Applied Research On Fuzzy Neural Network Of Traditional Chinese Medicine Rheumatoid Arthritis Based On Genetic Algorithms

Fuzzy technology, neural network and heredity algorithm are three information science of computation intelligence. Moreover, they are important tools used to develop and implement intelligent control systems. Today more and more researchers have concerned about a problem that how to combine fuzzy technology, neural network and heredity algorithm. The thesis is aimed at helping people to make an accurate diagnosis for rheumatoid arthritis from the point of intelligence control. The contents of the thesis areas follows.First, the thesis introduces the theory and basis of heredity algorithm, common heredity algorithm, adaptive proportion and the method of combining fuzzy logic and neural network.. In addition, the thesis provides a composing method of fuzzy neural network based heredity algorithm. According to this composing method, thesis provides the steps of optimizing network architecture and parameters : 1. initializing parameters 2.optimizing network architecture. 3. optimizing network parameters.Second, thesis discusses the feasibility of applying heredity algorithm and neural fuzzy technology to traditional medicine diagnosis. Traditional medicine is a science full of fuzziness. All kinds of phenomenon when people is ill, such as the character of body , tongue and pulse are described using fuzzy language. The relationship among all kinds of pathology is also fuzzy, and it is difficult to find a standard to describe these pathology. However, it is necessary to diagnose accurately in the clinic. So searching for accuracy from fuzzy is a important problem in the process of traditional medicine diagnosis.The thesis presents a structure of diagnostic system based on fuzzy neural network for rheumatoid arthritis. The structure ofthis system has become simple after using heredity algorithm to compress the diagnostic data, so computation has become easy and cost less time. The diagnostic system has proven to be very useful in the diagnosis for rheumatoid arthritis.Finally, the thesis comes up with an auxiliary prediction and diagnosis system for rheumatoid arthritis, which is developed based on Visual C++, SQL Server2000 and matlab6. 0. At the end of thesis, author indicates the research direction in the next stage and presents a method of on-line detecting pathology parameters of rheumatoid arthritis.