The Research Of Analog Circuits Fault Diagnosis Based On Fractal Theory And Fuzzy Neural Network

The analog fault diagnosis research already formed a series of diagnoses theory and the method, since 20th century 60's. However, complicated diagnosis question make these method's application poor. Along with development of electrical technology, especially VLSI and mixed circuit, it brings up new challenge to fault diagnosis of analog circuits. It is in great need of new theory and approaches to it.In recent years, researchers paid more attention to ANN than before. Fuzzy neural networks technology application is a new approach in the arena of the analog fault diagnosis. In the paper, through method combined with Fuzzy theory and Artificial Neural Network, an approach for the fault diagnosis of analog circuit based on clustering and fuzzy neural network is presented, and elaborated basic design thought and the algorithm realization of diagnosis system. Because ANN can be used without the specific model of object, and it stores useful information as distributed manner, we usually make use of the topological structure and weights of neural network to realize nonlinear mapping, which make those weights full of meaning. As it is known, we hope to get the universal best answer, so in the processing of weight training, we adopt an unsupervised clustering algorithm to avoid the defect of BP algorithm, which is very easy to get into the local best answer, so this method surpass traditional diagnosis method. The simulation experiment had indicated this method in the accurate rate and the diagnosis speed in the fault diagnosis compares the traditional method to have the further enhancement.Based on fractal theory, fractal diagnosis method is developed through characteristic parameter of calculating dimension of signal fractal. Based on the principles of Fuzzy theory and neural network as well as the conception of fractal dimension subordinating degree, fractal fuzzy neural network, input to which is fractal dimension subordinating degree, is constructed. A fault identification approach for nonlinear analogue systems is presented and it is examined by simulated tests. According to the principles above, an intellective diagnosis system is formed and studied. This research is the new exploration to methods of fault diagnosis of analog circuits.