专用线接触网的人工智能控制研究

Research of artificial intelligence controlling technology for dedicated railway catenary

铁路专用线接触网接地和负荷电流超载,都将影响干线铁路安全供电.在干线接触网与专用线接触网的"T"接处设置开关站,不仅可以快速切除专用线故障,而且可以限制负荷电流超载.在供电和用电双方都能够接受的前提下,提出了先警告,后惩罚性断电的控制策略.应用神经网络BP算法给定样本信号,通过神经网络的学习和训练,每次对神经网络的权值修正量为0.1%,控制检测信号输出与样本之间差值的平方和小于5%,就能够比较准确和快速地判断专用线接触网负荷电流是否超载.按照BP算法编制的计算机诊断程序,在实际开关站中进行应用,效果良好,实现了专用线接触网供电的智能控制管理.

Dedicated railway catenary grounding and its overload current can affect main railway line ＇s safety power supply. Setting the switch station in ＂T ＂joint,which is between main contact line and dedicated contact line,can not only cut-off the dedicated line fault rapidly but also limit the load current overload. This research brings forward an effective control strategy with acceptable premise between power supply and power user. The control strategy will firstly be warming both sides,then secondly cut-off the power supply as the punishment cause the overload line current. The paper uses neural network BP algorithm to give out sample signal,the weight modification of each neural network is 0. 1%,control the sum of squares of difference between detection signal output and sample signal less than 5%. According to the above neural network learning and training,the BP algorithm can rapidly and accurately determine whether dedicated railway catenary＇s load current overload. The computer diagnosis program,which was programing as BP algorithm theory,is good working in the actual switch station application. It realizes the artificial intelligence control management in dedicated railway catenary power supply.