Optimization of the Control Variables of Indoor Thermal Comfort Based on Genetic Algorithm and Neural Network

In order to improve thermal environment of office and decrease energy costs, office model is built with computational fluid dynamics (CFD) software, artificial neural network is trained with data from CFD model, and CFD model is surrogated by neural network model. Objective function of the algorithm is established and weighting factors in the objective function adjusted to obtain different optimization results. Optimal solution of indoor control variables are obtained when computational cost is decreased without loss of precision. Time-consuming computation of the algorithm proposed is greatly reduced compared with the algorithm to CFD model. Most of indoor personnels can feel comfortable by modifying distribution of outlets of air-conditioner and control variables. The experimental results indicated that the present choice of objective function and optimization approach are able to obviously improve the thermal comfort in indoor environment, and the energy cost is decreased accordingly.