Data-driven fault detection for linear systems: A q-step residual iteration approach

摘要

This article studies the problem of data-driven (DD) fault detection (FD) for linear systems. First, based on the DD realization of kernel representation, new residual generators are designed via a q-step residual iteration method. Then, it is proved that the proposed residual generators guarantee the stability andL(1)performance of the FD error systems, and the presented residual design method is more sensitive to faults than the existing ones constructed directly from available process data. Finally, two simulation examples are provided to verify the effectiveness and advantages of the designed method.