The Research Of Grounding Grid Fault Diagnosis Based On Information Entropy And Evidence Fusion

Grounding grid is the important equipment to meet the grounding requirements and ensure normal operation of power plants and substations.It provides a reference potential for the secondary system in the whole power system.The quality of the grounding grid is directly related to the safety of the power system and the operators.However,circuit branch of grounding grid,as buried under ground,can be subjected to corrosion or fracture because of the soil corrosion and the electromotive force of the short-circuit current.The resulting power system accidents had brought great personal and property losses.Therefore,it is essential that we establish an efficient grounding fault diagnosis system to diagnose various faults of the grounding grid rapidly and accurately.Using grounding grid as the main research object,this thesis has further studied the fault feature extraction method of grounding grid based on information entropy theory in depth,the information fusion method based on evidence theory and the fault diagnosis of grounding grid based on RBF neural network.On the basis of previous research,the fault diagnosis model of grounding grid under high-frequency excitation is established.In order to better reflect the grounding grid fault symptom,a fault feature extraction method based on information entropy theory is proposed by introducing information entropy theory in the fault diagnosis of grounding grid.Obtaining the voltage signal of touchable node under high frequency excitation,its singular spectrum entropy,power spectrum entropy and wavelet packet energy spectrum entropy are extracted from the time domain,the frequency domain and the time-frequency domain respectively according to the information entropy theory.The simulation results show that the method can obtain the state characteristic information from multi-direction and multi-levels.The theory of the information fusion and D-S evidence theory are applied to establish a multi-feature fusion method,therefore it can effectively reduce the dimension of feature space.The extracted three characteristics from the time domain,the frequency domain and the time-frequency domain are used as the evidence body.By means of the evidence theory,the feature information of different fault is merged to obtain the joint feature vector.The simulation results indicate that the obvious fault characteristics of information fusion are satisfactory.The fusion results are used as input into the RBF neural network to identify the different faults.It can effectively improve the performance of neural network by introducing the error correction algorithm to optimize the RBF neural network.The improvement on the structure and performance of the neural network is simulated as non-random error correction algorithm can adjust the network parameters without learning process.The simulation results show that the reliability and accuracy of the fault diagnosis of the grounding grid based on multi-feature fusion are obviously improved compared with the single feature fault diagnosis method.