轨道工程机械远程在线监控与智能故障诊断系统关键技术研究

In order to improve the level of remote on-line state monitoring and fault diagnosis about our nation’s large rail maintenance machinery and to reduce the enterprises’ cost of production, this project proposed several related critical problems in view of the realities of situation of remote on-line state monitoring and fault diagnosis and discussed them from the following aspects. .This project will uses Wireless Sensor Network(WSN) to build the model of on-line state monitoring system of the Relevance Vector Machine (RVM). It will research on WSN of high-speed data acquisition and real-time and reliable data transmission and power management. The monitoring data gathered from the wireless sensor has larger redundancy and error, which affects the data’s reliability. However, the requirements or the data accuracy are very low because of the REM with many characteristics. To realize REM dynamic real-time data fusion and enhance the real-time data precision, we need to take working process data as the testing site to propose a kind of data fusion method with real-time data and utilized the Grubbs criterion for data preprocessing and application of the adaptive weighted algorithm for the data fusion. The Grubbs criterion could effectively eliminate the large error. . In order to comprehensively and reasonably utilize much feature information of the REM to improve the accuracy of fault diagnosis. A method of intelligent fault diagnosis was proposed based on RVM and Improved Evidence Theory (IET). The reliability of local diagnosis evidence of each RVM for every failure mode was acquired with a confusion matrix to give different weight coefficient. The basic probability assignments constructed with a hard output decision matrix from the local diagnosis of each RVM were processed weighted to realize the effective combination of RVM and IET in intelligent fault diagnosis.. It will also improve the RVM and IET's fault symptoms extraction precision and calculation speed and optimize the RVM’s nuclear parameters to enhance the accuracy in fault diagnosis. An algorithm which can quickly extract the REM's fault symptoms and accurately diagnose. it's fault will be designed. The algorithm combines the IET and the RVM algorithm to achieve the REM's fault diagnosis. The project will not only ultimately solve the REM remote on-line state monitoring and fault diagnosis's related critical problems at last, but also establish a hybrid theoretical model based on several related critical technology such as WSN, The auto-adapted weighted average data fusion algorithm and RVM with improved evidence algorithm. Then the project will build an on-line state monitoring and fault diagnosis experimental system and make sure the system be effective. The on-line state monitoring and fault diagnosis system will be integrated into the whole REM. This project provides a new train of thought and theoretical basis for the further research and more application.

远程在线状态监控与智能故障诊断技术已成为目前国内外研究的热点，它对于设备的制造、使用、维护等有着重要的意义。项目探索在复杂环境下采用WSN技术在线采集设备状态数据，研究实时数据及自适应加权平均算法的效率，提出适用于轨道工程机械特点的多数据融合模型。重点研究设备故障状态对各状态特征信息的依赖程度，提出基于RVM并结合加权思想的故障诊断理论模型，以解决故障识别率低的关键难题。通过对关键技术的不断优化，项目提出的理论模型将会显著提高故障识别的可靠性和准确性。为基于私有云服务的轨道工程机械智能故障诊断提供理论基础。课题从WSN在线数据采集着手，研究了多数据融合的难点及关键技术，从而提出新的故障提取模型的框架理论，试制远程在线监控与智能故障诊断系统的实验模型。最终和原有的轨道工程机械车载系统实现统一集成，为企业云服务系统的建设和轨道工程机械故障诊断技术的进一步研究与应用提供新的思路和理论依据。

课题从提高我国轨道机械设备远程在线自动化控制水平、实现实时智能故障诊断及提高轨道维护效率的需求出发，针对复杂环境下轨道设备实时在线状态数据采集难及故障数据复杂的问题，设计了包括数据感知层、网络通信层和平台应用层的远程实时智能故障诊断故障诊断系统，利用无线传感技术实现数据的无线远程通信，极大提高数据传输效率，将获取的各种轨道设备数据进行预处理和相关性分析，根据轨道机械数据的特点进行多元数据融合，构建基于私有云服务的数据安全平台，对处理后的数据进行数据存储和智能故障诊断分析，为了解决轨道机械故障识别率低的问题，重点研究轨道设备故障状态对各状态特征信息间的关系，提取故障特征，应用故障诊断算法建立基于轨道数据安全传输的实时在线远程监控系统与智能控制的轨道机械故障诊断系统，实现远程故障诊断、故障告警、故障分析和故障修复等功能，实验结果表明，与传统的故障诊断方法相比，远程智能故障诊断系统的诊断效率和精确度都更高。