基于Pajek的沥青摊铺机故障网络分析与诊断  

Asphalt-paver fault network analysis and diagnosis based on the Pajek

作　　者：唐红雨[1] 田磐[1] 王志英[2] 

机构地区：[1]镇江高等专科学校电子与信息工程系,江苏镇江212003 [2]江苏科技大学经济管理学院,江苏镇江212003

出　　处：《陕西理工学院学报（自然科学版）》2014年第2期24-29,共6页Journal of Shananxi University of Technology:Natural Science Edition

基　　金：镇江市工业科技支撑计划项目(GY2012005);镇江市科技支撑计划软科学项目(RK2013030);2013年江苏省高等职业院校国内高级访问学者计划项目

摘　　要：摊铺机是复杂的机电液系统,因受环境因素影响,容易出现故障,影响施工质量,需建立摊铺机的故障诊断系统。在建立摊铺机的故障征兆网络后,对采样信号进行3层小波包分解,用小波系数模极大值法提取故障的特征状态信息,用Pajek软件分析9个子系统故障的度和聚类系数,得到故障的可能部位,降低系统维数,减少后续计算量。用PSO优化神经网络算法进行状态样本数据学习,训练网络结构并优化,进而得到故障部位信息。样本测试表明,和神经网络相比,系统的迭代次数有效减少,实际输出结果和期望区间相符。The paver is a complex machinery-electro-hydraulic system .Due to environmental factors , the paver is prone to failure , affecting the quality of construction and resulting in property damage .Therefore, a diagnosis system for paver needs to be established .Firstly, the sampling signal is decomposed by three-layer wavelet packet , and fault feature status information is extracted by the coefficient modulus maxima method af -ter the establishment of fault symptoms network for the paver .Then failure degree and clustering coefficient of nine sub-system is analyzed by Pajek software to obtain the possible site of the fault , and reduce the system di-mension , and the amount of subsequent calculations .Finally, the state sample data is trained by PSO optimiza-tion neural network algorithm , then the structure of the network and fault information are achieved .The test da-ta show that comparing with the neural networks , the system is effective in reducing the number of iterations , and the actual output corresponds with the desired line interval .

关 键 词：故障诊断 小波包 神经网络 摊铺机 

分 类 号：TP277[自动化与计算机技术—检测技术与自动化装置]