Fuzzy Logic and Neural Network Approximation to Indoor Comfort and Energy Optimization

Research about indoor environmental satisfaction has indicated that allowing building occupants to adjust their local environment to their preferences increases thermal satisfaction and human performance at the workplace. However, such systems have been considered as a reason of possible increase in energy consumption of environmental control systems. In our previous study, we minimized the energy consumption of distributed environmental control systems without increasing occupant thermal dissatisfaction using gradient-based optimization. We then approximated the optimal models using fuzzy systems based on the nearest neighbors. Those fuzzy models were specific to different outside temperatures. This required a considerable storage space to keep all the fuzzy rules of each outside temperature. In this study, we generated a fuzzy system and a neural network system, which are generated for any outside temperature. Our results show these two models approximate the results of gradient-based optimization in a practically feasible fashion