THE RESEARCH OF PROCESS MONITORING BASED ON DATA FUSION THEORY

This paper explores the multisensor process monitoring system which sampling is synchronous and transmission delay is less than one sampling period; A new process monitoring algorithm is put forward through introducing recursive least square estimate and combing the multisensor data fusion theory with the traditional process monitoring technology based on principle components analysis. Firstly, the paper analyzes the existing questions about the traditional process monitoring methods based on measurement in detail; secondly, in single sensor case it extends the analysis founded on measurement matrix to that based on state estimate value matrix by using least square estimate. So it can filter the measurement noise effectively. Afterwards it proposes that it can fuse the measurement of every sensor which gets to central processor step by step through using recursive least square estimate in multisensor case; finally, we can use the principle components analysis to realize real-time monitoring for complex process. The proposed method can improve the accuracy and enforceable ability of process monitoring technology effectively, and reduce false alarm. Computer simulations show the validity of the proposed method.