【作者】 王孟莲；

【导师】 高孝洪； 陈辉；

【作者基本信息】 武汉理工大学 ， 舰船电力推进及自动化技术， 2013， 博士

【摘要】 随着电力电子技术的高速发展以及大功率交流电机变频调速技术日趋成熟,船舶电力推进技术在国内外受到高度重视并得以迅速发展。船舶电力推进系统具有许多传统推进系统无法比拟的优势,是目前国内外船舶行业研究的热点。据当前形式看,电力推进将成为今后船舶制造行业的主流发展方向。因此,深入研究船舶电力推进系统状态评估有重要意义。电力推进系统状态评估是操纵、故障诊断和状态检修等的决策支持,建立科学、全面、合理的评估指标体系及电力推进系统的预测模型,选择合适的评估方法并最终开发切实可行的状态评估系统是电力推进系统状态评估研究的必经步骤。本文根据船舶电力推进系统的运行原理,在对电力推进系统状态信息分析的基础上建立了状态评估指标体系,并应用组合预测的方法建立了电力推进系统的预测模型,然后应用智能方法,如模糊神经网络和支持向量机,建立了船舶电力推进系统状态评估模型,设计了基于分布式架构的舰船电力推进状态评估系统,进行了状态评估模型仿真实验。本文完成的主要研究工作如下：(1)以船舶电力推进系统为研究对象,提出了电力推进系统状态评估指标的选取原则,在分析电力推进系统各组成设备运行原理的基础上,建立了船舶电力推进系统的状态评估指标体系,该体系能在一定程度上满足电力推进系统状态评估的分析要求。(2)为了预测船舶电力推进系统的运行状况,在分析各单项预测方法和组合预测方法的基础上,采用最优组合预测方法建立了船舶电力推进系统的预测模型,并在MATLAB中进行了仿真,结果显示所建立的预测模型具有较高的准确性。(3)研究了船舶电力推进系统的状态评估流程及状态评估训练样本的格式和获取方法,在此基础上分别建立了基于模糊神经网络的状态评估模型和基于支持向量机的状态评估模型。(4)研究了分布式的船舶电力推进系统状态评估系统和软件的设计方法,基于MATLAB与Visuale C++交互技术设计了简化的仿真实验系统。进行了模糊神经网络状态评估模型和支持向量机状态评估模型的仿真实验,结果表明了两种评估方法的准确性,并各有特点。更多还原

【Abstract】 With the great development of power electronic technology and high-power AC motor variable frequency drive technology, ship electric propulsion comes into a new flourishing period of development. Ship electric propulsion has many advantages over traditional propulsion, and is a hotspot research area in ship building. Currently ship electric propulsion will likely be one of the main trends in ship building. It is thus vitally important to study the condition assessment for ship electric propulsion.The conditional assessment for electric propulsion system is the basis for operation and fault diagnosis and status maintenance. To establish a scientific, comprehensive and rational evaluation index system and a prediction model of electric propulsion system, it is essential to choose a proper assessment approach and develop a realistic condition assessment system. It is a critical step in conditional assessment of electric propulsion systems condition. According to the running principle of marine electric propulsion system, a system of condition assessment has been established after analyzing condition information, and a prediction model of marine electric propulsion has been built after analyzing optimal combined forecasting. Then the Fuzzy Neural Network and Support Vector Machine in artificial intelligence techniques have been used to build up condition assessment model of marine electric propulsion system. The marine electric propulsion system condition assessment system was designed. Finally we conduct a simulation of condition assessment model with the test.The main results in the current dissertation are described as follows:The paper studies the marine electrical propulsion system, and puts forward the selection principle of condition assessment index. It also establishes the condition assessment Index system of marine electrical propulsion on the basis of running principle study. The Index system meets the requirements in aspects of analysis of electrical propulsion system condition assessment in certain degree.In order to predict running status of electrical propulsion system, after a further analysis of single forecasting methods and optimal combined forecasting method, the paper establishes the prediction model of marine electric propulsion system, then, emulates the model using MATLAB, the emulation result shows that the prediction model is better effective.After a further study of condition assessment process of marine electric propulsion system and format and acquisition method of condition assessment Training Sample, the paper establishes condition assessment model and used Fuzzy Neural Network and Support Vector Machine.The paper designs the marine electric propulsion system condition assessment system and its software, establishes a simplified laboratory test system based MATLAB and Visual C++. The simulation experiment of fuzzy neural network assessment model and support vector machine assessment model are implemented, and the results demonstrate the accuracy and characteristics of the two kinds of assessment methods.更多还原