用人工神经网络识别三相放电的数值仿真研究

为诊断三相大型电力设备的放电故障，采用数值仿真方法，研究了人工神经网络识别电模式的能力，提出了一种任务分解的识别三相设备电模式的神经网络组。用Ｍｏｎｔｅ－Ｃａｒｌｏ方法，产生不同相别，类型，发展程度的放电模拟数据，并对网络组进行训练。网络组对三相设备中的一相，两相或三相放电的识别结果表明，用这种网络组进行三相电力设备放电识别、类型，程度的分层次识别是可行的。

STUDY ON RECOGNITION OF 3 PHASES DISCHARGE BASED ON ANN USING MONTE-CARLO METHOD

In order to diagnose discharge faults of 3 phases electrical power equipment the recognition ability of artificial neural networks for discharge pattern was studied through numerical simulation method.A kind of task decomposition based artificial neural network group is presented,which is suitable for discharge pattern recognition of 3 phases electrical equipment. The discharge data of different phases,types and serious levels were simulated through Monte Carlo Method and were used to train the network groups.Under circumstances of 1,2, or 3 phases discharge of 3 phases equipment the recognition results of the artificial neural network groups proved the feasibility of hierarchical recongnition to discriminate the phase discharge occurred and the different discharge types and serious levels of 3 phases electrical equipment.