Interaction Analysis and Decomposition Principle for Control Structure Design of Large-scale Systems

Industrial processes are mostly large-scale systems with high order. They use fully centralized control strategy, the parameters of which are difficult to tune. In the design of large-scale systems, the decomposition according to the interaction between input and output variables is the first step and the basis for the selection of control structure. In this paper, the decomposition principle of processes in large-scale systems is proposed for the design of control structure. A new variable pairing method is presented, considering the steady-state information and dynamic response of large-scale system. By selecting threshold values, the related matrix can be transformed into the adjoining matrixes, which directly measure the couple among different loops. The optimal number of controllers can be obtained after decomposing the large-scale system. A practical example is used to demonstrate the validity and feasibility of the proposed interaction decomposition principle in process large-scale systems.