Analog Circuit Fault Diagnosis Research Based On Fusion Of Multi-signal Characteristics

With the rapid development of electronic technology, the circuit scale and integration are unprecedented developed. Analog and mixed circuit module is still irreplaceable part for a long time, therefore, fault diagnosis of analog circuits has important theoretical significance and application value. But until now there is no effective use of the analog circuit fault diagnosis method.This paper introduces the analog history and current situation of circuit fault diagnosis, and feature extraction and waveform classification method is also been researched. Based on the detailed analysis of the signal waveform classification's research and applications in various fields, we focus on the waveform template matching and auto-correlation analysis method.Considering the shortcomings of the DC fault dictionary and the specialty of the circuit data collection in actual circuit automated test, in order to increase signal discrimination, we introduced a fusion of multi-signal characteristics of analog fault diagnosis method, which encompasses signal feature extraction method, the waveform classification and recognition method, multi-dimensional vector division of fuzzy sets, transient fault dictionary generation and optimization, as well as the organizational structure of the standard interface file.Finally, we have implemented the integration of multi-signal characteristics of the transient fault dictionary module, and the module's software architecture is introduced. We use autocorrelation analysis, combined with zero-crossing and curve fitting, histogram, discrete cosine transform method to extract signal features. Based on the results of feature extraction we establish waveform templates, calculate the similarity, and combine statistical analysis to achieve waveform classification. On this basis, the multi-dimensional feature vector is established, and we complete the division of fuzzy multidimensional vector and transient fault dictionary generation and optimization. After generating a standard analog circuit fault diagnosis interface file, we use it in ATE test. We first propose multi-dimensional feature vector fuzzy set and transient fault dictionary in this article, and the transient fault dictionary is better than DC fault dictionary.