Circuit Fault Diagnosis Based On Information Entropy Feature Extraction

As the modern electronic technology develops, the reliability and the security ofthe circuits are highly required. Therefore, circuit fault detection and diagnosis havebecoming increasingly important. Information entropy is a quantitative evaluation indexof the uncertainty of system status and has a strong ability to describe system innerinformation. This paper focuses on the application study of information entropy incircuit fault diagnosis, and puts forward non-contact signal acquisition method based onelectromagnetic radiation.Firstly, the paper introduces the theory and the development of circuit faultdiagnosis technology and information entropy analysis method, reviews the conceptsabout circuit fault diagnosis, and discusses the advantages and disadvantages abouttypical methods of fault diagnosis. Secondly, it presents several different entropy-basedfeature extraction methods after the introduction of the concept of information entropy,then applies these methods into circuit fault analysis with concrete examples. Finally,toward switching power supply circuit board, a complete diagnosis process is proposed,including non-contact based magnetic radiation signal acquisition, parameter extraction,signal processing about magnetic radiation signal, time-domain entropy and spectralentropy feature extraction, and the establishment of fault characteristics table and faultrules. Its biggest advantage is in solving the problem of the circuit board testabilitywhile making full use of signal information acquired. It needs fewer test nodes and isespecially suitable for the circuits lack of prior knowledge and circuit information. Inaddition, the study also puts up a non-contact software and hardware fault diagnosisplatform based on information entropy feature extraction. Diagnostic test results showthat the proposed entropy-based diagnostic method is effective and practical.