Objective: Solve the problems of difficulties of Daqu fermentation detection, judge the fermentation state judgement, and controlling. Methods: The Tent-SSA optimized BP neural network algorithm Daqu fermentation humidity prediction model and dynamic threshold control algorithm were proposed to realize real-time judgment of Daqu state and Daqu fermentation control during Daqu fermentation process. Results: The error predicted by The simulation prediction model for humidity prediction had low error (0.596%), good robust performance and fast convergence speed. Conclusion: The Daqu monitoring system based on this model is accurate and reliable.