Simulation of Model-Based Temperature and Humidity Control Strategies for Soy Sauce Koji Production

To address the strong coupling, nonlinearity, and disturbances caused by microbial biochemical reactions in temperature and humidity control during soy sauce koji production, this study presents an engineering application and comparative investigation of three progressive control strategies: feedforward decoupling PI (FFD-PI) control, particle swarm optimization-based PI (PSO-PI) control, and fuzzy gain-scheduled PI (FGS-PI) control. A dynamic temperature–humidity model is first identified, and a multi-input multi-output closed-loop system based on PI control is established. Feedforward decoupling is then introduced to reduce the interactions between the temperature and humidity control loops, followed by offline optimization of the two PI controllers using particle swarm optimization. Based on the optimized parameters, fuzzy gain scheduling is further applied to adjust the controller gains online. Simulation results show that all three strategies achieve effective temperature and humidity setpoint tracking. PSO-PI improves the dynamic response and disturbance rejection capability, whereas fuzzy gain-scheduled PI provides better overshoot suppression and lower error indices. Under the koji-turning disturbance, the temperature and humidity overshoots are reduced by 62.2% and 48.3%, respectively. These results indicate that the PSO-initialized fuzzy gain-scheduled PI method improves temperature and humidity control performance and provides a useful reference for the intelligent control of koji production environments.