k-nearest neighbor fault monitoring based on multi-block information extraction and Mahalanobis distance

Summary: For traditionalkem>In the neighbor fault monitoring algorithm, only the observation information of the neighbor samples is considered, and a fault monitoring method based on multi-block information extraction and Mahalanobis distance is proposed.By mining the accumulated information and rate of change information implicit in the original data, it improves the traditional kNearby-neighbor fault monitoring algorithm has a good effect on faults such as small offset and pulse oscillation.Monitoring performance.Construct three types of information sub-blocks combined with observation data at the same time, based on Mahalanobis distanceIntegrate with Bayesian strategy to construct new statistics for monitoring.Numerical simulation of the proposed method and application to Tennessee—Eastman (TE) process and blast furnace ironmaking process fault monitoring, the simulation results verify the effectiveness of the method and monitoring performance.

<