Stochastic Iterative Learning Model Predictive Control based on Stochastic Approximation

Iterative learning model predictive control (ILMPC) is an effective control technique for improving the performance of a batch process under model uncertainty and rejecting real-time disturbances. Industrial batch processes often have stochastic disturbance and noise and ILMPC cannot guarantee convergence for such systems. In this work, we propose a novel stochastic ILMPC that combines stochastic approximation with ILMPC algorithm. The proposed algorithm ensures the almost sure convergence property. In comparison with the ILMPC, the proposed control algorithm also shows better performance in terms of the tracking error.