Data Driven Fault Detection For Multi-rate Systems

In this paper,we mainly study the fault detection problem of data driven multi-sampling rate system.Firstly,a multivariate sampling rate conversion algorithm is proposed based on the multi-sampling rate conversion technique in digital signal processing.The multi-sampling rate data matrix is transformed into a complete data matrix,and then combined with the kernel principal component analysis algorithm,Which can solve the nonlinear problem which can not be solved by the traditional multi-sampling rate principal component analysis algorithm,and simulate it on the TE industrial process platform.The simulation results verify the problem of the proposed method.The effectiveness of the algorithm.However,when the process variables and quality variables related to the process,the need for partial least squares algorithm,because the partial least squares algorithm can not solve the nonlinear problem,so the introduction of the kernel function,proposed a basis for the least squares Sampling rate system fault detection algorithm,the algorithm not only solve the process variables and quality variables related to the problem,but also solve the nonlinear problem,through simulation verification,and achieved good detection results,the simulation results also illustrate the proposed Fault detection algorithm can be effectively applied in fault detection of multi-sampling rate system.