Objective In response to the limitations of traditional temperature control methods (such as PID control), including slow response, significant overshoot, and the inability to fully utilize predictive information in dynamic, multi-variable complex scenarios, an intelligent food temperature control method combining fuzzy control and the Transformer prediction model is proposed. This method aims to improve temperature control accuracy, response speed, and energy consumption efficiency.Methods The Transformer model is used to predict the trend of future multi-time-step ambient temperature and target temperature, and real-time temperature deviation control is implemented based on fuzzy logic. The proposed method is experimentally validated using pizza billet temperature control as an example.Results The experimental results show that the proposed intelligent temperature control method outperforms traditional methods in both prediction and control performance. Compared with other prediction models, the mean absolute error (MAE) of the Transformer model is reduced by 21.30% (from 0.19 ℃ to 0.15 ℃), and the root mean square error (RMSE) is reduced by 16.67% to 25.00% across three time-series predictions. In the cold storage temperature control scenario, compared with the other four methods, the temperature overshoot is reduced by 15.73%~39.27%, the response time is shortened by 14.24%~33.52%, and the steady-state error is reduced by 33.30%~62.50%. In the oven temperature control scenario, compared with the other four methods, the temperature overshoot is reduced by 11.24%~33.05%, the response time is shortened by 11.54%~33.03%, and the steady-state error is reduced by 40.00%~71.43%.Conclusion The intelligent temperature control method combining fuzzy control and Transformer shows excellent prediction accuracy and control performance in complex food storage and processing scenarios, with significant performance advantages over other methods.