可评价保护系统动作行为的双重神经网络电网故障诊断模型

Dual Neural Network-based Fault Diagnosis Model of Power Grids for Evaluating the Action Behavior of Protection Systems

神经网络应用于电网故障诊断时不能很好地实现对保护系统动作行为的评判,因此提出一种可评价保护系统动作行为的双重神经网络电网故障诊断模型。在准确诊断故障元件的基础上,通过增加另外一重神经网络,利用0、1、-1对保护系统正确动作、误动、拒动情形进行编码,结合原有故障报警信息和诊断出的故障元件,进一步判断保护系统是否发生拒动、误动。算例仿真结果表明:改进后的模型,不仅可正确诊断故障元件,还可有效评价保护系统的动作行为,更好地为故障处置提供决策依据。

Neural network is not well suited for the evaluation of the action behavior of protection systems when it is applied to grid fault diagnosis.Dual neural network grid fault diagnosis model with the function of evaluating the action behavior of the protection system is proposed.By adding the BP neural network,and using 0,1,-1 to encode the correct action,misoperation and rejection of the protection system,it can determine whether the protection system action occurs rejection or misoperation according to the original fault alarm information and the diagnosed fault area when the original BP neural network grid fault diagnosis model accurately diagnoses the fault area.The simulation results show that the improved model can not only identify the fault area,but also effectively evaluate the action behavior of the protection system,so as to better provide decision-making basis for fault handling.