State Assessment And Life Prediction Of Power Transformers

The capacity of transformers in the local power system is gradually increasing,along with a rising period of the scale of power system.Therefore,a reasonable use of the characteristic quantities scientific state assessment and life prediction of the transformer,can not only reduce the maintenance cost of the transformer,but also increase its economic benefits,which is significant to the safe and stable operation of the power system.This paper is devoted to the state assessment system of the transformer in terms of hierarchical classification,determination of index weights,streamlining of state quantities,and the accuracy of the life prediction model etc.The analytic hierarchy process,which can be further detailed into three levels:program level,criterion level and the target level,is put to use to analyze the state characteristics and fault types of transformer.Using electrical tests,chemical tests,and other methods to extract 24 characteristics as a transformer program layer,including dielectric loss,core ground current,oil breakdown voltage,winding DC resistance,gas content in oil,furfural content,cardboard Degree of polymerization and so on,nine fault types such as arc discharge,insulation aging,winding failure,core failure,partial discharge,insulation oil degradation,current loop failure,oil flow discharge,and insulation moisture are selected as criteria layers for transformer state evaluation,while the final state of the transformer is taken as Transformer state assessment target layer.This paper proposes a transformer state assessment algorithm based on analytic hierarchy process and association rules.In premise of a large amount of test data,together with the study of the relationship(referring to the association rules)between state characteristics and fault types and the confidence degree,the weight of the single state characteristic in fault type rating is determined.When in consideration of the weight of each fault type in the overall rating of a transformer,it is better to take account of the possibility of causing a significant accident,replacing the relevant parts of the transformer,and maintenance cost,professional practice,pair wise based on the AHP.Due to a large amount of experimental data in judging fault occurrence,the state assessment is of objective regularity,while professional practice in judging fault severity degree endows it subjective initiative.Furthermore,the number of 24 state quantities selected for transformer state assessment is Slightly large,unfavorable distribution of the relationship between failure type and state quantity,a state feature simplified algorithm,which is based on improved Apriori algorithm,Is presented in order to reduce the state assessment workload.On this foundation,arcing,insulation aging,insulation oil degradation,and winding faults is picked up as important elements in the life prediction model of the transformer,based on GM(1,1)model.The transformer failure rate is introduced as a parameter to realize the convergence of the prediction of the transformer’s physical life to the prediction of the economic life,which provides a certain basis for the maximum use of the transformer value.Finally,according to study of the transformer state assessment and life prediction system established in this paper,it turns out that these methods can accurately reflect the true state of the transformer operation,which is of great significance to transformer state maintenance and decommissioning management.