Research On Condition Assessment Method For Power Transformer

Transformer is the most important equipment in a power system. The reliability of theoperating transformers has great influence on the safety and stability of the power system.With the increase of the capability and scale of the power system in china, the consequenceof the fault of running transformer will be more and more serious. It is significant for theimprovement of the reliability of the power system to realize condition based maintenanceof the transformers.Firstly, some typical faults and their mechanism are analyzed in this thesis. Combinedwith the analysis of transformer faults in the Guide for Condition Evaluation ofOil-immersed Power Transformers (Reactors)(Q/GDW169－2008), nine synthetic statusparameters are chosen, including moistened insulation, core fault, current circuit fault,winding fault, partial discharge, oil-flow discharge, arc discharge, insulation aging and oildeterioration. Independent status parameters related to each synthetic status parameter aredetermined.Secondly, association rules and variable weight coefficients theory are used todetermine the weight coefficient. By comparing calculation of the support degree andconfidence degree of the association rules, the constant weight coefficients of independentstatus parameters are got. The problem that the accuracy of weight coefficient is influencedby expert opinion or subjective experience can be solved. The synthetic status parametersare variable according to their scores. It can avoid the problem that the true state of thetransformer may not be accurately reflected under the condition of constant weightcoefficients.At last, a transformer condition assessment system is established based on the abovetheories. Furthermore, fuzzy theory, which include membership, relative impairment gradeand membership function of semi-ladder and triangular, is used to dispose the assessmentresult. This method makes the transformer s final score coincide well with its fault state.Two examples are analyzed, one of which is multipoint earth faults in magnetic coreand another is insulation aging. The results show that the assessment method proposed can reflect the true state of the transformer accurately. The validity and advantage of thismethod are also verified in practical application by comparing with the methods based onstatus-parameters scoring and support vector machine.