Circuits Fault Diagnosis Based On Virtual Instruments

The theory of analog circuit fault diagnosis is very important and significative, now it has become a hot studied project. The traditional methods of fault diagnosis are performed only if the faults of the circuits are those hard faults, such as open-circuit, short-circuit, etc. Those soft faults aroused by the tolerance of circuit components cannot be easily discovered. In this paper, fault feature extraction has been realized by analysis process of the circuit signal in time domain, frequency domain and time-frequency domain. Moreover, fault diagnosis methods based on neural network have realized successfully. The whole fault diagnosis system is based on Virtual Instruments which has realized a better man-machine interaction.In this paper, it is principally studied applying neural network to analog circuit fault diagnosis. The main contents are as follows. First, the intact diagnosis technique and its implement scheme are proposed. The paper investigates the design and algorithm of RBF neural network, and its feasibility applying to fault diagnosis is analyzed. Second, some fault feature extraction methods are studied. As the traditional fault diagnosis based on neural network trains only within one network, and the outputs of the network depend on the patterns of fault, namely, the fault diagnosis is realized just by one network. In this paper, several networks which have single-output are trained, the number of which is determined by the fault patterns, therefore, a system which synthesizes the fault in circuits is composed by all these networks. The paper compared the results of fault diagnosis used RBF neural network with k-means algorithm and statistic to classify the feature vectors when all the samples are known. At last, an actual circuit is selected to test these methods and process. The simulation results of the example show that the fault diagnosis methods based on neural network have good diagnosis effect and feasibility in tolerant circuit. In conclusion, the method of synthesis fault diagnosis by several single-output networks is better than the traditional method of fault diagnosis based on neural network.