Accurate temperature control in tea water-removing is the key to determine the quality of tea. The PID controller is universally used in this field. However, the choice of controller parameters has a direct effect on performance. Traditional method based on manual experience normally leads to subjectivity, long cycle, and large fluctuations. Therefore, this paper proposes an improved sparrow search algorithm to tune the quantization and scale factor of the fuzzy PID controller. Firstly, the adaptive fuzzy controller for the tea cylinder water-removing machine is established. Secondly, in addition to the levy flight strategy, random mutation and perturbation are introduced into the sparrows’ behaviour to enrich their diversity. Moreover, dynamic staged and real-time feedback adjustment strategies are also proposed to increase the diversity of the population size. Finally, traditional methods and four prior arts are compared with the proposed method to verify the superiority. The simulation results show that the fuzzy PID controller optimized by CSDFSA has excellent performance in terms of stability and dynamic response. Additionally, this controller also exhibits significant robustness under external signal interference, which can improve the performance and intelligence level of the tea water-removing temperature control system. In conclusion, this manuscript proposes an intelligent fuzzy control system through the intersection of computer science and agricultural engineering and applies it to the processing of tea. With the help of computer technology to promote the development of the famous tea industry.