Cleanroom HVAC systems are inherently energy-intensive, as they are required to maintain precise control over both temperature and humidity within stringent limits. Manual controls are no longer adequate, and they cannot adjust to changes in the environment in real time. To provide
intelligent, real-time control of cleanroom HVAC, we developed a lightweight AI system using the Classification and Regression Tree (CART) method. Based on real-time inputs, the CART model is used to identify the current system condition and to accordingly predict the optimal operating parameters for the AHU (Air Handling Unit). The entire computation is carried out locally, which enables faster response times and eliminates dependence on cloud-based systems. The model’s relatively simple and interpretable structure makes it easy for operators to understand
and makes it suitable for edge-level implementation. This leads to the improved overall system performance in terms of robustness and reliability. We tested the proposed system in an ISO Class 8 cleanroom under controlled conditions. During the trials, the temperature remained
stable and the humidity stayed within the required range. In addition, the system consumed less energy compared to the earlier setup. The results show that the approach can enhance cleanroom performance while reducing energy consumption.