Simulating fuzzy systems (Q705678): Difference between revisions
From MaRDI portal
Changed an Item |
Set profile property. |
||
Property / MaRDI profile type | |||
Property / MaRDI profile type: MaRDI publication profile / rank | |||
Normal rank |
Latest revision as of 02:02, 5 March 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | Simulating fuzzy systems |
scientific article |
Statements
Simulating fuzzy systems (English)
0 references
11 February 2005
0 references
The first objective of this book is to explain how many systems naturally become fuzzy systems. The second objective is to show how regular (crisp) simulation can be used to estimate the alpha-cuts of the fuzzy numbers used to analyze the behavior of the fuzzy system. Consider any system that uses probability distributions in its description. Quite often probabilities and/or parameters in the probability distributions (like the mean of the normal) have to be estimated from data or expert opinion. Usually one obtains point estimates of these unknown probabilities and/or parameters for input to the system. However, it would be better to use confidence intervals. But which confidence interval should one use? The author suggests in Chap. 3 to use an infinite set of confidence intervals, stacked up one on tope of another, producing fuzzy number estimators. Fuzzy estimators produce fuzzy probability distributions in Chap. 4. In Chap. 4 the author looks more closely at the fuzzy binomial, fuzzy Poisson, fuzzy normal, fuzzy exponential and the fuzzy uniform distribution. These are the fuzzy distributions that will be used in the rest of the book. Now the system is described by fuzzy probabilities and fuzzy probability distributions. It is a fuzzy system (Chap. 5). Things one want to compute, like response time, throughput, etc. all become fuzzy. Consider any system whose description employs probability theory. Assume that some of these probabilities must be estimated from crisp data, and/or some of the parameters in the probability densities must be estimated from crisp data. Using the fuzzy estimators of Chap. 3, one can obtain fuzzy probabilities and fuzzy distributions. For some fuzzy systems the computation of the fuzzy descriptors becomes difficult. A faster way to approximate these fuzzy numbers than multiple calculations using the extension principle is needed. In Chap. 7 it is argued that if there is a one-step function, and there usually is, that can produce the crisp form of the system descriptor, then simulation can be used to approximate the alpha-cuts of the extension principle extension of this function. Simulation may also be used to approximate some multi-step fuzzy calculations. But to be able to use simulation to approximate the alpha-cuts of these fuzzy system descriptors there are certain optimization problems to solve. This is discussed in Chap. 8 and continues in Chaps. 9-26. Also, the author shows how to use simulation to estimate the fuzzy sets/numbers needed to describe the fuzzy system in a great variety of applications in Chaps. 9-26. The applications encompass production lines, emergence rooms, inventory control process failure, oil tanker, supermarket, bank teller and project network model.
0 references
fuzzy sets
0 references
fuzzy estimation
0 references
fuzzy probability
0 references
fuzzy systems simulation
0 references