Study On The Intelligent Design Method For Microwave Filter

The modern design theory was introduced in the design of the planar microwave filters. This thesis focuses on the representation, capturing, the reasoning of the knowledge, and the comprehensive evaluation to the design schemes in planar microwave filter intelligent designing and the corresponding solutions are proposed, which is of important of significance to realize the intelligent design for microwave filter products and to the intelligent design theory.A product model with frame data structure was set up to change finding the design solution to pattern matching or recognition. Semantic network, frame, rule, procedure were combined to represent the knowledge in different structures, and case-based reasoning, rule-based reasoning and model-based reasoning were used to find the conceptual design solution at the basis of non-precise logic. Analogical inference, fuzzy clustering analyzing and statistics methods are used to capture the knowledge from those existing success cases, the self-learning was realized.A generalized closeness degree algorithm was proposed suitable for three kinds of variables of general real number, the section on the real number domain and fuzzy number. It was used in conceptual design based on fuzzy pattern recognition and fuzzy case-based reasoning to deal with fuzzy information effectively.A comprehensive evaluation model with weighted evidence was introduced to deal with various uncertainty factors. From the output of this model, the measurement of the certainty or uncertainty of the evaluation result can be obtained, which is the basic evidence to adopt or discard the design scheme, to find the optimal design or to revise the existing design.