项目名称: 数据驱动的非线性多模态复杂系统性能退化故障预测方法研究
项目编号: No.61273173
项目类型: 面上项目
立项/批准年度: 2013
项目学科: 自动化技术、计算机技术
项目作者: 马洁
作者单位: 北京信息科技大学
项目金额: 80万元
中文摘要: 以工业生产过程中带有性能退化部件的复杂系统为研究对象,基于多测量变量数据统计建模和故障重构技术提出一套用于非线性、多模态复杂系统的故障预测研究框架,最终得到系统整体性能的故障预测方法和剩余有效寿命估计技术,开发适合实际应用的算法和软件系统。一般复杂系统都具有高度非线性、变量多、耦合强、干扰多、运行模式多变等特性,无法建立精确静态的过程模型,因此基于模型的故障监测技术难以直接使用。而数据驱动的统计过程监控技术则得到了广泛应用。围绕复杂系统的故障预测与和剩余有效寿命估计问题,拟研究和解决以下几个方面问题:(1)提出基于非线性数据模型的故障描述,并研究非线性系统的故障预测技术;(2)提出基于多模态数据模型的故障描述,包括有监督的和无监督两类,并研究多模态系统的故障预测技术;(3)提出非线性多模态系统的故障预测技术。
中文关键词: 非线性;多模态;故障重构;故障预测;剩余有效寿命估计
英文摘要: Aiming at the complex systems with performance degrading components in industrial processes, a prognosis and health management frame is proposed based on multivariable statistical modeling and fault reconstruction, and the fault prognosis and remaining useful life estimation for the whole system performance is obtained. At last, the algorithm and software realization are developed for practical use. Most practical systems are highly nonlinear, multivariable, strongly disturbed, and multimodal. It is hard to build explicit static models for these systems and use model based fault detection methods directly. On the other hand, data-driven statistical process monitoring technologies are widely used. Towards the topic of fault prognosis and remaining useful life estimation for complex systems, several problems will be studied and solved: i) propose the fault description based on nonlinear data models, study the fault prognosis for nonlinear systems; ii) propose the fault description for multimodal data models, including the supervised and unsupervised cases, and study the fault prognosis for multimodal systems; iii) propose the fault prognosis for nonlinear multimodal systems.
英文关键词: nonlinear;multimodal;fault reconstruction;fault prognosis;remaining useful life estimation