Intelligent Dustbin Temperature Control System Based on Evaporative Cooling: Enhanced Prediction and Control of Temperature Systems Using Fuzzy PID strategies

Food waste, characterized by its high perishability, odor emission, and environmental impact under high temperatures, necessitates storage at low or normal temperatures. Traditional cooling methods, such as air conditioning, consume substantial energy. To address this, we designed an evaporative cooling dustbin. Given the strong influence of external environmental factors on cooling efficiency, we collected experimental data and trained three neural network models to improve the accuracy of bin temperature prediction and regulation. We enhanced the conventional proportional-integral-derivative (PID) control algorithm with neural networks, developing three distinct control strategies. Model performance was evaluated based on prediction accuracy and control efficacy. Results indicated that augmented with feature encoding, the long short-term memory (LSTM) model achieved the highest prediction accuracy with a mean error of ±0.3°C. The fan speed prediction model also demonstrated a strong correlation, with an R² value of 0.9804. Optimal fan speed control was achieved using a fuzzy PID model informed by the LSTM algorithm. Validation tests during operational hours showed temperature errors of 0.45°C and 0.54°C for two different periods. These results highlight the ability of the enhanced LSTM model to accurately predict bin temperature, while the optimized PID strategy effectively stabilizes temperature fluctuations. Additionally, two working modes—performance and economic—were established, both of which can maintain the average temperature of the garbage box at 25 ± 0.3°C. In the performance mode, the overshoot time of the system cooling is at least 46 s, and the response is rapid and has high stability. This work advances precise prediction and control methods for nonlinear temperature systems in energy-efficient cooling applications.