The Research Of Fault Diagnosis Of Analog Circuits Based On Principal Component Analysis And Probabilistic Neural Networks

The applications of large-scale integrated circuits allow the scale and the structure of a network to be more and more functioned and modularized with the development of modern electronic industry. So it is an urgent subject in practical project to study how to identify the fault sub-circuits and the fault elements correctly by using modern diagnosis technique for large-scale circuits with tolerance, and it is also a key step for practical application of fault diagnosis for analog networks theory and technology.After more than 20 years of development, analog circuit fault diagnosis has been formed a series of diagnosis theories and methods. However, due to complex theory itself and not universal practical methods of diagnosis at present, it's far from application prospects and people's expectations.With the technology development of artificial neural networks(ANNs),it's widely applied to Modeling and identification of non-linear system.The variety of faults in analog circuit makes the number of training samples of neural network greatly increase. Structure of BP network tends to be complex and training rate is greatly reduced. Against the shortcomings of Back-propagation Neural Network (BPNN),which include slow learning speed of convergence and the nature which is easy to fall into local minimum value, Probabilistic Neural Networks(PNN) conbined with Principal Component Analysis(PCA) based diagnostic method for faults of analog circuit with tolerance is proposed.The classical Probabilistic Neural Network based on Radial Basis Function Network is a kind of simple. Compared with classical BP neural network, there are three advantages. Simply network learning process and quickly training. Excellent fault-tolerant network, outstanding pattern classification ability, good convergence. Flexible network architecture design.The paper concentrates on main works as follows:1. On fault feature extraction method for analog circuits. The feature parameters of the node voltage response are compressed using principal component analysis (PCA). It has many good properties, such as simplifying the structure of NNs, improving the training speed and fault coverage.2. On the general methods and diagnosis steps of analog circuit fault diagnosis based on neural network, PNN networks constructure design, training algorithm and the advantages in fault diagnosis.3. Combined the advantages of PCA with PNNs, chooseing a typical pure resistor circuit, PSPICE and MATLAB are applied to the simulation of fault diagnosis process specificly.The experimental results show that the fault feature can be classify and identify effectively by PNNs, it is good for rapidly fault location, the purpose of diagnosis and fault coverage.In a word, based on the methods of PNNs, that PCA extracted fault features from analog circuit is effectively the solution of analog circuit fault data redundancy and fault diagnosis speed rate.