Robust Model Predictive Control For Industrial Systems With Large Operating Regions

In many industrial systems, the dynamic characteristics varies with different working conditions, hence the model parameters changes at different operating points. In this paper, we propose a novel Robust Model Predictive control strategy based on Linear Matrix Inequalities to deal with the industrial process with wide operating regions. This new control strategy will reduce the conservative of the classical RMPC and consider the practical problems in real processs (hard and soft constraints, the unobserable states).First, piece-wised linear time varying system with polytopic uncertainty is built to better approximate the industrial systems. Systems with wide operating regions can be expressed as a series of the linear time-varying systems around the operating point of. Then, we build local RMPC controller with explicit stability region for each operating region and expand each stability region of local RMPC controllers to determine the whole stability region. In this way, the control of systems with large operating regions has been simplified into a convex optimization problem with a set of LMI. The output feedback is used with the state observer to deal with the unobserverable states in most industrail systems.The proposed industrial robust predictive control algorithm has some theoretical significance because the parameters time-varying Lyapunov function is used instead of a single Lyapunov function and the additional degrees of freedom is added to input instead of a fixed feedback control to reduce the conservative and improve the control performance. On the other hand, the algorithm also has the engineering value, and can handle common industrail systems with large operating regions, mixed constraints (hard and soft constraints) and unobserverbale states.