Power electronic converter fault diagnosis method based on cerebellar model neural network

• Main Authors: LIN QIONGBIN, WANG WU, CHEN SHICAN, WAN ZHISONG, CAI FENGHUANG, CHAI QINQIN
• Format: Patent
• Language: Chinese; English
• Published: 01.01.2019
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; HANDLING RECORD CARRIERS; MEASURING; MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES; PHYSICS; PRESENTATION OF DATA; RECOGNITION OF DATA; RECORD CARRIERS; TESTING

The invention relates to a power electronic converter fault diagnosis method based on a cerebellar model neural network (CMNN). Data acquisition and noise reduction processing are carried out to obtain a sample with fault information; a few of fault features are extracted by using methods like time-domain and frequency-domain analysis and then a data sample library is established; with a CMNN fault classifier, offline training is carried out based on a Back-Propagation algorithm and then various faults included by the training sample and specific fault locations are classified accurately; an optimal parameter of the fault classifier is extracted and the extracted optimal parameter is directly assigned to a classifier network to carry out classifier testing work; and then the classifier network with the optimal parameter is implanted into a DSP and fault diagnosis and positioning of the actual circuit are carried out to realized quick self checking of the converter circuit. According tothe invention, the healt