Fault Diagnosis Of Electronic Instrument Transformer Based On Wavelet Neural Network

Comparing to the traditional electromagnetic transformer, electronic instrument transformer, which represents the developing trend of high-voltage power grid and voltage measurement tools, gains the advantage in terms of bandwidth, insulation and cost. With the gradual extend of smart grid and digital substation technology, electronic instrument transformer becomes a popular research, many new technologies have been introduced in the designation of electronic instrument transformer in order to induce the system operation cost, and to decrease the pressure on environment. However, as a new thing of power system, whether can work under the poor condition continuously is a core of practical problem on the road. It is valuable to do the research on the fault diagnosis of electronic instrument transformer. The first point is how to realize the problem while the electronic instrument transformer is working inaccurately; the secondly point is how to distinguish the abnormal change of power signal which will be directly reflected in the output transformers and the fault signal which occurred by transformer itself.The thesis is working on the basis of analyzing the current research of electronic instrument transformer and fault diagnosis technology, and then put forward the fault diagnosis method which based on the Wavelet Neural Network. The fault diagnosis method is a combination of knowledge processing and signal processing method, among them, artificial wavelet analysis is used to pick up the signal frequency domain, neural network is used to realize and classif. The feasibility was validated by the simulation experiment. The specific research contents are as followed:â‘ To introduce the related knowledge of neural network and wavelet theory and focusing on the combination mode of the wavelet and neural network and the advantage in the area of fault diagnosis.â‘¡To analyze the existent fault of electronic instrument transformer comprehensively, focus on the several typical fault modes of transformer, which are classified as mutant fault and graded fault, according to the characteristics of signal change. Illustrating the applicability that wavelet neural network method works on the fault diagnosis of electronic instrument transformer.â‘¢To construct the fault diagnosis system of electronic instrument transformer, finishing the designation of hardware and software part. Via three typical imitate experiments of diagnosis, the diagnosis method which based on the wavelet neural network shows remarkable effect on single transformer.â‘£The diagnosis result of single transformer might be affected by power factors. According to analysis of the different operating conditions of electric power capacity, the thesis proposes a joint diagnostic program that uses reluctant massage of several transformers. The simulation results show that the project can identify the abnormal power grid and transformer fault effectively.