1991 Fiscal Year Final Research Report Summary
Improvement of Fuzzy Control Methods and Development of Tools for Fuzzy Controls
| Project/Area Number |
02680032
|
|---|
| Research Category |
Grant-in-Aid for General Scientific Research (C)
|
| Allocation Type | Single-year Grants |
|---|
| Research Field |
Informatics
|
|---|
| Research Institution | Osaka Electro-communication University |
|---|
Principal Investigator |
MIZUMOTO Masaharu Osaka Electro-Communication University, Faculty of Engineering, Professor, 工学部, 教授 (40029541)
|
|---|
| Project Period (FY) |
1990 – 1991
|
| Keywords | Fuzzy reasoning / Fuzzy controls / Min-max-gravity method / Product-sum-gravity method / PID control / Singleton type fuzzy reasoning / FLIL / Tool for fuzzy controls |
|---|
| Research Abstract |
1. New fuzzy reasoning method for fuzzy control, called product-sum-gravity method, is proposed and compared with Mamdani's fuzzy reasoning method known as min-max-gravity method. It is shown that the product-sum-gravity method is intuitive in nature and can get better control results than min-max gravity method. Moreover, several new fuzzy reasoning methods such as (min/product) -max-gravity method, bounded product-algebraic sum-gravity method and so on are proposed and can get better control results than min-max-gravity method.
2. Emphatic effects on fuzzy inference results are observed when the identical fuzzy control rules are used twice or more in the fuzzy reasoning process under product-sum-gravity method. Inhibitory effects are also realized by employing fuzzy control rules whose consequence part is characterized by a negative valued membership function. It is shown that good control results are obtained by using such fuzzy control rules of emphatic and inhibitory types.
3. The properties of multi-fold fuzzy reasoning and multi-stage fuzzy reasoning are investigated. Inference results of multi-fold fuzzy reasoning are obtained by using several implications and it is shown that implication rules based on Godelian sequence and standard sequence can get good inference results. The formulation of n-stage syllogism and multi-fold multi-stage fuzzy reasoning are executed under Mamdani's fuzzy reasoning method.
4. A tool for the implementation of fuzzy controls is developed by using the results obtained above. It is written in C language and can manipulate fuzzy variables, membership function, fuzzy control rules and fuzzy inference engine. Moreover, we developed new tool written in FLIL language (Fuzzy Logic Interpreter Language) which is implemented for only fuzzy controls. This new tool can treat membership function and fuzzy control rules more easily.
|
