基于特种车的故障预测与健康管理系统研究

发布时间：2018-05-13 09:45

本文选题：故障树 + 专家系统　；参考：《南京理工大学》2017年硕士论文

【摘要】：本文研究的特种车是航天领域的重要设备,其性能的可靠性对航天任务的完成至关重要。由于频繁使用和设备复杂化的缘故,特种车在任务的执行过程中会经常出现故障,影响任务的正常进行。随着特种车设备集成化、综合化和智能化水平的提高,设备研制的风险越来越大、周期越来越长、费用越来越高,同时,对设备运行状态的监测以及维修手段也提出了更高的要求。为了实现对特种车的综合健康管理,本文在研究故障树分析法、专家系统分析法以及BP神经网络的基础之上,提出了一种新的混合故障诊断方法,即基于故障树和规则的专家系统,同时,建立了特种车的BP神经网络预测模型,对特种车的可能故障进行预测。论文首先在分析特种车实际需求的基础之上,设计了特种车故障预测与健康管理系统的体系结构和软件框架,并通过分析特种车的原始测试数据,实现了测试数据的导入和相关文件的配置;然后,在分析特种车故障模式的基础之上建立了特种车的故障树模型,并通过对故障树节点属性的配置,实现了基于CStatic控件的图形绘制与显示功能;最后,在分析特种车历史运行数据的基础之上,建立了特种车的健康等级样本,通过建立的样本对特种车的BP神经网络预测模型进行训练和优化,将训练好的BP神经网络预测模型进行固化并嵌入到软件平台。论文的创新点主要有两个方面:一是特种车故障预测与健康管理软件平台的通用性,通过导入一系列的配置文件,如算法配置文件、规则配置文件以及故障树配置文件等,使得软件平台不依赖于具体的某一种特种车,只要具备数据文件和相关的配置文件,就可以对任意系列的特种车进行健康管理;二是基于CStatic控件的故障树图形显示方法,结合故障树的节点属性和CStatic控件,实现了故障树图形的可视化和多元化,通过该方法可实现故障树的全自动绘制,而且用户可以根据自己的需求浏览故障树。通过特种车的历史运行数据对故障预测与健康管理系统进行实际验证,软件的诊断和预测结果体现了系统的通用性和实用性,符合实际需求。
[Abstract]:The special vehicle studied in this paper is an important equipment in spaceflight field. The reliability of its performance is very important to the accomplishment of space mission. Because of the frequent use and complicated equipment, the special vehicle will often break down during the task execution, which will affect the normal operation of the task. With the integration, integration and intelligent level of special vehicle equipment, the risk of equipment development is increasing, the period is getting longer and longer, the cost is getting higher and higher, at the same time, Higher requirements are also put forward for the monitoring and maintenance of equipment operation status. In order to realize the comprehensive health management of special vehicles, a new hybrid fault diagnosis method is proposed based on the research of fault tree analysis, expert system analysis and BP neural network. That is an expert system based on fault tree and rules. At the same time, the BP neural network prediction model of special vehicle is established to predict the possible faults of special vehicle. On the basis of analyzing the actual demand of special vehicle, this paper designs the system structure and software framework of the special vehicle fault prediction and health management system, and analyzes the original test data of the special vehicle. Then, on the basis of analyzing the fault mode of special vehicle, the fault tree model of special vehicle is established, and the node attribute of the fault tree is configured. The graphics drawing and displaying function based on CStatic control is realized. Finally, on the basis of analyzing the historical running data of special vehicle, the health grade sample of special vehicle is established. The BP neural network prediction model of the special vehicle is trained and optimized by the established samples, and the trained BP neural network prediction model is solidified and embedded into the software platform. There are two main innovations in this paper: first, the generality of special vehicle fault prediction and health management software platform, such as the introduction of a series of configuration files, such as algorithm configuration file, rule configuration file and fault tree configuration file, etc. So that the software platform does not depend on a specific special vehicle, as long as there are data files and related configuration files, any series of special vehicles can be managed healthily. The second is the graphical display method of fault tree based on CStatic control. Combined with the node properties of the fault tree and the CStatic control, the graph of the fault tree can be visualized and diversified. Through this method, the automatic drawing of the fault tree can be realized, and the user can browse the fault tree according to his own requirements. The fault prediction and health management system is verified by the historical running data of the special vehicle. The diagnosis and prediction results of the software reflect the generality and practicability of the system and meet the actual needs.
【学位授予单位】：南京理工大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：V55;TP311.52

【参考文献】