Study On Maintenance Decision In Transformer

With the rapid growth in power demand, many problems arise gradually as thescheduled maintenance system has been adopted for Transformer for a long time,including maintenance excess, deficiencies and cost increment. On the other hand,with the rapid development of computer technology and sensing technology in recentyears, the advanced condition-based maintenance technology has been promotedgreatly, which was paid attention to by many researchers. This dissertation focuses onthe condition-based maintenance problems of the Transformer, especially on theapplication of transformer, including maintenance strategies selection of Transformer,condition assessment and fault diagnosis of transformer, and finally the design and thedevelopment of maintenance management system for the power transformationequipment. Details are described as follows:A systematic study of the optional maintenance strategies of the Transformer isconducted and a set of general factors which should be considered in the maintenancestrategies is established. A Fuzzy Analytic Hierarchy Process (FAHP) based on thegoal programming is proposed in view of the defects of the traditional Fuzzy AnalyticHierarchy Process (FAHP), and the specific application process of this method in thetransformer maintenance strategies selection is described. In addition, a comparisonwith the standard Analytic Hierarchy Process is made to clarify the validity of themethod proposed in this dissertation.The criteria system and criteria normalization method for the conditionassessment of Transformer is developed and membership function of the criteria isestablished. The establishment of Expert Weight is decided jointly by subjectiveweight and objective weight which is based on entropy weight thoughts, while thecriteria weight is gained by the weight which is derived from the standard AnalyticHierarchy Process and the Expert Weight. A Comprehensive Condition AssessmentModel of Transformer based on OWA(Ordered Weighted Averaging) operator andfuzzy assessment is proposed and divided into two layers, fuzzy polymerization isadopted for the second layer sub-criteria to get the fuzzy membership of the first layermain criteria, while OWA operator polymerized with the final comprehensivecondition assessment of the transformer is adopted for the main criteriaof the first layer. In order to fully consider the impact of criteria weight and membership on thecondition assessment result, a fuzzy polymerization conversion function based onOWA operator is introduced in the model, so as to integrate the attribute informationfrom various important information sources. Case analysis indicates that the conditionassessment of transformer can be carried out by this method; the reasonable andobjective conclusion is close to the true condition of transformer.The Chaos Immune Evolutionary Fuzzy Clustering algorithm which is based onGrey Relational Degree is proposed and applied to the fault diagnosis of transformer.Compared with Fuzzy C-Means algorithm and Fuzzy C-Means algorithm which isbased on immune, Fuzzy Clustering algorithm has faster convergence speed andhigher accuracy of fault classification. Chaotic disturbance for initial antibody groupis made by the model, two factors including Grey Relational Degree and FuzzyClustering Objective Function are taken into full consideration as an assessmentstandard of affinity degree of antibody and antigen, adaptive cauchy mutationoperator and chaotic clone operator are adopted, so as to reduce the impact of initialvalue on Fuzzy Clustering algorithm, improve the local search capability at laterstages and accelerate the convergence speed. What’s more, it can also effectively stopthe "degeneration" phenomenon generated in the immune process. The experimentalanalysis indicates that the Fuzzy Clustering algorithm can improve the accuracy offault classification.Power transformation equipment maintenance management system is designedand developed. Analysis of requirements, general design and detailed design based onMVC and key technology for implementation is described in detail.In this Dissertation, there are thirty-seven figures, fifty-four tables, and onehundred and seventy-six reference documents.