Thermal comfort is an evaluation index of indoor environment comfort. Since the calculation of thermal comfort is a complex nonlinear iterative process, it is inconvenient to apply to air conditioning real-time control system. In order to solve this problem, use the BP neural network algorithm to predict thermal comfort. However, in order to improve the slow convergence rate of traditional BP neural network, the bird swarm algorithm (BSA) is used to optimize the initial weights and thresholds of BP neural network. Finally, the BSA algorithm is compared with the similar particle swarm optimization (PSO) algorithm. MATLAB software is used to simulate, and the simulation results of BSA-BP prediction model are compared with the simulation results of the basic BP neural network prediction model and the PSO-BP prediction model. The results show that the BSA-BP algorithm has faster convergence speed and higher prediction accuracy.