Fuzzy Control Rules
Fuzzy Control Rules is the core of the fuzzy controller, and its correctness directly affects the performance of the controller, its number The polyworld is also an important factor in measuring the performance of the controller.
Fuzzy control rules are part of the knowledge base in the fuzzy controller, and the fuzzy control rules are based on language variables. Language variables are "large", "medium", "small", etc., each fuzzy subset suggests that the exact value of the basic arguments belongs to the extent of the fuzzy subset. Therefore, in order to establish a fuzzy control rule, the exact value on the basic discussion is required in accordance with the membership function to each fuzzy subset, so that the exact value is replaced with the language variable (large, medium, small, small, etc.). This process represents the fuzzy division of the observed variable and control amount during the control. Since each variable ranges the value range, each of the basic arguments is first mapped to a standardized argument in different correspondences, respectively. Typically, the correspondence is taken as a quantization factor. For ease of processing, separate the standards and other separation, then fuzzy divide, define the fuzzy subset, such as NB, PZ, PS, etc. The same fuzzy control rule library, different vague division of basic aromas, and the control effect is also different.
As shown in Figure 1, the basic structure of the fuzzy controller includes a knowledge base, fuzzy reasoning, ambiguity, and output quantification.
(1) Knowledge Base
Knowledge Base includes a fuzzy controller parameter library and a fuzzy control rule base. Specifically, corresponding relationships, standard arguments, fuzzy subsets, and affiliate functions of each fuzzy subset have great impact on control. These three types of parameters are the same importance, so they consume them into the fuzzy control rule library to form a knowledge base with the fuzzy control rule bank.
Change the accurate input amount into ambiguated amount F has two methods:
a. Conversion of exact amount to standard theory Fuzzy single point set on the domain.
The exact amount of X is converted to the basic element on the standard aroma X by the correspondence relation to the corresponding relationship G.
b. Convert the exact amount to the fuzzy subset of standard arises.
The accurate amount is converted to the basic element on the standard discussion in the corresponding relationship, and the fuzzy subset of the maximum membership degree, that is, a fuzzy subset of the accuracy.
(3) Fuzzy reasoning
The most basic fuzzy reasoning form is:
premise 1 if A THEN B
Prerequisites 2 IF A '
Conclusions Then B'
where A, A 'is the fuzzy subset of the domain u, B, b' is the fuzzy subset of the field V. Prerequisites 1 are called the fuzzy implications, remember to be A → B. In practical applications, it is generally performed for each rule, and then the final reasoning result is obtained by the total reasoning result.
The fuzzy subset of reasoning to be converted to an exact value to obtain the final control amount output Y. Commonly used two precision methods:
a. Maximum membership method. In the reasoning the fuzzy subset, select the average of the largest standards for the largest level as the accuracy.
b. gravity method. The grade function of the affairing subset of fuzzy subsets and the criteria of the center of gravity of the horizontal coordinate are as accuracy results. After obtaining the accuracy of the reasoning result, it should also be based on the corresponding relationship to obtain the final control output Y.
Fuzzy control rules Source
Fuzzy control rules to obtain method:
(1) Experts and knowledge
Fuzzy control Expert system, expert experience in the control system, and knowledge in its design. Humanity is judged in the daily life, using language qualitative analysis than numerical quantitative analysis; and fuzzy control rules provide a natural architecture that describes human behavior and decision analysis; experts' knowledge can usually be expressed by IF .... THEN .
By inquiring experienced experts, obtaining the system's knowledge, and the knowledge is changed to if ....Then's type, the fuzzy control rules can be constituted. In addition, in order to obtain the best system performance, it is often necessary to use a mission method multiple times to correct the fuzzy control rules.
(2) Operator's operating mode
popular expert system, its idea only considers knowledge. Experts can subtly operate complex control objects, but to log consider experts, this requires consideration of techniques to consider. Many industrial systems cannot do correct control with general control theory, but skilled operators can successfully control these systems without mathematical mode: this inspires us to record operator's operating mode, and organize it to IF ... .Then's type can constitute a set of control rules.
In order to improve the performance of the fuzzy controller, it must make it self-learning or self-tissue, enabling the fuzzy controller to increase according to the set target Or modify the fuzzy control rules.
Fuzzy control rules
Fuzzy control rules can be divided into two:
(1) status assessment fuzzy control rules< P> State Evaluation Fuzzy Control Rules Similar to human intuition, it is used by most fuzzy controllers, its forms are as follows:
(2) Target evaluation Fuzzy control rule
target The Object Evaluation Fuzzy Control Rule can evaluate the control target, and predict the future control signal, the form is as follows:
Fuzzy control rule table
actual application fuzzy control, the initial problem It is the design of the controller, that is, how to design a fuzzy control method.
Inside the fuzzy controller, this table is a fuzzy rule table, where e represents the error, EC represents the error change, u is the output variable, the first column (NB, NM, ....) is E The language variable, the same, the first line is the language variable of EC. It has three methods for establishing this fuzzy rule:
(1) is based on control of engineering knowledge and maturity.
(2) is based on the actual control process of the operator.
(3) process fuzzy model.