Detection of control loop interaction and prioritization of loop maintenance

Abstract

In automatic control system of a chemical process, tens or hundreds of control loops are used in a multiloop control con guration. Loop interaction arises in these loops when changes are made in one of the parameters of the control loops. It is di cult to measure the degree of interactions between the control loops and rank the loops according to their extent of interactions. This study presents two novel data driven techniques to determine control loop interactions and rank the loops according to their importance. In rst approach, two new indices have been developed using integral of absolute or squared error (IAE or ISE) criteria to quantify loop interaction and determine rank of the loops. The rst index calculate the presence of interaction from one loop to another using IAE or ISE of the two interacting loops. Then by using these loop interaction values a loop interaction matrix has been formed. The second index uses the loop interaction matrix to calculate the rank of the loops. In another approach, a novel method based on canonical correlation analysis has been developed to calculate interaction among the loops. An index has been proposed in this method to determine the ranking of the loops and the ranking has been done with respect to the maximum canonical correlation value of the control loops. The proposed methods have been evaluated using both simulation and experimental studies. Both simulation and experimental results demonstrate the validity and e cacy of the proposed methods.