一种基于数据驱动的故障诊断与容错控制方法研究

With China's rapid economic development, the complex industrial system is playing a very important role in our national industrial and economic construction. The diagnosis of non-normal operation conditions with the safe operation methods and implementation technology have become key issues for both industrial and academic domains. Although the model-based fault diagnosis theory has been well established for the past two decades, the physical modeling of complex industrial systems is still very difficult in practice. This proposed project aims to solve key theoretic and technical problems of diagnosing non-normal conditions and making safe operation for complex industrial processes. Since it is very difficult to have an accurate physical model of a complex system, based on a great number of historic and online data with combining fault mechanisms and advanced information processing methods, the main objective of this proposed project is to achive efficient fault diagnosis and tolerant control in an integerated framework. Based on it, the prediction of non-normal conditions as well as optimal maintenance time prediction method will be further researched. It will build up a solid theoretical foundation of solving the problem of safe and highly efficient operations of complex industrial processes.

随着我国经济的高速发展，复杂工业系统在国民经济中所占比重日益增加，其非正常工况诊断与系统安全运行方法已成为国内工业界和学术界亟待解决的关键课题。尽管基于机理模型的故障诊断理论在过去二十年中有了长足的发展，但复杂工业系统的机理建模在实际中依然十分困难。本课题针对复杂工业系统的非正常工况诊断与安全运行核心理论及关键技术展开研究。在复杂工业系统难以建立精确机理模型的前提下，基于大量历史数据及在线数据，融合关键参数辨识理论，提出基于数据的故障诊断与容错一体化设计框架。并在此框架下，研究非正常工况预报及最优维护时机选择方法，为复杂工业系统的安全、高效运行奠定理论基础。

随着我国经济的高速发展，复杂工业系统在国民经济中所占比重日益增加，其非正常工况诊断与系统安全运行方法已成为国内工业界和学术界亟待解决的关键课题。尽管基于机理模型的故障诊断理论在过去二十年中有了长足的发展，但复杂工业系统的机理建模在实际中依然十分困难。本课题针对复杂工业系统的非正常工况诊断与安全运行核心理论及关键技术展开研究。通过本项目，对数据驱动的控制器关键参数设计问题、基于数据的故障检测系统参数直接设计问题及滤波问题、基于关键性能指标的故障诊断及预测问题等进行了研究并取得了一系列良好的成果，成功建立了基于数据的故障诊断与容错一体化设计框架。