船舶动力系统故障诊断方法与趋势预测技术研究

发布时间：2018-01-21 05:51

本文关键词： 船舶动力系统故障诊断专家系统 SOM 神经网络趋势预测　出处：《武汉理工大学》2014年硕士论文　论文类型：学位论文

【摘要】：船舶动力系统作为整个船舶的心脏与动脉，包括主推进装置、辅助供能装置、用于保证船舶安全运行的设备、满足船员正常生活的设备和环境保护设备等。由于船舶动力系统的运行条件苛刻，并具有强烈的时变性，一旦发生故障，往往会造成严重的后果，其安全可靠的运行将直接影响到船舶运行的安全。在船舶动力系统趋于自动化和智能化的背景下，对船舶动力系统进行智能化故障诊断是保证该系统安全可靠运行的重要方法之一，具有重要意义。本文在分析国内外船舶动力系统故障诊断系统的发展现状的基础上，针对目前存在的在线诊断能力薄弱等问题，研究了船舶动力系统智能故障诊断方法。首先采用规则引擎技术，研究了船舶动力系统故障诊断专家系统的推理方式以及诊断规则库的构建，制定了以Drools为推理引擎的专家系统方案。通过对船舶动力系统的主要故障模式进行分析总结，以此为基础构建诊断知识库。为解决专家系统在实施过程中存在的知识获取瓶颈、不完整性信息处理能力较差等问题，进行了数据驱动的故障诊断方法的研究，并通过SOM神经网络构建故障诊断模型来弥补专家系统的不足。同时，为了实现事后诊断向预诊断方式的转变，在故障诊断的基础上，研究了船舶动力系统主要状态参数的趋势预测方法，，采取了ARMA模型和小波神经网络模型。在对比两种模型的特点和适用范围的基础上，针对不同的状态参数选取不同的模型进行了趋势预测，可以实现对异常参数变化的提前报警，对船舶动力系统的日常维护具有一定的指导意义。本文以“东海救117”为应用对象，在原有船舶动力系统状态监测系统的基础上，利用该系统采集的船舶动力系统状态参数进行故障诊断。根据对故障诊断方法以及趋势预测技术的研究，对故障诊断系统的功能进行了设计，对其实现方法进行了研究。以Windows操作平台为运行平台，Java为开发语言，利用Eclipse作为开发工具完成了故障诊断专家系统功能的实现，为轮机工作人员对船舶动力系统的维护提供了一种新手段。
[Abstract]:Ship power system as the whole ship's heart and artery, including the main propulsion device, auxiliary energy supply device, used to ensure the safe operation of the ship equipment. Because of the harsh operating conditions of the ship's power system and its strong time-varying characteristics, once it breaks down, it will often cause serious consequences. Its safe and reliable operation will directly affect the safety of ship operation. Intelligent fault diagnosis of ship power system is one of the important methods to ensure the safe and reliable operation of the system. Based on the analysis of the development of fault diagnosis system for marine power system at home and abroad, this paper aims at the existing problems such as weak on-line diagnosis ability and so on. The intelligent fault diagnosis method of ship power system is studied. Firstly, the reasoning method of fault diagnosis expert system of ship power system and the construction of diagnosis rule base are studied by rule engine technology. An expert system scheme with Drools as the inference engine is developed, and the main fault modes of ship power system are analyzed and summarized. In order to solve the problems such as the bottleneck of knowledge acquisition and the poor processing ability of incomplete information, the data-driven fault diagnosis method is studied. The fault diagnosis model is constructed by SOM neural network to make up for the deficiency of expert system. At the same time, in order to realize the transformation from post-diagnosis to pre-diagnosis, it is based on fault diagnosis. The trend prediction method of the main state parameters of ship power system is studied. The ARMA model and the wavelet neural network model are adopted, and the characteristics and applicable scope of the two models are compared. According to different state parameters, different models are chosen to predict the trend, which can alarm the abnormal parameters in advance, and have certain guiding significance for the daily maintenance of ship power system. This paper takes "Donghai Rescue 117" as the application object, on the basis of the original ship power system condition monitoring system. According to the research of fault diagnosis method and trend prediction technology, the function of fault diagnosis system is designed. The implementation method is studied. The Windows operating platform is used as the operating platform and the Java language is used as the development language. The function of fault diagnosis expert system is realized by using Eclipse as a development tool, which provides a new method for the maintenance of marine power system.
【学位授予单位】：武汉理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：U664.81;U672.74

【参考文献】