An Enhanced 2D-PID Adaptive Strategy for Batch Processes through Set-point-Tuning Indirect Iterative Learning Control

Abstract

To optimize productivity growth in batch processes, it's imperative to effectively manage nonlinearities and dynamically shifting process parameters. A cutting-edge approach to tackle this challenge involves integrating an auto-tuning-neuron-based proportional-integral-derivative (ANPID) system with an indirect PID-type iterative learning control (ILC) method, resulting in an innovative two dimension proportional-integral-derivative (2D-PID) adaptive recipe. This intensified two-dimensional (2D) control strategy offers a robust solution for addressing the complexities inherent in batch processes, ultimately fostering enhanced efficiency and performance. This method targets industrial processes characterized by nonlinearities and time variations across multiple batches. The ANPID addresses intra-batch nonlinearities and time variations autonomously. Additionally, an adjustable set-point-related PID-type ILC improves local tracking capability between batches. Historical batch data iteratively informs productivity improvements. Initial PID and ILC parameters are optimized via Particle Swarm Optimization (PSO) procedure. A fermenting reactor simulation illustrates the proposed concept's potential application. The performance metric "Average Absolute Tracking Errors" (ATE) is frequently used in batch processes to evaluate the efficacy of tracking control. Comparing the enhanced 2D-PID to the conventional 2D-PID, which has an ETA of 0.0223, the latter has the lowest ETA, and the conventional 2D-PID demonstrates superior tracking and control effects by 0.0389. The finding indicates that enhanced 2D-PID adaptive controller adapts more effectively to the dynamic conditions of the process, a property that could be further exploited to optimize batch cycle times and throughput.