Research On Fuzzy Reasoning Algorithm Based On B-spline Membership Function

Fuzzy reasoning is the core of fuzzy systems and fuzzy control, and it is paid widespread attention. In the fuzzy inference system, it has been a difficult issue to determine the fuzzy membership functions and to establish fuzzy inference rules, and it's a key part of the fuzzy inference system. B-spline function which has excellent mathematical properties is widely used in engineering and research projects. Some scholars introduce the B-spline function into neural network system and constructe a fuzzy neural network system with B-spline transfer functions. However, the research on the combination of B-spline function and fuzzy inference system, including the reasoning of B-spline fuzzy membership function and the construction of B-spline fuzzy inference system are still relatively small.The starting point of this dissertation is the fuzzy reasoning module designing and implemention in decision support system. In order to achieve that purpose, issues including B-spline fuzzy membership function determining, controls of B-spline membership function fitting error and the implementation of B-spline fuzzy reasoning algorithm were studied. Its main contents are as follows:(1) Based on the introduction to the basic concept and theory of fuzzy reasoning, some main reasoning algorithms has been analyzed in this dissertation and the conventional methods for determining the membership function have been illustrated at the same time.(2) B-spline function and its nature are introduced. In this dissertation, F statistical method and least squares fitting method are proposed to extract fuzzy membership function from dataset. Adaptive iterative fitting error control method used among adjacent points is proposed. Experiments show the effectiveness of these methods.(3) From the engineering point of view, reasoning algorithm for CRI reasoning method based on B-spline fuzzy membership function is studied and its computation process for finding fuzzy rule's R value is induced in detail and, its final mathmatical format result is got in this dissertation. And, reasoning algorithm for similarity reasoning method based on B-spline fuzzy membership function is studied too and its computation process for finding fuzzy rule's R value is induced detailly and its final mathmatical format result is got. The reduction and conflict set of fuzzy rules in fuzzy inference system are discussed too. (4) Finally, application of the fuzzy reasoning in disaster relief decision support system has been studied.