Research On Supervisory Control Theory Of Fuzzy Discrete Event Systems With Constraints

Discrete event systems (DES) exist in real world widely, whose state revolutions are driven by a serial of discrete events and operation process should follow the complicated logic rules defined by humans. As a kind of artificial systems, DES have a lot of uncertainty. As an extension and generalization of DES, fuzzy discrete event systems (FDES) describe fuzzy uncertainty of DES effectively by fuzzy states and fuzzy events. For a class of FDES with constraints, whose transitions of states need to meet certain state constraints and actions on events also need to meet input or output constraints, this dissertation studies the problems of modeling and supervisory control of this kind of systems from the sides of a fuzzy automata and fuzzy logic rules based on the definition of an effective fuzzy state and proposes the corresponding supervisory control theory. Furthermore, aiming at the problem of determination for the membership functions of fuzzy sets which have a closed relation with the realization of the supervisory control theory, it presents a determination method of membership functions based on the fusion of multi-feature similarity. The works of this dissertation can be cataloged as follows:By the definition of a set of effective fuzzy states and its evaluation function used for characterizing the meeting degree of the constraints in the transitions of states, this dissertation presents a modeling method with the constraint automata based on the traditional fuzzy automata. By the definition of fuzzy events and the controllability of fuzzy event strings, it also presents the properties of intersection and union operations for the controllability of fuzzy languages and the definition and operation properties of supremal controllable fuzzy sublanguage and infimal controllable fuzzy superlanguage. Furthermore, this dissertation builds the supervisory control theory for FDES with constraints from the sides of a conventional supervisor and a constraint supervisor.By analyzing the five level structure of the traditional rule-based description method, this dissertation formulates a dynamic hierarchical structure. By extending ordinary factors to fuzzy factors in the logic rule description language, it defines a generalized logic rule description language and a generalized hierarchical rule-based model which can be used to model DES and FDES. Furthermore, a constraint generalized hierarchical rule-based model can be built by assigning a constraint degree for execution of logic rules.By the associating fuzzy logic rules with fuzzy events, this dissertation formulates the controllability of fuzzy logic rule strings and the fuzzy language and presents a method for constructing a fuzzy supervisor. It proposes an adaptive closed-loop supervisory control structure which fuses a fuzzy supervisor and a fuzzy controller for FDES and presents the detail realization methods for the fuzzy inference process, the defuzzification process and the dynamic mechanism of a fuzzy logic rule base. Thus, a new FDES supervisory control theory has been built on a constraint generalized hierarchical rule-based model.Based on the analysis and summary of the shortage of the existing membership functions of fuzzy sets, this dissertation defines representative elements and the feature vector of fuzzy sets and a distance between an object element and a fuzzy set. According to the distribution of feature elements in the feature space, it presents a method for computing the distance. By the normalization of distances and the definition of the similarity between an object element and a fuzzy set, it proposes a determination method of membership functions based on the fusion of multi-feature similarity.In a word, the works and finds of this dissertation around FDES with constraints not only meet the nees of modeling and control of this kind of systems, but also improve and develop the theory system of FDES.Furthermore, they lay the foundation for the better engineering applications of the relevant theories and methods.