Design and Implementation of Multi -Criterion Decision Systems Based on Artificial Intelligence in Optimizing Production Process in Food and Drinks Industries

This study introduces an innovative, multi-layered artificial intelligence (AI) framework designed to optimize food and beverage production processes, with an emphasis on sustainability, safety, and operational efficiency. The framework integrates hierarchical fuzzy analytic hierarchy process (fuzzy AHP) for multi-criteria decision-making, advanced data preprocessing techniques including auto encoders and feature engineering, and diverse AI models such as deep reinforcement learning (DRL), convolutional neural networks (CNN), and gradient boosting machines (GBM). A comprehensive digital twin environment simulates real-time plant operations, enabling validation and scenario analysis. The proposed methodology employs a ten-round decision refinement process, incorporating expert judgment, sensor data, visual defect detection, and predictive analytics to dynamically control process parameters, quality assurance, and resource allocation. The case study conducted within a Tehran dairy processing plant demonstrates substantial improvements in operational metrics: a reduction of energy consumption by 8%, microbial counts by 15%, waste by 10%, and zero safety violations over a three-month period. The integration of explainable AI (XAI) techniques enhances interpretability and stakeholder trust. The findings underscore the potential of such integrated AI-driven systems to revolutionize Industry 4.0 practices in food manufacturing offering pathways toward smarter, safer, and more sustainable production paradigms. This research provides a scalable, adaptable blueprint for future deployment across diverse industrial contexts.