摘要
为提升铁路信号设备的故障处理效率,在对故障文本信息分析的基础上,提出一种基于关联规则的铁路信号设备故障诊断方法。由于故障文本信息存在不规范性和高维性,首先采用TF-IDF(Term Frequency and Inverted Document Frequency)算法提取故障文本信息中的故障特征,根据故障特征、故障类型、故障原因建立故障诊断模型。采用改进的FP-Growth(Frequent Pattern Growth)算法,通过权重定义项的支持度,依据频繁1-项集划分各项的数据库子库,并构造每项的条件FP-Tree,减少内存占用空间,提高运行速率,挖掘出具有维修指导意义的关联规则,进行故障诊断与维修决策。研究结果表明,本方法运行时间优于传统的FP-Growth算法,平均诊断准确率比案例推理算法和贝叶斯网络算法提高了10.35%和11.44%,可用于故障文本信息的潜在规律挖掘,简化故障诊断流程。
In order to improve the efficiency of fault handling of railway signal equipment,a fault diagnosis method of railway signal equipment based on association rules is proposed after the analysis of fault text information.Due to the irregularity and high dimension of the fault text information,the TF-IDF algorithm is used to extract the fault features in the fault text information,and the fault diagnosis model is established according to fault features,fault types and causes.The improved FP-Growth algorithm is used to define the support of items by weight,divide the sub-databases of each item according to the frequent 1-item set,and construct the conditional FP-Tree of each item.This method can reduce the memory occupation,improve the running speed,establish association rules with the significance of maintenance guidance,and carry out fault diagnosis and maintenance decision.The research results show that the running time of the method in this paper is better than that of the traditional FP-Growth algorithm,and the average diagnosis accuracy rate is 10.35%and 11.44%respectively higher than the Case-Based Reasoning algorithm and Bayesian networks algorithm.Therefore,the method can be used to find the potential laws of the fault text information and simplify the fault diagnosis process.
作者
张振海
张湘婷
ZHANG Zhenhai;ZHANG Xiangting(School of Automation and Electrical Engineering,Lanzhou Jiaotong University,Lanzhou 730070,China)
出处
《铁道标准设计》
北大核心
2022年第4期175-181,共7页
Railway Standard Design
基金
国家自然科学基金项目(61763025)
中国博士后科学基金项目(167306)。