Research on Fault Diagnosis Expert System Fusing the Neural Network Knowledge

Abstract

For a complicated system based on high technology, once a part breaks down, the entire system can not work normally. Moreover, due to the complexity of its structure and fault causes, the fault diagnosis of the system is also complex and indeterminate, a single test equipment can hardly finish a difficult diagnose task, and fault diagnosis expert system can resolve these problems effectively. The traditional diagnose expert systems have many problems such as the bottleneck of knowledge acquisition, the fragility of knowledge, the pool ability of self-study, the inefficient reasoning, and the monotonicity of reasoning, so there are certain limitations. But the artifical neural networks technology is a new system, it is an mathematical model that applies the structure like the joint of synapses in hypothalamic neurons, which has the strong ability to study, and can learn from samples, obtain knowledge, store it in the network in the form of weight and threshold, and it is easy to implement the parallel processing, has the character of association memory, own the better robust. it ability of adaptive self-study is manifested mainly in adjusting the weight of network according to the change of enviroment by learning algorithms, so as to adapt to the environmental change. But the neural network can not explain its own reasoning. Therefore we will apply the neural network to the expert knowledge system, which can make them learn each other's good points mutually for common progress, constructing the new neural network expert system. The system is applied to the power fault diagnosis, achieving good results.