Electricity Demand Forecasting Using Fuzzy-Neural Network

6
Electricity Demand Forecasting Using Fuzzy-Neural Network Naren Chandra Kattla Department of Computer Science Engineering VIT University, Chennai-600128, India Phone: +91-9765434698 Email: [email protected] Abstract—Demand forecasts are important for energy sup- pliers, electricity generation, transmission, distributors. Neuro- fuzzy systems have attracted growing interest of researchers in various scientific and engineering areas due to the increasing need of intelligent systems. The methods considered are fuzzy neural network (FNN) and an artificial neural network (ANN) trained using backpropagation (BP) algorithm. The implementation and forecasting hybrid fuzzy neural technique, which combines neural network modelling, and techniques from fuzzy logic and fuzzy set theory for electric load forecasting. The strengths of this powerful technique lie in its ability to forecast accurately. I. I NTRODUCTION Expanding access to electricity means including the rural areas people of 1.2 billion and still 0.6 billion there is no access to electricity and 0.4 billion people there is access to unreliable electricity networks. We need machine technique and practical approaches, as a mechanism that receives inputs directly and transmits to development, electricity plays a major role in both fighting against poverty and climate change. Forecasting using Artificial neural network(ANN) and Fuzzy Neural Network(FNN) had achieved satisfactory results. The implementation and comparing different techniques are followed by Artificial Neural Network Draw Back Of Neural Network The Proposed Approach Fuzzification Neural Network Defuzzification II. ARTIFICIAL NEURAL NETWORK A. Data Encoding And Normalization One of the essential keys to working with artificial NN is data encoding and normalization. After all data has been collected we may have some data in non-numeric data and high magnitude data. After all data has been converted to Encoding non-numeric output data to numeric values by using 1-of-N encoding, encoding non- numeric input data to numeric values by using 1-of-(N-1) encoding and apply normalization to high magnitude using any one out of two different types of normalization, Gaussian normalization and min-max normalization. In general, the min-max normalized value for some value x is (x - min) (max - min) (1) very simple, the Gaussian normalized value for some value x is (x - mean) Standard Deviation (2) B. Neural Network An artificial neural network will accept one or more inputs and produces one or more outputs. The basic neural network input to process (hidden layer) and process to output computation is known the feed-forward NN mechanism. Fig. 1: Neural Network Architecture Understanding the feed-forward NN mechanism is impor- tant to know how to create NN, which will make predictions. A NN computes in several stages to get output and the feed- forward NN mechanism is explained by using (Fig. 1), there are six inputs each line connecting one node to another repre- sents a weight constant. The weight is labelled IHWEIGHT [0] [0] it represents the weight from input GDP to hidden 1 and the weight in the lower right corner is labelled HOWEIGHT

Transcript of Electricity Demand Forecasting Using Fuzzy-Neural Network

Page 1: Electricity Demand Forecasting Using Fuzzy-Neural Network

Electricity Demand Forecasting Using Fuzzy-NeuralNetwork

Naren Chandra KattlaDepartment of Computer Science Engineering

VIT University, Chennai-600128, IndiaPhone: +91-9765434698

Email: [email protected]

Abstract—Demand forecasts are important for energy sup-pliers, electricity generation, transmission, distributors. Neuro-fuzzy systems have attracted growing interest of researchers invarious scientific and engineering areas due to the increasing needof intelligent systems. The methods considered are fuzzy neuralnetwork (FNN) and an artificial neural network (ANN) trainedusing backpropagation (BP) algorithm. The implementation andforecasting hybrid fuzzy neural technique, which combines neuralnetwork modelling, and techniques from fuzzy logic and fuzzy settheory for electric load forecasting. The strengths of this powerfultechnique lie in its ability to forecast accurately.

I. INTRODUCTION

Expanding access to electricity means including the ruralareas people of 1.2 billion and still 0.6 billion there is noaccess to electricity and 0.4 billion people there is access tounreliable electricity networks. We need machine techniqueand practical approaches, as a mechanism that receives inputsdirectly and transmits to development, electricity plays a majorrole in both fighting against poverty and climate change.Forecasting using Artificial neural network(ANN) and FuzzyNeural Network(FNN) had achieved satisfactory results.

The implementation and comparing different techniques arefollowed by

• Artificial Neural Network

• Draw Back Of Neural Network

• The Proposed Approach

• Fuzzification

• Neural Network

• Defuzzification

II. ARTIFICIAL NEURAL NETWORK

A. Data Encoding And Normalization

One of the essential keys to working with artificial NN isdata encoding and normalization.

After all data has been collected we may have some datain non-numeric data and high magnitude data. After all datahas been converted to Encoding non-numeric output data tonumeric values by using 1-of-N encoding, encoding non-numeric input data to numeric values by using 1-of-(N-1)encoding and apply normalization to high magnitude using

any one out of two different types of normalization, Gaussiannormalization and min-max normalization.

In general, the min-max normalized value for some valuex is

(x−min)

(max−min)(1)

very simple, the Gaussian normalized value for some value xis

(x−mean)

StandardDeviation(2)

B. Neural Network

An artificial neural network will accept one or moreinputs and produces one or more outputs. The basic neuralnetwork input to process (hidden layer) and process to outputcomputation is known the feed-forward NN mechanism.

Fig. 1: Neural Network Architecture

Understanding the feed-forward NN mechanism is impor-tant to know how to create NN, which will make predictions.A NN computes in several stages to get output and the feed-forward NN mechanism is explained by using (Fig. 1), thereare six inputs each line connecting one node to another repre-sents a weight constant. The weight is labelled IHWEIGHT [0][0] it represents the weight from input GDP to hidden 1 andthe weight in the lower right corner is labelled HOWEIGHT

Page 2: Electricity Demand Forecasting Using Fuzzy-Neural Network

[12] [0] it represents the weight from hidden 7 to output. Allhidden and output layer nodes has an arrow pointing to it thisare called the bias values.

Fig. 2: Activation Functions

The first stage in the feed-forward NN mechanism is tocompute for all hidden nodes. The value of hidden 1 nodeand is computed as the product of each input value and itsassociated weight are summed then associated bias value isadded. From (Fig. 2) Pre-activation sum function applied bythe activation function, it will be explained in detail nextsection, but for now it’s enough to say that the activationfunction is the hyperbolic tangent function, which is usuallyabbreviated tanh.

HiddenNode 1 = tanh(PreActivationSum) (3)

After all, hidden node output values have been computed,these values are used as inputs to the output nodes and it’scomputed slightly differently from the hidden nodes. The pre-liminary output sums, before activation, for output nodes arecomputed the same way as hidden node sums, the activationfunction for the output layer is called the softmax function.

For 1st Output Node:

output =ePreActivationOutput Sum

ePreActivationOutput Sum(4)

For nth Output Node:

output =a

b(5)

Form equation 5, a = ePreActivationOutput Sum and b =ePreActivationOutput 1Sum + ...+ ePreActivationOutput nSum

If we look at (Fig. 2), we will see these are the final outputvalues computed by the neural network. Where exp functionis the math constant e = 2.71828. . . raised to the xth power.Notice the output values sum is 1.0, which is not a coincidencethat is the point of using the softmax function.

1) Activation Functions: In (Fig. 2) for hidden layer nodeactivation function is hyperbolic tangent and for the outputlayer node activation functions is softmax function are usedand there is a third common activation function called thelogistic sigmoid function. In general, the hyperbolic tangentfunction is the best choice for hidden layer activation. Foroutput layer activation, if neural network is performing classi-fication where the dependent variable to be predicted has threeor more values softmax activation is the best choice. If neu-ral network is performing classification where the dependentvariable has exactly two possible values the logistic sigmoidactivation function is the best choice for output layer activation.

C. Training

The ultimate goal of a neural network is to make aprediction. In order to make a prediction, a neural networkmust first be trained. Training a neural network means findinga set of good weights and bias values so that the known outputsof some training data match the outputs computed using theweights and bias values. The resulting weights and bias valuesfor a particular problem are often collectively called a modeland it’s used to predict the output for unseen inputs that iscomputed for known output values.

The data set consists of a total of 5 items data set andsplits it randomly into a 4-item subset (80%) to be used fortraining and a 1-item subset (20%) to be used for testing,that is, to be used to estimate the probability of a correctclassification on data that has not been seen before. Traininga 6-7-1 fully connected feed-forward NN using the back-propagation algorithm in conjunction with a technique that iscalled incremental training.

After training has completed, the accuracy of the resultingmodel is computed on the training set and on the test set.

III. DRAW BACKS OF USING NEURAL NETWORK

These ANN based methods had given better performancewhen compared to traditional techniques, the accuracy islimited due to the following limitations in ANN:

• The accuracy is depended on the training data set, Soselecting correct training data to cover the entire posesproblems is difficult.

• The neural networks need to re-training due to changein seasons.

These methods lack when compared to common-senseknowledge frequently used by human experts to forecast loadprofiles. The predictive accuracy of forecasting models de-pends on projecting values and interrelationships of importantparameters (weather).

In order to overcome problems and to improve the fore-casting, a unique approach by combining the powers of neuralnetwork and fuzzy logic techniques. This hybrid techniques isgood by combining both features to overcome the limitationsof each.

Page 3: Electricity Demand Forecasting Using Fuzzy-Neural Network

IV. THE PROPOSED ARCHITECTURE

The main idea behind this hybrid approach is : the fuzzyknowledge-base models the knowledge about the system andits input parameters, quantitative as well as qualitative, and theneural network captures the inexplicable relationship betweenfuzzy inputs and outputs.

Fig. 3: Flow of processing

Form (fig. 3)inputs to this hybrid model include the dailytemperature, rain forecast, seasons, day type. A fuzzy front-endprocessor has been developed for preprocessing these inputswith the application of fuzzy rules. The front-end processoreffectively handles the different types of inputs, both numericas well as fuzzy, and produces a fuzzy output vector which isthen fed to a three- layer back propagation neural network.During training, the neural network captures the unknownmapping between these input variables and the target outputs.Once trained, the outputs of the neural network, interpreted asfuzzy membership functions of the target load, are defuzzifiedto obtain the load profile for the following day.

V. FUZZIFICATION

A. Fuzzification variables

The inputs to the fuzzy consist of the daily temperature,rain forecast, seasons, day type. The first step consists ofdefining all inputs and outputs as fuzzy sets and variablesvalues are defined range of possible values for each variable(Fig. 4).

Fig. 4: Fuzzification of input

The output variables is electricity consumed.

B. Membership Function

A membership function (MF) is a curve that defines howeach point in the input space is mapped to a membership value(or degree of membership) between 0 and 1. For the followingvariable the membership function as show below:

1) Temperature Membership Function: The daily temper-ature for the day are represented range is [25 45] and theseare divided into three linguistic classes each [low (L), medium(M), high (H)] as in the (Fig. 5).

Fig. 5: Membership function of Temperature

2) Rain Forecast Membership Function: The rain forecasthas been found to play a dominant role in the accuracy of theforecast.The weather forecast in terms of rain is representedas fuzzy subsets [very heavy (VH), heavy (H), medium (M),light showers (LH), no rain (NR)] (Fig. 6).

Fig. 6: Membership function of Rain Forecast

3) Seasons Membership Function: In seasons the rangeis [1 4]and these are divided into four linguistic classes forspring[March, April, May], summer[June, July, August], au-tumn[September, October, November], winter[December, Jan-uary, February] (Fig. 7).

4) Day Type Membership Function: The distance of aweekdays (or) weekends (for example, from a Saturday orMonday to Sunday) the range is [0 4] and it is divided intofour segments[war (F), near (N), very near (VN), day (D] (Fig.8).

5) Electricity Consumed Membership Function: Thechange in load is represented by a fuzzy variable P, divided intofive referential sets [negative big (NB), negative small (NS),zero (ZE), positive small (PS), positive big (PB)], as shown in(Fig. 9).

Page 4: Electricity Demand Forecasting Using Fuzzy-Neural Network

Fig. 7: Membership function of Seasons

Fig. 8: Membership function of Day Type

Fig. 9: Membership function of Electricity consumed

C. Identification of Day

In graph on x-axis is weekly cycles and on y-axis iselectricity consumed load curve, with random variations asseasons changed the electricity consumed load is increas-ing(winter, spring, summer). Based on the load curve, day typeare grouped in to two separate classes:

1) Weekday (Monday to Friday): The shape of the loadcurve on all weekdays, Friday is high when compare to otherdays (Fig. 10)

2) Saturday and Sunday: The load-curve on Saturdays,most businesses are open for the first half of the day, is

Fig. 10: Typical load shape for weekdays

substantially different from the rest of the days,due to thishigh load demand in the morning.The load curve on Sundays,where less when compare to Saturday (Fig. 11). By this loadon holidays is very less compare to businesses days.

Fig. 11: Load on weekends

D. Fuzzy Rules

In most case, more than one fuzzy rules are applied forsame output variables. For example, the fuzzy rules:

Rule 1 : IF Temperature isHighTHEN Load isHigh

Rule 2 : IF Rain is V ery Heavy THEN Load isLow

E. Training Patterns

The training patterns in the learning of the fuzzy-neuralnetwork consist of Sugeno-Type Fuzzy Inference System(FIS)is used in this input variables applied with membership valuesof p and fuzzy inference rules. The outcome of fuzzy rulesas described above (Fig. 12). The Target load is output ofelectricity consumed and it also load into FIS with membershipfunction (Fig. 9). This processed data combines different kindsof knowledge before being fed to the neural network.

Page 5: Electricity Demand Forecasting Using Fuzzy-Neural Network

Fig. 12: Fuzzy rule based system

Fig. 13: preparation of Traning Set

VI. NEURAL NETWORK

The role of the neural network is to capture the unknownmapping between fuzzy input and output vectors. Form (Fig.14) First layer is associated with each input variables, Secondlayer for each input membership function is determined by in-putmf, Third layer is rules by the fuzzy truth table proposition,In Fourth layer for each Sugeno-Type Fuzzy Inference Systemthe output membership value will be generated and Fifth layeris defuzzification for generated outputmf.

The 3-layer neural networks able to capture the electric-ity consumed for Weekdays and weekends. The activationfunction for hidden and output is Sigmoid transfer. Back-propagation type algorithm, which calculates the error rates,defined as the derivative of the squared error with respect toeach node’s output, recursively from the output backward tothe input nodes is used for learning.

The number of neurons for representing these variables is5, 3, 4, 4 and 240 respectively. The training process is shownin (Fig. 15).

Fig. 14: Adaptive Neuro-Fuzzy Inference System

Fig. 15: Neural Network Training

VII. DEFUZZIFICATION

This is the final level, at this point output value from theneural network is a fuzzified in the range [0 - 1]and it willindicates the degree of membership function of the outcomein the range [Xmin -Xmax]. This result is defuzzified byconverting it in this pre-specified range to obtain the dailyload value (in MW) for each week of seasons.

X = Xmin +Xn ∗ (Xmax −Xmin) (6)

where Xn is the predicted output of the NN and X is thecorresponding load value in MW.

VIII. RESULTS

This fuzzy neural network has been implemented usingMatlab (Fig. 16). The knowledge base and the membershipfunctions along with 240 rules were train by 10 epochs andError tolerance 0.By repeatedly Fine training from observa-tions (Fig. 17) hybrid system giving better results by usingclimate conditions.

Page 6: Electricity Demand Forecasting Using Fuzzy-Neural Network

Fig. 16: Neuro-Fuzzy Designer

Fig. 17: Training Result

IX. CONCLUSIONS

A forecasting technique based on ANN and FNN resultsform this hybrid technique approach can be used to elec-tricity load forecasting with greater accuracy than the ANNtechnique. The FIS can be easily developed and modified toreflect changes in different seasons. The basic concepts of thishybrid technique(FNN) are clearly presented in this paper andto implement on a personal computer and allows for operatorintervention is very simple. The results obtained based onweather parameters are found to be very accurate and superiorwhen compared to the ANN approach.

REFERENCES

[1] A.Kumar Singh, Ibraheem, S. Khatoon, Md. Muazzam, ”An Overviewof Electricity Demand Forecasting Techniques, National Conferenceon Emerging Trends in Electrical, Instrumentation and CommunicationEngineering, vol. 3, no.3, 2013.

[2] R. Xu, G. K. Venayamoorthy, and D. C. Wunsch, Modeling of generegulatory networks with hybrid differential evolution and particle swarmoptimization, Neural Networks, Vol. 20, pp. 917 927, 2007.

[3] Azadeh, S. F. Ghaderi, S. Tarverdian, and M. Saberi, Integration ofartificial neural networks and genetic algorithm to predict electrical energyconsumption, Applied Mathematics and Computation, Vol. 186, pp.17311741, 2007.

[4] F. Hobbs, S. Jitprapaikulsarn, and S. Konda, Analysis of the value forunit commitment of improved load forecasts, IEEE Trans on PowerSystems, Vol. 14, no. 4, pp. 13421348, 1999.

[5] S. A. Billings, and X. Hong, Dual-orthogonal radial basis functionnetworks for nonlinear time series prediction, Neural Networks, Vol.18, pp. 479493, 1998.

[6] L. Yue, Y. Zhang, Q. Zhong, and Z. Wang, Mid-long term load forecast-ing based on fuzzy logic clustering neural network approach, Journalof North China Electric Power University, Vol. 35, no. 2, pp. 4246, 2008.

[7] A. Kazemi et al., A Multi-Level Artificial Neural Network for GasolineDemand Forecasting of Iran, In Second International Conference onComputer and Electrical Engineering, pp. 61-64, 2009.

[8] A. Ghanhari, A. Naghavi, S.F. Ghaderi and M. Sabaghian, ArtificialNeural Networks and regression approaches comparison for forecastingIrans annual electricity load, Proceeding International Conference onPower Engineering Energy and Electrical Drives. pp. 675-679, 2009.

[9] Sackdara, et al., Electricity Demand Forecasting of Electricity DU Lao(EDL) using Neural Networks, TENCON 2010, pp. 640-644, 2010.