为了在电力系统不同故障位置、故障时刻和噪声环境中准确识别暂态故障类型,提出基于机器学习的电力系统暂态故障事件智能识别方法.将暂态故障结构特征值作为量子粒子群优化径向基神经网络模型的输入向量,通过选取合适的参数编码策略、适应度函数以及终止条件,输出优化后径向基神经网络最优参数,完成故障事件智能识别.仿真实验结果表明,该方法采用量子粒子群优化算法(QPSO)优化径向基函数(RBF)神经网络可以获取最佳训练参数,训练时间为3.561 s,训练误差为0.000 257 7,可在不同故障位置、故障时刻和噪声环境下正确识别暂态故障类型,且识别效率优势显著.
In order to accurately identify transient fault types in different fault locations,fault moments and noisy environments,an intelligent identification method of transient fault events in power system based on machine learning is proposed.The eigen-values of transient fault structure are used as the input vectors of the quantum particle swarm optimization and radial basis func-tion neural network model,the optimal parameters of the optimized radial basis function neural network are used as output by selecting the appropriate parameter coding strategy,fitness function and termination conditions,the system then completes the intelligent identification of fault events.The simulation results show that the optimal training parameters can be obtained by u-sing quantum particle swarm optimization(QPSO)to optimize radial basis function(RBF)neural network.The training time is 3.561 s and the training error is 0.000 257 7.This method can correctly identify the transient fault type in different fault loca-tion,fault time and noise environment,and the identification efficiency is significant.