Application Of Bias Network And D-S Evidence Theory In Fault Diagnosis Of Power System

With the development of modern science and technology,the fault diagnosis technology of power grid has been considerable development,but in practice still faces many challenges,especially the emergence of complex faults such as the cascade trip,the dispatch center will instantly receive massive fault information,to dispatch personnel caused great difficulties in diagnosis,the traditional diagnostic methods are based on diagnosis switch,diagnostic information is too difficult to meet the modern single fault diagnosis accuracy requirements,a comprehensive diagnosis using multi-source information method for fault diagnosis and has certain practical significance.The Bayesian network is suitable for solving non deterministic reasoning problems,and can effectively deal with the incomplete fault information,data fault tolerance problems,so this dissertation uses the Bayesian network for fault diagnosis;in addition,the fault diagnosis process using information fusion technology,make full use of the redundancy of the multi-source fault information.In the information fusion algorithm,the D-S evidence theory does not require a priori information,is a kind of uncertain reasoning algorithm is accurate,applicable to solve the uncertainty problem of multi-source information,so this article launches the research on D-S evidence theory and its application in power system,the main research contents include:(1)Study the validity of evidence conflict coefficient.According to the traditional theory of evidence conflict coefficient unreasonable problem,by introducing Mahalanobis distance tria ngular norm operator and evidence redefines the conflict coefficient,reasonable degree of evidence conflict to be distinguished,improve the effectiveness of evidence conflict coefficient.(2)To study the effective fusion of high conflict evidence.In vi ew of the traditional combination rule can not effectively combining the conflict evidence,this dissertation proposed two amendment evidence combination method,combination of body weight to evidence,and trust coefficient of each step fusion results to i mprove the fusion accuracy of discount.(3)The application of Bayesian network and improved D-S evidence theory in fault diagnosis of power grid.Aiming at the fault diagnosis based on Bayesian network diagnosis method mostly adopts the centralized diagno sis,caused by low efficiency and diagnosis of vulnerable network structure change,this dissertation adopts distributed diagnosis,the distributed diagnosis greatly improves the diagno sis efficiency;The traditional diagnostic methods rely on the switch c ause diagnostic accuracy the problem of high analysis of electrical quantities is introduced in this dissertation,by using the improved D-S decision fusion rule,diagnosis example shows that the proposed diagnosis method has better fault tolerance and acc uracy.