基于反相传播神经网络的黑启动三相合闸统计过电压快速预测

在黑启动方案制定的过程中,需要对方案进行反复的空充输电线路操作过电压统计计算和校验,带来沉重的工作负担。提出利用人工神经网络快速预测黑启动空载合闸统计过电压的方法。通过选择一组有效的输入特征并采用过电压时域仿真程序建立训练样本集,误差反向传播神经网络被用来构造输入特征和最大统计过电压之间的映射关系,从而实现黑启动合闸统计过电压的快速预测。算例表明方法的有效性。

Fast forcasting of statistical no-load switching overvoltages in black start process based on back propagation artificial neural networks

In the black-start scheme design,a majority of the work is to calculate and assess the switching overvoltages for transmission line energization.A fast forecasting method to determine the no-load switching statistical overvoltages of black-start is proposed by use of artificial neural networks.By selecting a group of effective input features and building a training sample set by a time-domain simulation program of no-load switching overvoltage,the back-propagation neural network is trained to constructing the mapping between the input features and the peak values of the statistical overvoltages.The trained neural network is used to determine the switching statistical overvoltages quickly without time-consuming simulation.The effectiveness of the proposed method is validated by a numerical example.