أنت هنا

AN IMPROVED METHOD FOR THE DESIGN OF FUZZY CONTROLLERS

التبويبات الأساسية

Issam  M. KOUATLI

 

Univ.

Birmingham

Spec.

Mechanical Engineering

Deg.

Year

#Pages

Ph.D.

1990

174

 

In recent years there has been an increased interest in the application of fuzzy logic to the control of industrial processes. Fuzzy control is one type of "intelligent system" that can be used to imitate the action of human operators. The advantage of fuzzy control in comparison with conventional control techniques that are able to deal with uncertainty in the system. In practice, the mathematical model of the process to be controlled is often inaccurate and in other cases does not exist. In such cases fuzzy logic based control offers an effective strategy.

Before a fuzzy controller can be designed the process information has to be gathered. This is a difficult task due to the differences in expertise that individuals bring to the process. Also, human beings usually find it difficult to explain their actions in instances in which intuitive skills are used. Systemic analysis in the form of input/output diagrams have been suggested and used in this thesis as a useful tool in extraction of operator's knowledge.

In the language of fuzzy logic theory, the f first step in the design of a fuzzy controller is to define the fuzzy variables that relate to any of the inputs or outputs e.g. Small, Medium, Large are examples of fuzzy variables. In the terminology of Set Theory the possibility of all of the values that a control variable might have are often referred to as "Universe of Discourse" and thus partitioning the Universe of discourse of both the inputs and the outputs is required in order to define the fuzzy variables (Small, Medium ... etc). "Fuzzimetric Arcs" are introduced in this thesis as an effective and new technique of partitioning the universe in a logical and powerful way, which allows the selection of the optimum shape of the fuzzy sets used in the control system.

The implementation of fuzzy control to a multivariable system requires a substantial amount of memory, which often leads to memory overload. A strategy for avoiding this difficulty is suggested and discussed in this thesis where it is suggested to use Sub‑algorithms to avoid the complexity in developing the multivariable fuzzy algorithm, which provides a simpler technique to apply and consequently an equation describing the inference of a specific output from a number of inputs is derived. This is also simpler to use and has a higher level of confidence than other techniques.

Both of the techniques proposed i.e. the Fuzzimetric Arcs and the Multivariable inference strategy are built into an expert system for fuzzy control which has been termed "FuzCon" and is proposed as a means of providing the control engineer with (i) A guide to the suitability and implementation of fuzzy control technique. (ii) Development of the required fuzzy relations and their consequent inference from a specific inputs. (iii) If connected to a dynamic plant to provide feedback of the system performance, thus allowing the conditions to be modified to achieve a satisfactory result.

Control of a particular industrial process is implemented as a means of demonstrating the effectiveness of such techniques.