| Currently, fuzzy systems and control has been widely used in industrial areas. In sociology, economics, environmental science, biology, medicine and other fields, it has also many successful applications, some of which are even groundbreaking exploratory applications. But fuzzy systems and control has always been a controversial field, the theoretical foundations still need to improve, the practical applications also need further study. In this dissertation, the inherent flaws of fuzzy inference methods which have been widely used in engineering fields are investigated: rule output sets are often determined only by one of the antecedents, ignoring the impacts of other antecedents; or, although the fuzzy inference considers all the antecedents, the results are not produced by each of them in accordance with its degree of influence on the consequent. To solve these problems, a consequent-oriented fuzzy inference mechanism is proposed for type-1, interval type-2, and general type-2fuzzy logic systems. Further, the common grounds of the inference of the three fuzzy logic systems are abstracted to introduce the concept of interrelated fuzzy sets.The innovative work of the dissertation is as follows. A fuzzy inference mechanism conducted in an environment where fuzzy sets are interrelated with each other is proposed. The method can overcome the defects of missing information caused by traditional fuzzy inference methods which are carried out under a circumstance where antecedents and consequents are independent of each other. In addition, it can capture more fuzzy information of the fuzzy rules, and can provide more choices for one to design a fuzzy logic system (including type-1, interval type-2, and general type-2). The main work can be summarized as follows.(1) A concept of object-oriented transform (OOT) on fuzzy sets is introduced based on the consideration that the correlation information between antecedents and consequents are not used in most fuzzy inference methods. A fuzzy set transformed by OOT includes the correlation information of reference objects which are chosen for some certain situation. A consequent-oriented fuzzy inference approach is proposed by replacing the antecedents with their OOT fuzzy sets in the process of fuzzy inference, which is suitable for type-1, interval type-2, and general type-2fuzzy logic systems.(2) The correlation information between fuzzy sets is investigated systematically. For the case that the correlation between fuzzy sets can be described by a crisp number, a COFI with CRD (consequent-oriented fuzzy inference with crisp relation degree) is addressed and is applied to type-1, interval type-2, and general type-2fuzzy logic systems.When the correlation is hard or inappropriate to be defined as a crisp number, the above COFI with CRD is unable to bring such fuzzy correlation into the fuzzy inference. For this case, the concept of OOT is extended and the compound type-2fuzzy set model is proposed, based on which, a COFI with FRD (consequent-oriented fuzzy inference with fuzzy relation degree) is presented and is applied to type-1and interval type-2fuzzy logic systems.(3) The common characteristics of the COFI with CRD and FRD for three kinds of systems are abstracted to be a concept of interrelated fuzzy set. Such fuzzy sets can make one study the fuzzy sets and fuzzy logic systems under an environment in which fuzzy sets are interrelated with each other. An interrelated type-1fuzzy set, interrelated interval type-2fuzzy set, and interrelated general type-2fuzzy set are proposed for type-1, interval type-2, and general type-2fuzzy logic systems, respectively. Some basic concepts of ordinary sets, such as, inclusion, union, intersection, complement, are extended to the interrelated fuzzy sets. Meanwhile, some of their properties and applications to the above three kinds of fuzzy logic systems are explored in a preliminary way. |
PDF Full Text Request