针对基于传统深度学习进行输电线路故障测距时直接采用电压、电流学习而导致模型鲁棒性差的问题,提出了一种采用故障量相对偏移作为故障特征进行学习的方法,该方法显著提高了模型的鲁棒性;针对在通过支持向量回归(support vector regression,SVR)进行深度学习时,传统灰狼优化(grey wolf optimization,GWO)算法对初始参数的寻优容易陷入局部最优问题,提出了一种改进的自适应灰狼优化算法以确定SVR的初始参数.对IEEE33节点模型的仿真结果表明,相较于传统方法,该方法的准确性更高,且当线路电压与长度发生改变时,精度相差甚微,具有较好的鲁棒性.
A two-terminal fault location method for transmission lines based on relative deviation of fault quantity and improved GWO-SVR
Aiming at the problem of poor model robustness caused by the direct use of voltage and current learning in fault location of transmission lines based on traditional deep learning,this paper proposes a learning method that uses the relative deviation of the fault quantity as the fault feature to learn,which significantly improves the robustness of the model.In view of the fact that the traditional grey wolf optimization(GWO)algorithm is easy to fall into the local optimal problem in the optimization of the initial parameters when using support vector regression(SVR)for deep learning,an improved adaptive gray wolf algorithm is proposed to determine the initial parameters of the SVR.The simulation results of the IEEE33 node model show that compared with the traditional method,the accuracy of the method proposed in this paper is greatly improved,and when the line voltage and length change,the accuracy difference is very small,so the model proposed in this paper has better robustness.
relative deviation of fault quantityimproved gray wolf optimization algorithmsupport vector regressionfault location
NETL
NSTL
万方数据
维普
