摘要
文章针对FTGS轨道电路故障的原因和特点,提出一种基于多分支BP神经网络的FTGS轨道电路故障诊断方法。该方法通过对BP神经网络进行改进,建立多分支的BP神经网络,用于FTGS轨道电路的故障分类仿真。结果表明,多分支BP神经网络在精度和效率上都比专家系统、基于Agent方法和遗传算法更有优势,可以有效提高FTGS轨道电路故障诊断的准确性和可靠性。
Aiming at the causes and characteristics of FTGS track circuit failures,this article proposed an FTGS track circuit fault diagnosis method based on multi-branch BP neural network. This method established the multi-branch BP neural network through improving the BP neural network,which was used for fault classification simulation of FTGS track circuit. The results showed that the multi-branch BP neural network has more advantages in the precision and efficiency than expert systems,Agent-based method and genetic algorithms,and can effectively improve the accuracy and reliability of FTGS track circuit fault diagnosis.
出处
《西部交通科技》
2016年第3期76-79,共4页
Western China Communications Science & Technology
基金
广西高校科研项目"基于风险理论的城市轨道交通运营应急处置专家系统开发研究"(项目编号:2014YB565)
关键词
故障诊断
多分支BP神经网络
特征指标
决策融合
Fault diagnosis
Multi-branch BP neural network
Feature indicators
Decision fusion
