New Predictive Control of Systems through Band-Limited Networks

A method for closed-loop control of systems through band-limited networks is proposed. The proposed method is a Network Model-Based Predictive Control (NMBPC) methodology. In traditional NMBPC methods, using the recorded recent input/output history of the plant, and associated with various possible time delays, a vector of stabilizing control input signals are determined and sent in a control packet, to the plantside. This vector is received by a time delay compensator module, preceding the plant, which corresponding to the actually occurred time delay,selects just one entry of the control vector, and applies it to the actuator. In practice, difficulties may arise when the communication network is of low bandwidth. The reason is that the conventional Discrete-Time (DT) controllers used with the NMBPCs can guarantee the closed-loop stability only when the sampling frequency is high enough. Therefore, the time resolution, and hence, the required length of the control input vector need to be increased substantially. In order to overcome such a difficulty, a modified NMBPC with a new digital control design methodology is proposed here which can guarantee the closed-loop stability for large sampling periods. Case studies performed on well-known benchmark problems, demonstrate that the proposed approach is both accurate and effective enough for a real time implementation.

کلیدواژه ها:

Networked Control Systems ، ModelPredictive Control ، Digitally Redesign Method ، PlantInput Mapping (PIM)