm disastersrdquo Production and Operations Manage-ment vol 28 no 11 pp 2699ndash2715 2019

[2] N Altay and W G Green ldquoORMS research in disasteroperations managementrdquo European Journal of OperationalResearch vol 175 no 1 pp 475ndash493 2006

[3] M Janssen N Lee and A Cresswell ldquoAdvances in multi-agency disaster management key elements in disaster re-searchrdquo Information Systems Frontiers vol 12 no 1 pp 1ndash72010

[4] J Buckland and M Rahman ldquoCommunity-based disastermanagement during the 1997 red river flood in CanadardquoDisasters vol 23 no 2 pp 174ndash191 2010

[5] Y Shi ldquoRisk analysis of rainstorm waterlogging on residencesin Shanghai based on scenario simulationrdquo Natural Hazardsvol 62 no 2 pp 677ndash689 2012

[6] X Zhang G M Hu and Y Xu ldquoUrban rainwater utilizationand its role in mitigating urban waterlogging problems-A casestudy in nanjing Chinardquo Water Resources Managementvol 26 no 13 pp 3757ndash3766 2012

[7] X Wu Z D Yu and R L Wilby ldquoAn evaluation of theimpacts of land surface modification storm sewer develop-ment and rainfall variation on waterlogging risk in Shang-hairdquo Natural Hazards vol 63 no 2 pp 305ndash323 2012

[8] R-S Quan ldquoRainstorm waterlogging risk assessment incentral urban area of Shanghai based on multiple scenariosimulationrdquo Natural Hazards vol 73 no 3 pp 1569ndash15852014

[9] Y F Ning W Y Dong L S Lin et al ldquoAnalyzing the causesof urban waterlogging and sponge city technology in Chinardquoin Proceedings of the 2nd International Conference on Ad-vances in Energy Resources and Environment Engineering(ICAESEE) vol 59 2017 Article ID 012047

[10] Y Zhang P Luo S Zhao et al ldquoControl and remediationmethods for eutrophic lakes in recent 30 yearsrdquoWater Scienceamp Technology vol 81 no 6 pp 1099ndash1113 2020

[11] T Sewell R E Stephens D Dominey-Howes et al ldquoDisasterdeclarations associated with bushfires floods and storms inNew South Wales Australia between 2004 and 2014rdquo Sci-entific Reports vol 6 Article ID 36369 2016

[12] D Satoh R Y Takano and T Mochida ldquoReduction ofcommunication demand under disaster congestion usingcontrol to change human communication behavior withoutdirect restrictionrdquo Computer Networks vol 134 pp 105ndash1152018

[13] O Ergun L Gui J L Heier Stamm P Keskinocak andJ Swann ldquoImproving humanitarian operations throughtechnology-enabled collaborationrdquo Production and Opera-tions Management vol 23 no 6 pp 1002ndash1014 2014

[14] P Luo Y Sun S Wang et al ldquoHistorical assessment andfuture sustainability challenges of Egyptian water resourcesmanagementrdquo Journal of Cleaner Production vol 263 ArticleID 121154 2020

[15] B Su H Huang and Y Li ldquoIntegrated simulation method forwaterlogging and traffic congestion under urban rainstormsrdquoNatural Hazards vol 81 no 1 pp 23ndash40 2016

[16] Z Zheng S Qi and Y Xu ldquoQuestionable frequent occurrenceof urban flood hazards in modern cities of Chinardquo NaturalHazards vol 65 no 1 pp 1009-1010 2013

[17] Z Xudong Y Kun P Shuangyun et al ldquo-e study of urbanrainstorm waterlogging scenario simulation based on GIS andSWMM model-take the example of Kunming Dongfeng EastRoad catchment areardquo in Proceedings of the 21st InternationalConference on Geoinformatics (Geoinformatics) InternationalConference on Geoinformatics Kai Feng China June 2013

[18] S Chowdhury A Emelogu M Marufuzzaman S G Nurreand L Bian ldquoDrones for disaster response and relief opera-tions a continuous approximation modelrdquo InternationalJournal of Production Economics vol 188 pp 167ndash184 2017

[19] C Lai X Chen X Chen Z Wang X Wu and S Zhao ldquoAfuzzy comprehensive evaluation model for flood risk based onthe combination weight of game theoryrdquo Natural Hazardsvol 77 no 2 pp 1243ndash1259 2015

[20] M P Scaparra and R Church ldquoProtecting supply systems tomitigate potential disasterrdquo International Regional ScienceReview vol 35 no 2 pp 188ndash210 2012

[21] A M Caunhye X Nie and S Pokharel ldquoOptimizationmodels in emergency logistics a literature reviewrdquo Socio-Economic Planning Sciences vol 46 no 1 pp 4ndash13 2012

[22] C Wang J Wu X Wang et al ldquoApplication of the hiddenMarkov model in a dynamic risk assessment of rainstorms inDalian Chinardquo Stochastic Environmental Research and RiskAssessment vol 32 no 10 pp 2045ndash2056 2018

Complexity 11

[23] J Salmeron and A Apte ldquoStochastic optimization for naturaldisaster asset prepositioningrdquo Production and OperationsManagement vol 19 no 5 pp 561ndash574 2010

[24] M Mahootchi and S Golmohammadi ldquoDeveloping a newstochastic model considering bi-directional relations in anatural disaster a possible earthquake in Tehran (the Capitalof Islamic Republic of Iran)rdquo Annals of Operations Researchvol 269 no 1-2 pp 439ndash473 2017

[25] Y Sawada and Y Takasaki ldquoNatural disaster poverty anddevelopment an introductionrdquo World Development vol 94pp 2ndash15 2017

[26] J W Delleur and Y Gyasi-Agyei ldquoPrediction of suspendedsolids in urban Sewers by transfer function modelrdquo WaterScience amp Technology vol 29 no 1-2 pp 171ndash179 1994

[27] H Yu C G Huang and C Wu ldquoApplication of thestormwater management model to a piedmont city a casestudy of Jinan City Chinardquo Water Science and Technologyvol 70 no 5 pp 858ndash864 2014

[28] D S Bisht C Chatterjee S Kalakoti P Upadhyay M Sahooand A Panda ldquoModeling urban floods and drainage usingSWMM and MIKE urban a case studyrdquo Natural Hazardsvol 84 no 2 pp 749ndash776 2016

[29] X Pan K Han and K H Law ldquoA multi-agent basedframework for the simulation of human and social behaviorsduring emergency evacuationsrdquo Ai amp Society vol 22 no 2pp 113ndash132 2007

[30] C Fikar P C P Hirsch and P C Nolz ldquoAgent-basedsimulation optimization for dynamic disaster relief distri-butionrdquo Central European Journal of Operations Researchvol 26 no 2 pp 423ndash442 2017

[31] S Iwanaga and A Namatame ldquoContagion of evacuationdecision making on real maprdquo Mobile Networks and Appli-cations vol 21 no 1 pp 206ndash214 2016

[32] K Zhu H J Tang and J C LiGong ldquoUsing a combinatorialauction-based approach for simulation of cooperative rescueoperations in disaster reliefrdquo International Journal of Mod-eling Simulation and Scientific Computing vol 09 no 4Article ID 1850035 2018

[33] M Hashemipour J S Stuban and J Dever ldquoA disastermultiagent coordination simulation system to evaluate thedesign of a first-response teamrdquo Systems Engineering vol 21no 4 pp 322ndash344 2018

[34] J Tang H K Zhu and C C S LiaoGong ldquoUsing auction-based task allocation scheme for simulation optimization ofsearch and rescue in disaster reliefrdquo Simulation ModellingPractice and 8eory vol 82 pp 132ndash146 2018

[35] M Zhang S R Starbuck S S LeeHwang M Choi andH-S Lee ldquoDistributed and interoperable simulation forcomprehensive disaster response management in facilitiesrdquoAutomation in Construction vol 93 pp 12ndash21 2018

[36] A Park A Pietro P Kini et al ldquoA platform for disasterresponse planning with interdependency simulation func-tionalityrdquovol 417 pp 183ndash197 in Proceedings of the 7thAnnual IFIP Working Group 1110 International Conferenceon Critical Infrastructure Protection (ICCIP) Advances inInformation and Communication Technology vol 417pp 183ndash197 George Washington University WashingtonDC USA March 2013

[37] R J Dawson R Peppe andMWang ldquoAn agent-based modelfor risk-based flood incident managementrdquo Natural Hazardsvol 59 no 1 pp 167ndash189 2011

[38] Q Yang Y Sun X Liu et al ldquoMAS-based evacuation sim-ulation of an urban community during an urban rainstormdisaster in Chinardquo Sustainability vol 12 no 2 pp 1ndash19 2020

[39] Q Yang X Sun X Liu et al ldquoMulti-agent simulation ofindividualsrsquo escape in the urban rainstorm context based ondynamic recognition-primed decision modelrdquoWater vol 12no 4 Article ID 1190 2020

[40] Q Hu and N Kapucu ldquoInformation communication tech-nology utilization for effective emergency management net-worksrdquo Public Management Review vol 18 no 3pp 323ndash348 2016

[41] D Cohen and S Aner ldquoCommon-value group contests withasymmetric informationrdquo Economics Letters vol 192 ArticleID 109164 2020

[42] J Ni J Zhao and L K Chu ldquoSupply contracting and processinnovation in a dynamic supply chain with informationasymmetryrdquo European Journal of Operational Researchvol 288 no 2 pp 552ndash562 2020

[43] G F Nel E Smit M Leon and Brummer ldquo-e link betweenInternet investor relations and information asymmetryrdquoSouth African Journal of Economic and Management Sciencesvol 21 no 1 pp 1ndash10 2018

[44] V Grimm E Revilla U Berger et al ldquoPattern-orientedmodeling of agent-based complex systems lessons fromecologyrdquo Science vol 310 no 5750 pp 987ndash991 2005

[45] R M Axelrod ldquo-e complexity of cooperation agent-basedmodels of competition and cooperationrdquo Complexity vol 3no 3 pp 46ndash48 1998

[46] C M Macal and M J North ldquoTutorial on agent-basedmodelling and simulationrdquo Journal of Simulation vol 4 no 3pp 151ndash162 2010

[47] E Bonabeau ldquoAgent-based modeling methods and tech-niques for simulating human systems Proceedings of thesackler colloquium on adaptive agents intelligence andemergent human organization-capturing complexity throughagent-based modelingrdquo in Proceedings of the NationalAcademy of Sciences of the Unite States of America vol 99no 3 pp 7280ndash7287 Irvine CA USA October 2001

[48] V Grimm U Berger F Bastiansen et al ldquoA standard pro-tocol for describing individual-based and agent-basedmodelsrdquo Ecological Modelling vol 198 no 1-2 pp 115ndash1262006

[49] G Belyavsky N Danilova and G Ougolnitsky ldquoAmarkovianmechanism of proportional resource allocation in the in-centive model as a dynamic stochastic inverse stackelberggamerdquo Mathematics vol 6 no 8 Article ID 131 2018

[50] S Sharma D K Ogunlana and J Grynovicki ldquoModelinghuman behavior during emergency evacuation using intelli-gent agents a multi-agent simulation approachrdquo InformationSystems Frontiers vol 20 no 4 pp 741ndash757 2018

[51] S Taga T Matsuzawa M Takimoto et al ldquoMulti-agent baseevacuation support system using MANETrdquo in Proceedings ofthe 10th International Conference on Computational CollectiveIntelligence (ICCCI) Lecture Notes in Artificial Intelligencevol 11055 pp 445ndash454 Bristol UK September 2018

[52] J Z Leibo V Zambaldi M Lanctot et al ldquoMulti-agent re-inforcement learning in sequential social dilemmasrdquo inProceedings of the 16th International Conference on Auton-omous Agents and Multiagent Systems (AAMAS) pp 464ndash473 Sao Paulo Brazil May 2017

[53] N Hooshangi and A Asghar Alesheikh ldquoAgent-based taskallocation under uncertainties in disaster environments anapproach to interval uncertaintyrdquo International Journal ofDisaster Risk Reduction vol 24 pp 160ndash171 2017

[54] N Collier ldquoRepast an extensible framework for agent sim-ulationrdquo University Of Chicago Social Science ResearchBuilding vol 36 pp 371ndash375 2003

12 Complexity

[55] M J North N T Collier and J R Vos ldquoExperiences creatingthree implementations of the Repast agent modeling ToolkitrdquoACM Transactions on Modeling and Computer Simulationvol 16 no 1 pp 1ndash25 2006

[56] M J North T R Howe N T Collier et al ldquo-e Repastsimphony runtime systemrdquo in Proceeding of the Agent 2005Conference on Generative Social Processes Models andMechanisms pp 151ndash158 Chicago IL USA October 2005

[57] N Malleson L A Heppenstall and L See ldquoCrime reductionthrough simulation an agent-based model of burglaryrdquoComputers Environment and Urban Systems vol 34 no 3pp 236ndash250 2010

[58] F-Y Wang K M Carley D Zeng and W Mao ldquoSocialcomputing from social informatics to social intelligencerdquoIEEE Intelligent Systems vol 22 no 2 pp 79ndash83 2007

[59] F Y Mao X Wang L Li et al ldquoSteps toward parallel in-telligencerdquo IEEECAA Journal of Automatica Sinica vol 3pp 345ndash348 2016

[60] B Linghu F Chen X Guo et al ldquoA conceptual model forflood disaster risk assessment based on agent-based model-ingrdquo in Proceedings of the International Conference onComputer Sciences and Applications (CSA) pp 369ndash373International Conference on CSA Wuhan China December2013

[61] C F Camerer ldquoWhen does ldquoeconomic manrdquo dominate socialbehaviorrdquo Science vol 311 no 5757 pp 47ndash52 2006

[62] C E Fritz and E S Marks ldquo-e NORC studies of humanbehavior in disasterrdquo Journal of Social Issues vol 10 no 3pp 26ndash41 1954

[63] J Ghurye G Krings and V Frias-Martinez ldquoA framework tomodel human behavior at large scale during natural disastersrdquoin Proceedings of the 2016 17th IEEE International Conferenceon Mobile Data Management (MDM) pp 18ndash27 PortoPortugal June 2016

[64] N Parikh R J Hayatnagarkar M V Marathe and S SwarupldquoA comparison of multiple behavior models in a simulation ofthe aftermath of an improvised nuclear detonationrdquo Au-tonomous Agents and Multi-Agent Systems vol 30 no 6pp 1148ndash1174 2016

[65] A Dan E Owens and O Rozenbaum ldquoDo informationreleases increase or decrease information asymmetry Newevidence from analyst forecast announcementsrdquo Journal ofAccounting amp Economics vol 62 no 1 pp 121ndash138 2016

[66] J Martınez-Ferrero D Ruiz-Cano and I-M Garcıa-Sanchezldquo-e causal link between sustainable disclosure and infor-mation asymmetry the moderating role of the stakeholderprotection contextrdquo Corporate Social Responsibility and En-vironmental Management vol 23 no 5 pp 319ndash332 2016

[67] L Pearce ldquoDisaster management and community planningand public participation how to achieve sustainable hazardmitigationrdquo Natural Hazards vol 28 no 2-3 pp 211ndash2282003

[68] F -omalla E T Spanger-Siegfried and J G RockstromldquoReducing hazard vulnerability towards a common approachbetween disaster risk reduction and climate adaptationrdquoDisasters vol 30 no 1 pp 39ndash48 2006

[69] V Strandh and N Eklund ldquoEmergent groups in disasterresearch varieties of scientific observation over time andacross studies of nine natural disastersrdquo Journal of Contin-gencies and Crisis Management vol 26 no 3 pp 329ndash3372018

[70] I Noy ldquo-e macroeconomic consequences of disastersrdquoJournal of Development Economics vol 88 no 2 pp 221ndash2312009

[71] R J Burby ldquoHurricane katrina and the paradoxes of gov-ernment disaster policy bringing about wise governmentaldecisions for hazardous areasrdquo 8e Annals of the AmericanAcademy of Political and Social Science vol 604 no 1pp 171ndash191 2006

[72] B Raphael ldquoCrowds and other collectives complexities ofhuman behaviors in mass emergenciesrdquo Psychiatry Inter-personal and Biological Processes vol 68 no 2 pp 115ndash1202005

[73] A K Chakravarty ldquoA contingent plan for disaster responserdquoInternational Journal of Production Economics vol 134 no 1pp 3ndash15 2011

[74] S H Kang and M Skidmore ldquo-e effects of natural disasterson social trust evidence from South Koreardquo Sustainabilityvol 10 no 9 pp 1ndash16 2018

[75] Y Liu J Tian F Gengzhong et al ldquoA relief supplies pur-chasing model via option contractsrdquo Computers amp IndustrialEngineering vol 137 Article ID 106009 2019

Complexity 13

were stuck in the streets and could not get home until late atnight On July 21 2012 a more powerful rainstorm struckBeijing Serious waterlogging occurred in some low-lyingroad sections and some people even drowned in their cars[15] During the disaster 79 people lost their lives about 19million people suffered and the total economic damageswere almost 10 billion RMB [16] -us a series of problemscaused by urban storm waterlogging has become the re-search focus and more attention is paid to it [17]

Disaster management has everlasting attention [18] Inorder to reduce the damage caused by rainstorm disastersensure traffic safety and control emergency cost extensiverainstorm disaster models have been developed to assessdisaster risk and predict the impact of rainstorm disasters aswell [19ndash21]Wang et al [22] used the hiddenMarkovmodel(HMM) method to conduct a dynamic risk assessment ofnatural disasters in Dalian China Javier Salmeron et al [23]developed a two-stage stochastic optimization model toguide the relief assets allocation Mahootchi and Golmo-hammadi [24] extended the mathematical two-stage sto-chastic optimization model to provide appropriate responseand better service Sawada and Takasaki [25] established amicroeconomic analysis framework between disasters andpoverty revealing that mutual aid is critical to effectivepoverty alleviation ldquoLife firstrdquo is often seen as the principalrule in the disaster rescue so as in the rainstorm relief Whilehydrological analysis and facility planning occupied majorresearch position in urban rainstorm evolution like SWMM(Storm Water Management Model) [17] MOUSE (Modelfor Urban Sewers) [26] etc various mutual influencingfactors underlie the complex system of urban drainage [27]Actually owing to the increase of complexity of urbandevelopment the effectiveness and efficiency of modelling issubject to the dynamic parameter optimization [28]

-ere are many tools for simulating [29] human andsocial behaviours in emergencies -e use of these tools canhelp determine the locations of temporary shelter [30] andpredict disaster losses [31] which can improve the efficiencyof disaster response Search and rescue are the major parts ofdisaster response [32] in which agent-based simulation hasbeen proved as a useful method [33] Tang Jet al [34] op-timized search and rescue using auction-based task alloca-tion scheme through agent-based simulation Choi et al [35]developed a flexible and interoperable distributed simulationenvironment for comprehensive disaster response man-agement Alsubaie et al [36] constructed a disaster responseplan based on supervisory control and data acquisition(SCADA) systems Modelling by multi-agent system (MAS)has been a common tendency for the operation of social-economic system under changing natural conditions [37]like evacuation route choice of pedestrians in an urbanrainstorm [38 39]

Effective communication and coordination are crucialaspects of emergency management [40] However infor-mation asymmetry often manifests when one party has moreor better information than the other Agents with moreinformation will have higher expected payoff than those withless [41] Besides agentsrsquo decisions will have potential dis-tortion due to information asymmetry [42]-e difference of

decision among agents therefore has a positive correlationwith information asymmetry [43]

2 Research Method

21 MAS Modelling Agent-based simulation seems to be akind of dynamic networks of interacting agents [44] Byspecifying the rules between individuals computer simu-lations can be used to demonstrate complex behaviourpatterns from other simple models [45] Agent-based sim-ulation is a relatively new approach to modelling systemscomposed of autonomous interacting agents [46] Eachindividual is able to make decisions based on circumstancesand rules [47] Repetitive interactions are hard to implementwith mathematical methods which in contrast is a pecu-liarity of agent-based simulation Agent-based simulation iswidely used in many fields to deal with problems covered incomplex systems [48] like resource allocation [49] crowdevacuation [50] decision-making systems [51] and coop-erative game [52] -e application of agent-based simulationin disaster management can improve search and rescueefficiency [53]

Since the 1990s agent-modelling platforms such asSTARLOGO SWARM NETLOGO REPAST (RecursivePorous Agent Simulation Toolkit) etc have been developedREPAST has outstanding scalability for its JAVA pro-gramming [54] and has already been used in modellingsocial behaviour [55 56] and city management with geo-spatial design [57] REPAST can be redesigned for diversepractical needs Bridging the gap between agent-basedsimulation and actuality is a challenge for MAS modellingwhich is often conducted by parallel execution and adaptivecontrol between artificial and real world [58] through theartificial societies-computational experiments-parallel exe-cution (ACP) platform [59] to interact with the real socialsystem and provide reliable support for themanagement anddecision-making of real social scenes -erefore the evo-lution of the situation co-evolution and closed-loopfeedback are realized

For deepening our previous study on human reaction indisaster we followed the previous MAS modelling onrainstorm relief [38 39] Yang et al analysed the efficiency ofdisaster relief by considering the crowd psychology of res-idents and diverse density distribution of relief settlementHowever the study focuses on the outcome generated fromdiverse pedestrian categories and distribution of shelterswhich does not involve the mutual reaction from diversesubjects

Actually information asymmetry is inevitable in rain-storm the interaction between the government and resi-dents underlies the disaster relief efficiency Exploring theimpact on the disaster relief from different initiatives andinteractions helps implement disaster situation analysis anddisaster relief decision-making reasonably

22 Agent Attributes -e agent is a physical or abstractentity and the basic unit in multi-agent systems (MAS) [60]From a practical modelling standpoint the agent is required

2 Complexity















Complexity 3







PGt+1 P

Gt +


Gt1113872 1113873 E

PGminus E

G1113874 1113875

EPG

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (1)

EPG

1 minus PGt1113872 1113873 P

Rt λA minus αS1113872 1113873 + E

G (2)









ER N minus n

N (3)





Resident



Government



4 Complexity


PRt+1 P

Rt +


Rt1113872 1113873 E

PRminus E

R1113872 1113873

EPR

⎡⎢⎢⎣ ⎤⎥⎥⎦ (4)

EPR

1 minus PRt1113872 1113873 αP

Gt minus αβP

Gt minus α1113872 1113873L

+ PGt minus P

Gt ε + ε1113872 1113873I + E

R

(5)












Residentaction

Initialization

Move out

Arrive

Rescued

Wading


Failure

Yes

Yes No

YesNo

Yes

No

No

No

Terminated

Yes

Governmentaction

Rescue


Complexity 5








Waterlogging point





6 Complexity













700 800 900 1000 1100 1200 1300 1400 1500 16000

5

10

15

20

25

30

35

40

Am

ount



Complexity 7




01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t


17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 25 26 27 28 29 300

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


200022002400260028003000320034003600

012 020 032 040 052 060 070 080 090

Ave

rage

rem

aini

ng en

ergy

PG


2400

2600

2800

3000

3200

3400

3600

012 022 034 042 048 054 062 072 084

Ave

rage

rem

aini

ng en

ergy

PR


8 Complexity





01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

t


20

01

02

03

04

05

06

Ratio

07

08

09

1

(a)


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17t

180

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3 4 5 6 7 8 9 10 11

t


12

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11t

120

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


Complexity 9



4 Conclusions





Resident agent

Government agent

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

Resident agent

Government agent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 250

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3

t4

01

02

03

04

05

06

Ratio

07

08

09

1

(c)


10 Complexity





Data Availability




Acknowledgments


References























Complexity 11

































12 Complexity






















Complexity 13















Complexity 3







PGt+1 P

Gt +


Gt1113872 1113873 E

PGminus E

G1113874 1113875

EPG

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (1)

EPG

1 minus PGt1113872 1113873 P

Rt λA minus αS1113872 1113873 + E

G (2)









ER N minus n

N (3)





Resident



Government



4 Complexity


PRt+1 P

Rt +


Rt1113872 1113873 E

PRminus E

R1113872 1113873

EPR

⎡⎢⎢⎣ ⎤⎥⎥⎦ (4)

EPR

1 minus PRt1113872 1113873 αP

Gt minus αβP

Gt minus α1113872 1113873L

+ PGt minus P

Gt ε + ε1113872 1113873I + E

R

(5)












Residentaction

Initialization

Move out

Arrive

Rescued

Wading


Failure

Yes

Yes No

YesNo

Yes

No

No

No

Terminated

Yes

Governmentaction

Rescue


Complexity 5








Waterlogging point





6 Complexity













700 800 900 1000 1100 1200 1300 1400 1500 16000

5

10

15

20

25

30

35

40

Am

ount



Complexity 7




01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t


17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 25 26 27 28 29 300

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


200022002400260028003000320034003600

012 020 032 040 052 060 070 080 090

Ave

rage

rem

aini

ng en

ergy

PG


2400

2600

2800

3000

3200

3400

3600

012 022 034 042 048 054 062 072 084

Ave

rage

rem

aini

ng en

ergy

PR


8 Complexity





01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

t


20

01

02

03

04

05

06

Ratio

07

08

09

1

(a)


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17t

180

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3 4 5 6 7 8 9 10 11

t


12

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11t

120

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


Complexity 9



4 Conclusions





Resident agent

Government agent

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

Resident agent

Government agent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 250

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3

t4

01

02

03

04

05

06

Ratio

07

08

09

1

(c)


10 Complexity





Data Availability




Acknowledgments


References























Complexity 11

































12 Complexity






















Complexity 13







PGt+1 P

Gt +


Gt1113872 1113873 E

PGminus E

G1113874 1113875

EPG

⎡⎢⎢⎢⎢⎢⎢⎢⎢⎢⎢⎣⎤⎥⎥⎥⎥⎥⎥⎥⎥⎥⎥⎦ (1)

EPG

1 minus PGt1113872 1113873 P

Rt λA minus αS1113872 1113873 + E

G (2)









ER N minus n

N (3)





Resident



Government



4 Complexity


PRt+1 P

Rt +


Rt1113872 1113873 E

PRminus E

R1113872 1113873

EPR

⎡⎢⎢⎣ ⎤⎥⎥⎦ (4)

EPR

1 minus PRt1113872 1113873 αP

Gt minus αβP

Gt minus α1113872 1113873L

+ PGt minus P

Gt ε + ε1113872 1113873I + E

R

(5)












Residentaction

Initialization

Move out

Arrive

Rescued

Wading


Failure

Yes

Yes No

YesNo

Yes

No

No

No

Terminated

Yes

Governmentaction

Rescue


Complexity 5








Waterlogging point





6 Complexity













700 800 900 1000 1100 1200 1300 1400 1500 16000

5

10

15

20

25

30

35

40

Am

ount



Complexity 7




01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t


17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 25 26 27 28 29 300

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


200022002400260028003000320034003600

012 020 032 040 052 060 070 080 090

Ave

rage

rem

aini

ng en

ergy

PG


2400

2600

2800

3000

3200

3400

3600

012 022 034 042 048 054 062 072 084

Ave

rage

rem

aini

ng en

ergy

PR


8 Complexity





01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

t


20

01

02

03

04

05

06

Ratio

07

08

09

1

(a)


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17t

180

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3 4 5 6 7 8 9 10 11

t


12

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11t

120

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


Complexity 9



4 Conclusions





Resident agent

Government agent

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

Resident agent

Government agent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 250

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3

t4

01

02

03

04

05

06

Ratio

07

08

09

1

(c)


10 Complexity





Data Availability




Acknowledgments


References























Complexity 11

































12 Complexity






















Complexity 13


PRt+1 P

Rt +


Rt1113872 1113873 E

PRminus E

R1113872 1113873

EPR

⎡⎢⎢⎣ ⎤⎥⎥⎦ (4)

EPR

1 minus PRt1113872 1113873 αP

Gt minus αβP

Gt minus α1113872 1113873L

+ PGt minus P

Gt ε + ε1113872 1113873I + E

R

(5)












Residentaction

Initialization

Move out

Arrive

Rescued

Wading


Failure

Yes

Yes No

YesNo

Yes

No

No

No

Terminated

Yes

Governmentaction

Rescue


Complexity 5








Waterlogging point





6 Complexity













700 800 900 1000 1100 1200 1300 1400 1500 16000

5

10

15

20

25

30

35

40

Am

ount



Complexity 7




01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t


17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 25 26 27 28 29 300

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


200022002400260028003000320034003600

012 020 032 040 052 060 070 080 090

Ave

rage

rem

aini

ng en

ergy

PG


2400

2600

2800

3000

3200

3400

3600

012 022 034 042 048 054 062 072 084

Ave

rage

rem

aini

ng en

ergy

PR


8 Complexity





01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

t


20

01

02

03

04

05

06

Ratio

07

08

09

1

(a)


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17t

180

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3 4 5 6 7 8 9 10 11

t


12

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11t

120

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


Complexity 9



4 Conclusions





Resident agent

Government agent

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

Resident agent

Government agent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 250

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3

t4

01

02

03

04

05

06

Ratio

07

08

09

1

(c)


10 Complexity





Data Availability




Acknowledgments


References























Complexity 11

































12 Complexity






















Complexity 13








Waterlogging point





6 Complexity













700 800 900 1000 1100 1200 1300 1400 1500 16000

5

10

15

20

25

30

35

40

Am

ount



Complexity 7




01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t


17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 25 26 27 28 29 300

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


200022002400260028003000320034003600

012 020 032 040 052 060 070 080 090

Ave

rage

rem

aini

ng en

ergy

PG


2400

2600

2800

3000

3200

3400

3600

012 022 034 042 048 054 062 072 084

Ave

rage

rem

aini

ng en

ergy

PR


8 Complexity





01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

t


20

01

02

03

04

05

06

Ratio

07

08

09

1

(a)


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17t

180

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3 4 5 6 7 8 9 10 11

t


12

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11t

120

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


Complexity 9



4 Conclusions





Resident agent

Government agent

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

Resident agent

Government agent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 250

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3

t4

01

02

03

04

05

06

Ratio

07

08

09

1

(c)


10 Complexity





Data Availability




Acknowledgments


References























Complexity 11

































12 Complexity






















Complexity 13













700 800 900 1000 1100 1200 1300 1400 1500 16000

5

10

15

20

25

30

35

40

Am

ount



Complexity 7




01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t


17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 25 26 27 28 29 300

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


200022002400260028003000320034003600

012 020 032 040 052 060 070 080 090

Ave

rage

rem

aini

ng en

ergy

PG


2400

2600

2800

3000

3200

3400

3600

012 022 034 042 048 054 062 072 084

Ave

rage

rem

aini

ng en

ergy

PR


8 Complexity





01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

t


20

01

02

03

04

05

06

Ratio

07

08

09

1

(a)


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17t

180

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3 4 5 6 7 8 9 10 11

t


12

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11t

120

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


Complexity 9



4 Conclusions





Resident agent

Government agent

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

Resident agent

Government agent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 250

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3

t4

01

02

03

04

05

06

Ratio

07

08

09

1

(c)


10 Complexity





Data Availability




Acknowledgments


References























Complexity 11

































12 Complexity






















Complexity 13




01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t


17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 25 26 27 28 29 300

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


200022002400260028003000320034003600

012 020 032 040 052 060 070 080 090

Ave

rage

rem

aini

ng en

ergy

PG


2400

2600

2800

3000

3200

3400

3600

012 022 034 042 048 054 062 072 084

Ave

rage

rem

aini

ng en

ergy

PR


8 Complexity





01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

t


20

01

02

03

04

05

06

Ratio

07

08

09

1

(a)


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17t

180

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3 4 5 6 7 8 9 10 11

t


12

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11t

120

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


Complexity 9



4 Conclusions





Resident agent

Government agent

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

Resident agent

Government agent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 250

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3

t4

01

02

03

04

05

06

Ratio

07

08

09

1

(c)


10 Complexity





Data Availability




Acknowledgments


References























Complexity 11

































12 Complexity






















Complexity 13





01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19

t


20

01

02

03

04

05

06

Ratio

07

08

09

1

(a)


1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17t

180

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3 4 5 6 7 8 9 10 11

t


12

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

1 2 3 4 5 6 7 8 9 10 11t

120

01

02

03

04

05

06

Evac

uatio

n ra

tio

07

08

09

1

(b)


Complexity 9



4 Conclusions





Resident agent

Government agent

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

Resident agent

Government agent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 250

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3

t4

01

02

03

04

05

06

Ratio

07

08

09

1

(c)


10 Complexity





Data Availability




Acknowledgments


References























Complexity 11

































12 Complexity






















Complexity 13



4 Conclusions





Resident agent

Government agent

01 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16

t17 18 19 20 21 22 23 24 25 26 27 28 29 30

01

02

03

04

05

06

Ratio

07

08

09

1

(a)

Resident agent

Government agent

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16t

17 18 19 20 21 22 23 24 250

01

02

03

04

05

06

Ratio

07

08

09

1

(b)


01 2 3

t4

01

02

03

04

05

06

Ratio

07

08

09

1

(c)


10 Complexity





Data Availability




Acknowledgments


References























Complexity 11

































12 Complexity






















Complexity 13





Data Availability




Acknowledgments


References























Complexity 11

































12 Complexity






















Complexity 13

































12 Complexity






















Complexity 13






















Complexity 13

MAS-Based Interaction Simulation within Asymmetric ...

Documents

Transcript of MAS-Based Interaction Simulation within Asymmetric ...