基于概率图模型的电动汽车系统可靠性分析关键技术研究

项目名称： 基于概率图模型的电动汽车系统可靠性分析关键技术研究

项目编号： No.61203184

项目类型： 青年科学基金项目

立项/批准年度： 2013

项目学科： 自动化学科

项目作者： 钟小品

作者单位： 深圳大学

项目金额： 24万元

中文摘要： 系统可靠性分析是整个电动汽车产品周期中众多应用的核心支撑技术。由于复杂、耦合和未知噪声等特点，电动汽车产业化需要更精准的系统可靠性分析方法。除了组件的失效，电机、电控等系统的失效隐患更多来自于组件之间不确定的关联性和失效逻辑结构。因此本项目提出主动地构建失效逻辑结构的新思路，以概率图模型为理论框架，将系统可靠性分析与故障诊断等转换为不确定信息下学习和推理的问题。据此，本申请拟研究以下主动式系统可靠性分析的关键技术：(1)研究基于概率图模型的电动汽车系统可靠性表示技术；(2)研究基于机器学习的主动构建电动汽车系统失效逻辑结构的方法；(3)研究电动汽车系统中不确定性传递的机理以及在表示模型中快速推理的技术。在此基础上，研制一套系统可靠性分析系统，对提出的理论方法和技术方案进行验证和评估。预期在上述各方面获得理论和技术进展，对电动汽车等复杂机电产品周期中各种应用的研究起到积极的推动作用。

中文关键词： 电动汽车；系统可靠性分析；故障诊断；概率图模型；

英文摘要： System reliability analysis (SRA) is a core support technique of many applications in the entire product lifecycle of electric vehicle (EV). However, the large-scale industrialization needs an accurate SRA due to the features of complexity, coupling and unknown noise of an EV. Besides component failure, the latent failure of electrical motor, electrical control and other subsystems is more originated from the uncertain dependency between components and the uncertain logical structures of failure. Therefore, this research puts forward such a new idea: in the theoretical framework of probabilistic graphical model, the logical structures of failure are constructed proactively; SRA and fault diagnosis are converted to the problems of learning and inference with uncertainty. Accordingly, this research intends to study the key technologies of the proactive SRA as follows: (1) the representation technique of SRA for EVs based on probabilistic graphical model; (2) the proactive construction of EV system failure's logical structure based on machine learning; (3) the mechanism of uncertainty propagation in an EV system and the fast inference technique in the proposed SRA representation. On this basis, an SRA platform will be developed and used to verify the proposed methods and techniques. Theoretical and technical progre

英文关键词： electric vehicle；system reliability analysis；fault diagnosis；probabilistic graphical model；