摘要
针对轨道电路维修维护中存在维修不足和维修过剩的问题,建立一种基于遗传算法优化核极限学习机(ELM)的轨道电路故障预测模型.首先,通过在ELM的基础上以核函数映射取代隐含层节点映射,改进为核极限学习机(KELM);其次,以遗传(GA)算法优化核函数的参数及正则化系数,改进KELM的学习性能形成新的GA-KELM算法.通过实例验证了GA-KELM相比于其他算法在轨道电路故障预测中的高效性.最后,选择GA-KELM算法对两个实例进行了实际应用验证,表明了改进算法的可用性.
Aiming at the problems of insufficient maintenance and excessive maintenance in track circuit maintenance,a track circuit fault prediction model based on genetic algorithm optimized kernel extreme learning machine was established.By replacing the hidden layer node mapping with kernel function mapping on the basis of ELM,the kernel extreme learning machine was improved.The genetic algorithm was used to optimize the parameters and regularization coefficient of the kernel function and improve the learning performance of KELM to form a new GA-KELM algorithm.An example is given to verify the high efficiency of the GA-KELM in track circuit fault prediction compared with other algorithms.Finally,the GA-KELM algorithm is selected to verify the practical application of two examples,which shows the availability of the improved algorithm and provides a new way for the intelligent operation and maintenance of railway signals.
作者
李晓艳
LI Xiaoyan(College of Railway Power,Shaanxi Railway Institute,Weinan 714000,China)
出处
《大连交通大学学报》
CAS
2022年第2期115-119,共5页
Journal of Dalian Jiaotong University
基金
陕西省渭南市科技计划资助项目(2019-ZDYF-JCYJ-127)。
