Aiming at the refined requirements of temperature in the process of scented tea baking, a control system composed of fuzzy control, neural network and PID algorithm was proposed to realize the reliable control of temperature and humidity and to achieve the purpose of rapid dehumidification and drying of flowers. The system taken the temperature and humidity error e and error rate ec as input variables for fuzzy processing, established membership function, through fuzzy reasoning and fuzzy production into the output, for PID control system to provide optimal regulatory factor △K, accurate control of the executive unit. The system adopts RBP neural network algorithm to fine-tune the real-time variables in the fuzzy control process, and USES Delta (δ) function and gradient descent algorithm to adjust the network weighting coefficient w, central vector c_ij and base-width vector b_ij as learning rules. Data simulation system by using Matlab software, compare the traditional PID control and fuzzy control, and verify the superiority of the combination of fuzzy neural network PID control, the data analysis shows that the combination of temperature control system response speed and the accuracy of the data are obviously improved, the disturbance compensation and improve anti-jamming capability, and it has better robustness.