Research On A Neural Network Fault Diagnosis Approach For Analog Circuits

With the development of research on fault diagnosis of analog circuits in the past more than three decades, a series of diagnosis theory and methods has been formed. Because of the complication and difficulty of fault diagnosis for analog circuits, there are a lot of limitations and poor practicable in those theories and their applications. Especially for linear and nonlinear analog circuit fault diagnosis as well as a large scale integrated circuit, there are no generally proper maths model and common simulation program for networks. So the unified method is lacking for computing fault character of circuits. As increasingly maturation of neural networks theory and its application, it has been becoming a major research project to apply fault feature extraction and neural networks to diagnosis faults. This dissertation based neural networks profoundly studies neural networks design and fault feature extraction methods and neural networks ensemble methods to diagnosis faults. Different power supply leads to different networks.These networks are parallel.two algorithms are proposed to combine the network diagnosing faults. The system is capable ofdiagnosising multiple faults even if trained on single ones.The creative work of this dissertation includes three aspects1. Choose each different power supply configuration to resolve the ambiguity in the presence of faults which are indistinguishable and can not be detected.Train different networks and two algorithms are proposed to combine several neural...