Fault Diagnosis Research Of Wind Turbine Based On Evidence Theory

Fault diagnosis of wind turbine is an important technology in wind power generation. With the rapid development of wind power industry, fault diagnosis of wind turbine has new and higher requirements put forward. The fault diagnosis of wind turbines is of great significance. As a kind of uncertainty reasoning method, the evidence theory is more and more convenient and efficient in dealing with the uncertainty information, and it has been widely used in the fault diagnosis based on multi-source information. The evidence theory has been an appropriate choice in fusion diagnosis because of the characteristics of wind turbine fault that multisource information representation and uncertainty. The method is applied to the fault diagnosis of wind turbine, it includes the following three points:(1)The article introduces the basic principles and concepts of the information fusion technology, and the fusion model of information fusion in the data level, feature level and decision level. The related methods of fault diagnosis based on evidence theory have been summarized through the research of evidence theory framework and the corresponding combination rules. The possibilities that multi-source information fusion method based on evidence theory have been analyzed through the analysis of the methods currently used in fault diagnosis of wind turbine.(2)Through the research and application of the theory of relative entropy, the influence of the original evidences which are conflicting on the fusion result is solved. According to the different importance of evidences obtained from multiple sensors, relative entropy theory has been used to evaluating the size of influence factor of each evidence. Namely, by giving different weights for each evidence, the influence of each evidence would be weakened or enlarged. The method applied in bearing fault diagnosis of wind turbine obtains good diagnosis effect. The method extracts vibration signal of faulted bearing by the acceleration sensors, and empirical mode decomposition(EMD) method has been used to decompose the fault vibration signal. Envelop spectrum frequency has been collected to be the fault feature. A preliminary diagnosis results was acquired by comparing the characteristic frequency and the fault sample frequency; Relativity theory in grey system theory is introduced, the original evidences are obtained by calculating the correlation in the inspection feature frequency and the fault sample frequency; Corresponding weights are distributed original evidences by the application of the relative entropy theory. Finally, evidences modified are fused by the D-S combination rule.(3)In the reality fault diagnosis, there are some shortcomings in the application of the evidence theory because of the influence of the external environment, that is, the extraction of BPA(basic probability assignment function) from incomplete and uncertain information also has some limitations. The belief measure based on random sets and the plausibility measure based on random sets are processed by vague sets. BPA of the evidence theory could be well extracted by the method; Better diagnosis results could also be obtained in the fault diagnosis of generator in the wind turbine. By the comparison of the vague sets and fuzzy sets in processing the uncertainty information, the former can do it better and more objectively and comprehensively reflect the nature of the uncertain information. This method not only can be applied to the fault diagnosis, but also can be used to diagnose the fault of other mechanical and electrical equipment.