A Study Of Electronic Transformer Fault Diagnosis Based On Wavelet-fractal Theory

With the development of the Smart Grid, it would become inevitable that electronictransformer insteads of traditional electromagnetic transformer having the advantage ofsimpler insulation structure, smaller volume, lighter weight,no magnetic saturation andferro resonance. But electronic transformer has worked for a short time,the reliability ofelectronic transformer is unstable.The electronic transformer often occur wrong orabnormal signal result in measuring distortion and even the misoperation of protectivedevice,so the electronic transformer fault diagnosis technology is the importantmeasures to ensure the safe operation of power grid in addition to improve thereliability of electronic transformer.This paper proposes the diagnosis method of the electronic transformer fault basedon wavelet-fractal theory according to the analysis of the current research status ofelectronic transformer and fault diagnosis method. Through the analysis of structuretheory of electronic transformer, combined with the actual operation of electronictransformer,this paper research the operation weak point and influence factors ofelectronic transformer. Using RTDS (Real Time Digital Simulator) power systemdynamic simulation system and OET700type combined-type photoelectric transformerto build the electronic transformer experimental platform. Simulating the power gridfailure and electronic transformer fault to research the signal features with the powergrid failure and electronic transformer own fault,and establishing the mathematicalmodel.To research the electronic transformer fault diagnosis method based on themathematical model. this paper will be divided the diagnosis method into fault patterndetection, fault components location two stages to get the cause of electronictransformer fault gradually.Fault pattern detection method uses wavelet-fractal theory. the method combinesadvantages of the ability of the wavelet theory in detection signal singular value andability of fractal theory in the signal feature extraction. Analysing the fractal dimensionof electronic transformer’s fault model and establishing five kinds of fault model.According to the electronic transformer fractal dimension analysis of fault output signal,we can judge the fault model of the electronic transformer fault belonged to the fivefault models built.Using MATLAB software to simulate the fault mixed noisesignal.The result of fault simulation proves that the fault pattern detection method can effectively eliminate the affect of noise and collect the fault signal characteristic valuefor fault pattern diagnosis.Fault components location uses the fault tree theory to establish fault tree for thefive kinds of fault model. Analysing the features of the fault parts of electronictransformer, building the fault tree for each fault model, sorting fault tree in theimportance of minimum cut set.when fault location,Searching the cause of fault inimportant degree of the minimum cut set by comparing the fault signal characteristicswith features of the minimum cut set. And using the information fusion method to tacklethe collected fault information and builds fuzzy matrix and decision-making feasibledegree matrix with the fuzzy theory. This method distinguish the difficult fault bymerging all kinds of fault information in the data layer, characteristic layer,decision-making layer. Using the diagnosis method to diagnose the fault in the RTDSsimulation of the second chapter.The result verify this method can maximize miningvarious hiding information from fault information to locate fault,also verify the practicalvalue of the fault diagnosis method.