Risk Assessment Via Monte Carlo Simulation: Tolerances Versus … · 2003. 3. 10. · 1 Risk...
Transcript of Risk Assessment Via Monte Carlo Simulation: Tolerances Versus … · 2003. 3. 10. · 1 Risk...
1
Risk Assessment Via Monte Carlo Simulation: Tolerances Versus Statistics
B. Ross BarmishECE Department
University of Wisconsin, MadisonMadison, WI 53706
2
• A. C. Antoniades, UC Berkeley
• A. Ganesan, UC Berkeley
• C. M. Lagoa, Penn State University
• H. Kettani, University of Wisconsin/Alabama
• M. L. Muhler, DLR, Oberfaffenhofen
• B. T. Polyak, Moscow Control Sciences
• P. S. Shcherbakov, Moscow Control Sciences
• S. R. Ross, University of Wisconsin/Berkeley
• R. Tempo, CENS/CNR, Italy
COLLABORATORS
3
Overview
• Motivation
• The New Monte Carlo Method
• Truncation Principle
• Surprising Results
• Conclusion
4
Monte Carlo Simulation
• Used Extensively to Assess System Safety
• Uncertain Parameters with Tolerances
• Generate “Thousands” of Sample Realizations
• Determine Range of Outcomes, Averages, Probabilities etc.
• How to Initialize the Random Number Generator
• High Sensitivity to Choice of Distribution
• Unduly Optimistic Risk Assessment
5
It's Arithmetic Time
• Consider
1 + 2 + 3 + 4 + 5 + ! + 20 = 210
• Data and Parameter Errors
• Classical Error Accumulation Issues
(1 ± 1) + (2 ± 1) + (3 ± 1) + ! (20 ± 1) = 210 ± 20
6
Two Results
Actual Result Alternative Result
190 · SUM · 230
190 195 200 205 210 215 220 225 230
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
x
7
Circuit Example
+_
10 Ω
10 Ω
10 Ω5 Ω
R
R
R
R
2
3
I1I 2
I3
I 4
x
1
1 VVout+ _
Output Voltage
• Range of Gain Versus Distribution
8
More Generally
k1
b
k2
2b
k3
3b
k4
4
m =12m =11 m =13 m =14
b1
y
F
• Desired Simulation
9
Two Approaches
• Interval of Gain
• Monte Carlo
0.45 0.5 0.55 0.6 0.65 0.7 0.75 0.8 0.85 0.90
100
200
300
400
500
600
700
800
900
1000
Student Version of MATLAB
• How to Generate Samples?
Gain Histogram: 20,000 Uniform Samples
10
Key Issue
What probability distribution should be used?Conclusions are often sensitive to choice of distribution.
• Intractability of Trial and Error
• Combinatoric Explosion Issue
11
Key Idea in New Research
Classical Monte Carlo New Monte Carlo
RandomNumber
Generator
F
Samples of q
System Model
Samples of F(q)
Statistics of q
RandomNumber
GeneratorFSamples of q
System Model
Samples of F(q)
Statistics of q
NewTheory
Bounds on q
12
The Central Issue
What probability distribution to use?
13
Manufacturing Motivation
14
Distributions for Capacitor
-20 -15 -10 -5 0 5 10 15 200
0.01
0.02
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
15
Interpretation
16
Interpretation (cont.)
RandomNumber
GeneratorFSamples of q
System Model
Samples of F(q)
Statistics of q
NewTheory
Bounds on q
17
Truncation Principle
− r1 − t1 r1t1 − t2 t 2− r2 r2x1 x2
f x1 1( ) f x2 2( )
18
Example 1: Ladder Network
...
...
+
-
VoutVin
R1
R3
R2
R4
R6
R5
Rn-2
Rn
Rn-1
• Study Expected Gain
19
Illustration for Ladder Network
20
Example 2: RLC Circuit
Consider the RLC circuit
1
L
I
+
-
CC R V1 2 021R
21
Solution Summary For RLC
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5x 10 -7
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
x 10 -6
t1
t2
22
Current and Further Research
• The Optimal Truncation Problem
• Exploitation of Structure
• Correlated Parameters
• New Application Areas; e.g., Cash Flows