Improved cascade controller design methodology based on outer-loop decomposition

Abstract

In this study, a new serial cascade controller design methodology is proposed. The inner-loop process is taken as a stable first-order process with time delay and the outer-loop process as an integrating or non-integrating nth-order stable process with time delay. Classical Smith predictors are used to compensate for the time delay in each loop. In the outer loop when the process order is greater than 1, the process model transfer function is decomposed into first-order transfer functions of the required quantity. Next, for each first-order transfer function, a control loop is established and then for each loop an appropriate controller is built up. The controllers are designed to provide a simplified closed-loop transfer function for the existing loop using zero pole cancellation. The decomposed first-order transfer functions results in either an integrator form or a transfer function with one real pole. The resulting controller evolves to be a proportional-type controller for the loops involving the integrator and they appear as a PI-type of controller for the loops involving the real stable pole. To illustrate the effectiveness of the proposed methodology, it is compared with another method given in the literature via simulations.