基于概率图模型与序列蒙特卡洛方法的航天器自主故障诊断与健康监测技术研究

项目名称： 基于概率图模型与序列蒙特卡洛方法的航天器自主故障诊断与健康监测技术研究

项目编号： No.60874054

项目类型： 面上项目

立项/批准年度： 2009

项目学科： 交通运输

项目作者： 沈毅

作者单位： 哈尔滨工业大学

项目金额： 28万元

中文摘要： 针对航天器尤其是深空探测中的航天器，所处环境复杂、任务周期长，对安全性、可靠性和自主性要求高的特点，为了提高航天器的自主健康监测能力，深入研究了适用于非线性、非平稳信号自适应处理的经验模态分解方法以及相关的特征识别与分类方法，达到从大量数据中实现航天器状态信息的有效特征提取，并研究了冗余分析技术与经验模态分解结合的航天器故障诊断方法；采用概率图模型表征系统各组成部分之间的因果关系及结构信息，研究了基于系统历史与实时数据的故障推理方法，以提高航天器的自主故障诊断能力；基于预测滤波器和时间序列分析理论的长时间、分段式执行器故障重构与故障预报技术；从提高航天器自主性和可靠性的角度出发，研究了故障诊断与容错控制一体化设计方法、容错控制器的在线选择与管理策略，以满足多样的安全管理要求，达到提高航天器的自主监测能力和健康管理水平；针对于未来航天器可能使用无线传感网络的问题，本研究还探索了分布式无线传感网络的传感器配置方法，为将来航天器无线传感网络的应用奠定基础。

中文关键词： 航天器；故障诊断；健康监测；概率图模型；容错控制

英文摘要： Spacecrafts, especially the ones in deep space, are required to have high security, reliability, and autonomy due to the complexity of the environment and the long duty cycle. In order to improve the ability of self-health monitoring of the spacecraft, we have studied the empirical mode decomposition method for adaptively processing nonlinear and non-stationary signals, and have deeply investigated its associated feature recognition and classification methods as well. These methods are very effective to extract the useful status information of the spacecraft from a large number of data. Furthermore, by combing with the method of redundancy analysis, some new methods for the spacecraft fault diagnosis are proposed. By using the probabilistic graphical model to denote the relationship for different components of the system and the structure information, the fault reasoning method based on the system historic data and real-time data is put forward to improve the capability of autonomous fault diagnostic for spacecrafts. A new method concerning the long-time, segmented actuator fault reconstruction and actuator fault forecasting techniques is presented by utilizing the prediction filter and the time series analysis. From the aspects of improving the autonomy and reliability of spacecrafts, the integrated design of fault diagnosis and fault-tolerant control is studied. A general online selection and management of fault-tolerant controllers is also studied for different practical demands to improve the capability of self-health management of spacecrafts. On account of utilizing wireless sensor networks for future spacecrafts, the study also explores the methods of sensor configuration for the distributed wireless sensor networks as the foundation for the future applications of the spacecraft wireless sensor network.

英文关键词： spacecrafts; fault diagnosis; health monitoring; probabilitic model; fault tolerant control