Air conditioning systems with the classical ON/OFF control strategy depend on temperature sensor readings and a fixed set point. However, this may result in overheating or overcooling, as the sensor measures only its local temperature rather than the average temperature in the target zone (T avg ). To address this limitation, a neural network‐based control strategy has been proposed, developed, and tested, without modifying the air‐conditioning system components. The model, named an inverse neural network (INN), trained and tested using historical data, can reveal the time‐dependent correlations between T avg and the sensor temperature readings while the space is under dynamic conditions. Using these correlations, the set point is updated at each time interval to maintain T avg within a desired range. To validate the effectiveness of the proposed approach, a case study was conducted for an office room under dynamic thermal loads, including outdoor conditions and human activities. Although the outdoor conditions were derived from 2 years of Adelaide weather data across summer and winter months, minute‐by‐minute room data over this period were generated using computational fluid dynamics (CFD). To support indoor air quality (IAQ), the system was operated without air recirculation, such that outdoor air was supplied when the AC system was OFF. Data from the first year, simulated under classical control, were used to train and test the INN model. A comparative analysis was then carried out between the proposed INN‐based approach and the classical control method using the second‐year data. The results show that the INN‐based approach increases the percentage of time during which the T avg remains within the desirable range by 20%–70%, while also reducing the AC system′s energy consumption by approximately 5%–20%.
Nia et al. (Thu,) studied this question.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: