基于朴素贝叶斯算法的电网故障诊断研究

Research on Power Grid Fault Diagnosis Based on Naive Bayesian Algorithm

针对现有的电网故障诊断系统误差大、适应性差等问题,将朴素贝叶斯算法和D-S证据理论结合,并利用经验模态分解(EMD)方法将电网故障信号分解为有限个固有模态函数。针对电网设备并联型有源电力滤波器(SAPF)中功率开关的开路故障会加剧电网的谐波污染,提出一种基于三相四开关(TPFS)拓扑进行硬件容错。仿真结果表明,近5年内年平均精度达到94%,具有良好的诊断效果。

Aiming at the problems of large error and poor adaptability of the existing power grid fault diagnosis system,the na-ive Bayesian algorithm and D-S evidence theory are combined,and the empirical mode decomposition(EMD)method is used to decompose the power grid fault signal into finite intrinsic modal functions.Aiming at the open circuit fault of the power switch in the shunt active power filter(SAPF)of power grid equipment,which may aggravate the harmonic pollution of the power grid,a three-phase four-switch(TPFS)topology is presented for hardware fault tolerance.The simulation results show that the average annual accuracy has reached 94%in the past five years,and it has a good diagnostic effect.