SIMULATION CIRCUIT FAULT DIAGNOSIS METHOD BASED ON CONTINUOUS WAVELET ANALYSIS AND ELM NETWORK

• Main Authors: HE, Wei, YUAN, Zhijie, HE, Liulu, SHI, Luqiang, HE, Yigang, ZHAO, Deqin, LUO, Qiwu, LI, Zhigang, SHI, Tiancheng, WANG, Tao
• Format: Patent
• Language: Chinese; English; French
• Published: 05.07.2018
• Subjects: MEASURING; MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES; PHYSICS; TESTING

A simulation circuit fault diagnosis method based on continuous wavelet analysis and an ELM network, comprising: data acquisition: performing data sampling on output responses of a simulation circuit by means of Multisim simulation to acquire an output response data set; characteristic extraction: carrying out continuous wavelet analysis by taking the output response data set of the circuit as a training and test data set to acquire a wavelet time-frequency coefficient matrix, the coefficient matrix being segmented into eight sub-matrices of a same size, and performing singular value decomposition on the sub-matrices to calculate a Tsallis entropy of each sub matrix, so as to form a corresponding fault characteristic vector; and fault classification: submitting the characteristic vectors of all the samples to the ELM network so as to realize accurate and rapid fault classification. The method has good characteristic extraction effect on circuit faults, and can implement accurate and efficient classification o