Tuning strategy for dynamic matrix control with reduced horizons

In Dynamic Matrix Control (DMC) algorithm, the control signal is computed optimally based on the process model. In effect, the DMC algorithm allows for obtaining a better quality of control than conventional controllers, especially for plants with large time delays. However, in spite of these advantages, there are still some difficulties that can appear in the implementation of DMC in local control loops. This is due to limitations of the computational resources in industrial devices (e.g., Programmable Logic Controllers). To overcome these difficulties, we propose a tuning strategy for the DMC algorithm with reduced horizons. It is shown that a reduction in the length of prediction and dynamic horizons can reduce the required memory in industrial controllers without degrading the quality of control.