An industrial process optimization framework: from data to deployment with case studies in food production processes

Abstract

Control systems are crucial for ensuring operational efficiency, safety, and product consistency in industrial environments. Industrial AI, which integrates artificial intelligence technologies into traditional control systems, offers significant performance enhancement opportunities. However, current approaches face two fundamental challenges: theoretical research often develops advanced AI methods disconnected from practical industrial needs, while industrial implementations tend toward fragmented, case-specific solutions (addressing issues in isolation without systematic approaches), hindering standardization and scalability. To bridge these gaps, we propose an industrial AI development framework designed to systematically enhance system performance through integrated methodologies encompassing data acquisition and analysis, model development, and system deployment. The framework’s effectiveness and generalizability were validated through applications in two industrial processes: steam flow management in wort kettles and filtration process optimization. Empirical testing demonstrated the framework’s generalizability across different control paradigms, achieving substantial performance improvements compared to the conventional control system. Beyond operational improvements, this framework could enable substantial energy efficiency gains and carbon footprint reduction through optimized process control, addressing growing sustainability imperatives in energy-intensive industrial operations.