Data-driven Based Integrated Learning Controller Design for Batch Processes

The challenge of optimization control of batch processes is how to combine both discrete-time (batch-axis) information and continuous-time (time-axis) information into an integrated frame when designing optimal controller. By using data-driven technology, a novel integrated learning control system is proposed in this paper. Firstly, an iterative learning controller (ILC) is designed along the direction of batch-axis, and then an adaptive single neuron predictive controller (SNPC) that plays role of feedback controller along the direction of time-axis is devised accordingly. As a result, the integrated control system is very effective to eliminate modeling error and uncertainty, which is superior to traditional ILC. In addition, the self-tuning algorithm of SNPC controller is derived by a rigorous analysis based on the Lyapunov method such that the predicted tracking error convergences asymptotically. Lastly, to verify the efficiency of the proposed control scheme, it is applied to a benchmark batch process. The simulation results show that the proposed method has better stability and robustness compared with the traditional iterative learning control.