On-Line Fault Diagnosis Of Twenty Four-Wave Controlled Rectifier

On-line fault diagnosis for high-power power electronics equipments is a significative subject. It is an intersectional subject which comes into existance in order to adapt to the engineering demand. Multi-wave controlled rectifier is applyed to electric drive industry extensively, such as city subway, realway transportation etc, and which is also the key equipment during lots of industrial process. At present, the overseas and domestic scholars have done a great deal of research on six-wave and twelve-wave in this field and notable results have been obtained. However, multi-wave controlled rectifier with less coefficient of ripple is reported scarcely. Therefore, fault diagnosis of twentyfour-wave controlled rectifier is analyzed theoretically and validated by simulation in this thesis.The developing situation of fault diagnosis is introduced firstly, then the research purpose and meaning of this thesis are summarized, and the recent research condition and characteristics of the fault diagnosis for high-power power electronics equipments are analyzed, later, the scheme of this thesis is established, primarily includes the theories analysis and simulation two aspects. In theory analysis, the voltage waveforms of two different structures of 24-wave controlled rectifier under fault and normal condition are analyzed and researched in detail, all the faults states are summarized and classified according to the aberrant voltage waveforms, and two distinct methods are employed. So far as the rectifier without balanced reactor, the corresponding 24-dimension fault pattern vectors are founded using two numerical values, and logical fault tree is constructed, afterward the fault diagnosis oriental algorithm is summarized, then it realizes the direct shift from fault pattern vectors to oriental vectors. To the complicated twentyfour-wave controlled rectifier with many balanced reactance, it samples the output voltage of controlled rectifier, and then it pretreats the data, after that, it makes classifications roughly according to the number of fault eigen vector, then it confirms the one to use among these parallel BP sub-networks and identifies the fault category. Finally, based on the location of the occurrence of exceptional information in the three numerical values code for the first time, it diagnoses the specified fault state of fault category. The proposed diagnosis method solves the problem of fault diagnose of the complex rectifier successfully for the first time, and simulations are conducted by Matlab software. The result shows that the method can diagnose all kinds of faults of the complicated twenty-four wave controlled rectifier with many balanced reactance fast, accurately and simultaneously. The simulative results of two different topology twenty-four wave controlled rectifier show that the two methods are feasible, favorable and fast. The algorithm lays a solid condition for the industrial use. In addition, the fault diagnosis for controlled rectifier based on wavelet analysis is explored. The specified fault state of fault category is diagnosed by the maximum value of wavelet transform of abnormal point. We know that the small sort could be differentiated according to the simulation result.