Research On High Resistance Fault Diagnosis Of Power Distribution System

The distribution network is a window for power applications and users,and is an important breakthrough for grid intelligence.The power line is the lifeblood of the distribution network,and the normal operation of the power line is closely related to the safety and stability of the power system.Due to cable mixing,there are many branch lines,complex structures,complex and variable operating conditions,and many faults occur.High-impedance faults are a common type of fault in single-phase grounding faults of transmission lines.Because the fault characteristics are not obvious,the detection is difficult,and the high-resistance fault has always been a difficult point in the fault identification of the distribution network.Since the fault characteristics are not obvious,high-resistance fault has been an urgent issue in the fault identification of the distribution network.In actual power consumption,people use the trial and error method to detect the fault location of the transmission line,but this requires a lot of human resources and consumes a lot of time.Since reinforcement learning can achieve high-precision classification with a relatively small number of samples,this paper proposes a classification method for high-impedance faults in transmission lines based on reinforcement learning.First,empirical mode decomposition(EMD)is used to process the original voltage and current to generate an intrinsic mode function(IMF)containing local characteristic signals of the original signals at different time scales,and then these intrinsic mode functions are performed using the C4.5 algorithm.(IMF)Trim to be used as an input variable for the reinforcement learning fault classifier.This classifier is adaptive,and if the input sample is misclassified,it is punished to try to correct the error in the next decision stage for optimization purposes.During the iterative learning process,the classifier displays a vector value of the input sample including the intrinsic mode function(IMF)and attempts to identify the fault corresponding to the sample.If the classifier output matches the fault type,it will be rewarded,otherwise it will receive a penalty.Accompanied by this reward penalty identification procedure is the incremental method for adjusting the q value.As the algorithm progresses,the q value is gradually improved,and the classification accuracy of the transmission line fault is significantly improved.After training and testing a sufficient number of samples,the Intensive Learning Fault Classifier can accurately classify faults.The MATLAB simulation test tests the high-impedance fault of a lOkV bus and its three branches,preprocesses the phase voltage,phase current,zero-sequence voltage and zero-sequence current data,constructs and adjusts the reinforcement learning model for data training.Finally,locate the transmission line where high resistance fault occurs.The simulation results show that the high-resistance fault classification method for transmission lines based on reinforcement learning can accurately determine the line position where high-resistance faults occur.