多尺度强跟踪滤波及在非线性系统微小故障诊断中的应用

Recently there are more and more multiscale phenomenon or multiscale effect arising in practice, so it is more important to theoretically solve the multisensor fusion problem when sample at different rate. With the complexity of the object processed, and the nonlinear property of the dynamic model describing the object's characteristic, the solution of the multiscale data fusion in nonlinear dynamic system is going to be a research subject of topic. For the reasons above, the multiscale random dynamic model and the multiscale recursive data fusion algorithm have been established for the dynamic system with sensor of different characteristic on different scale in this task. In addition we put forward the multiscale strong tracking theory, establish the multiscale state fusion algorithm for multisensor and joint estimation method of state as well as parameter and the fault detection and diagnosis method for the nonlinear system. The application of this theory to practical equipment be carried out synchronously. One volume of monograph and about 12 articles in this field have been published or accepted. For the advantage above, the research based on this field will be of important scientific signification and widely application value.

针对现代、高性能、多层次、复杂系统中出现的各种各样故障，对其现象或过程进行不同尺度或分辨率观测、描述、分析和综合，从而诊断或预测出故障发生或可能发生的部位与时间。建立多尺度强跟踪滤波理论框架，提出基于随机动态系统的多尺度故障诊断算法，并应用于非线性动态系统微小故障的早期诊断，将实验验证后的部分成果应用于实际工业对象。