AI-Based Surrogate Models for the Food and Drink Manufacturing Industry: A Comprehensive Review

Abstract

Surrogate models provide virtual representations that mirror physical objects or processes, serving distinct purposes in simulations and digital transformation. This review article examines how integrating surrogate modelling with artificial intelligence (AI) techniques can facilitate the iterative development of surrogate models and identify instances where additional data acquisition is necessary to enhance the performance of a surrogate model. This demonstrates the potential of combining AI with surrogate modelling in addressing some of the key challenges in the food and drink manufacturing industry. The paper also provides an accessible examination of AI and surrogate modelling in the food and drink manufacturing industry, offering a summary of current applications and advancements within the field. The key areas addressed by this article include the application of AI and ML in process control, prediction, and modelling for food manufacturing, as well as the advantages and limitations of AI-based surrogate modelling (SM), among other issues addressed. Based on the literature reviewed herein, AI-based surrogate models can be employed to optimise production processes and reduce the need for extensive physical prototyping in the food and drink manufacturing industry. This review emphasises AI-based surrogate modelling techniques tailored for complex food processing systems and distinguishes itself by bridging method-specific insights with practical industrial relevance. Additionally, this article reviews challenges and limitations in the food and drink manufacturing industry and the application of surrogate modelling, along with future directions for research in this rapidly evolving field.

Carlos Eduardo da Silva

Senior Lecturer in Software Engineering and leader of ASERG

Leader of ASERG with research interests exploring the interplay between software engineering and cybersecurity.