摘要
针对轨道交通直流牵引供电系统列车启动负荷冲击电流引发的DDL保护误启动问题,提出了基于小波包变换和自组织映射(SOFM)神经网络的短路电流检测方法。该方法采用改进小波包变换将馈线电流信号映射至频域,消除频谱混叠干扰后,提取能够区分负荷冲击电流与远端短路电流的特征向量,将特征向量输入到SOFM神经网络实现短路电流检测。实验结果表明,该方法能够有效区分负荷冲击电流和远端短路电流,且识别准确率优于现有方法。
Aiming at problem of false start of DDL protection caused by train starting load impulse current in rail transit DC traction power supply system,a short circuit current detection method based on wavelet packet transform and self organizing map(SOFM)neural network is proposed.The improved wavelet packet transform is used to map the feeder current signal to the frequency domain.After eliminating the spectrum aliasing interference,the feature vector which can distinguish the load impulse current from the remote short circuit current is extracted.Then the feature vector is input into SOFM neural network to detect the short circuit current.The experimental results show that the method can effectively distinguish the load impulse current and the remote short circuit current,and the recognition accuracy is better than the existing methods.
作者
史天玉
SHI Tianyu(China Railway First Survey and Design Institute Group Co.,Ltd.,Xi’an 710043,China)
出处
《机械与电子》
2021年第3期29-33,38,共6页
Machinery & Electronics
关键词
直流牵引供电
短路检测
小波包
SOFM神经网络
DC traction power supply
short circuit detection
wavelet packet
SOFM neural network