Applications of metaheuristic optimization algorithms in model predictive control for chemical engineering processes: A systematic review

Abstract

The growing competitiveness of the chemical industry, along with sustainability demands and regulatory requirements, calls for optimized and well-controlled operations. Chemical engineering processes are often characterized by non-linearity, strong variable coupling, dead times, multiple inputs and outputs, and operational constraints, making control strategies challenging. Model predictive control is widely used for its advantages in optimal control, flexibility, robustness, and ability to handle multi-objective tasks. However, precise tuning and optimization are essential for implementing this strategy in real-time applications. Metaheuristic optimization algorithms offer an alternative to traditional optimization methods, as they can quickly reach near-optimal solutions and avoid local minima, making them well-suited for use with model predictive control. This study aims to analyze the application of metaheuristic optimization algorithms in conjunction with model predictive control in chemical engineering processes through a systematic review. The review considers three eligibility criteria: applying model predictive control for process control, utilizing metaheuristic optimization algorithm, and chemical engineering-related processes. A total of 46 studies were analyzed, revealing three main application areas for metaheuristic optimization algorithms in model predictive control: improving dynamic models used in the receding horizon, tuning model predictive control parameters, and serving as optimizers in the model predictive control formulation. Over 20 different metaheuristic optimization algorithms and various process models were identified, with typical applications including continuous stirred tank reactors, tank-level control, and column distillation. Genetic algorithms and particle swarm optimization were the most frequently used algorithms. This review concludes that metaheuristic optimization algorithms have been successfully applied to enhance model predictive control in several processes. It also highlights the benefits, weaknesses, and limitations of metaheuristic optimization algorithms applications in chemical engineering processes and provides recommendations for future research. We hope this study will be valuable to professionals and researchers in chemical engineering and process control.