Application Of Analog Circuit Fault Diagnosis Based On Data Reduction Algorithm

With the rapid development of electronic technology,lots of new methods of simulating circuit fault diagnosis are proposed.Many intelligent diagnostic algorithms have the advantage of not needing to master circuit knowledge,but the diagnosis effect of this algorithm has obvious relationship with the trained model.Taking the diagnosis of the nonlinear analog circuit as an example,the first step in this process is to collect a comprehensive amount of fault data.Due to the tolerance of a large number of components such as resistors and capacitors,the acquired data tends to exhibit high dimensional and nonlinear structural distribution.How to reduce the dimensionality of these high-dimensional data and obtain the optimal fault features from them have become a central issue in recent years.In this paper,the problem of fault diagnosis for complex analog circuit based on pattern recognition was discussed.The core is to discuss how to construct typical feature.Based on the wavelet theory and the typical dimension reduction algorithm,the optimal features are analyzed comprehensively.And the improved clonal selection algorithm is used to diagnose and verify the circuit.The main contents of this study are as follow:1.Study the feature extraction method based on wavelet theory.The wavelet packet decomposition methods are analyzed in detail.Based on the selection of wavelet basis function and the selection of wavelet decomposition level,this paper proposes a new way to reduce the problem of exponential increase in data volume.2.This paper analyzes the characteristics and applicable scope of each dimension reduction algorithm,combines the characteristics of analog circuit fault diagnosis engineering application,and selects suitable algorithms for engineering applications through theoretical analysis and experimental comparison.For the algorithm with better dimension reduction performance,the selection of parameters and the ratio of dimension reduction were further analyzed.Focus on the problem of low efficiency and complicated calculation of the current fault diagnosis method,in this paper,a fault diagnosis scheme for complex analog circuits based on wavelet packet decomposition and data reduction algorithm is proposed.The accuracy and computational complexity of the diagnosis were used as evaluation criteria.It is proved that this method can be used for fault diagnosis and successfully applied to the board level test of complex circuit.