Real-time dynamic estimation of occupancy load and an air-conditioning predictive control method based on image information fusion

Abstract Public buildings have large indoor personnel flow changes and complex backgrounds. Due to the large lag of air-conditioning systems, such systems in public buildings have difficulty adjusting to changes in indoor loads, resulting in untimely system control, poor thermal control of buildings' internal environments, and energy waste. This paper proposes a real-time estimation method of building space personnel load and an air-conditioning predictive control strategy by integrating image information. First, deep learning image detection technology is applied to establish an end-to-end building space personnel load dynamic estimation model based on a convolutional neural network. The model is employed to realize the real-time detection of the number of indoor personnel and estimate changes personnel load. Second, to propose air-conditioning prediction control strategies introduce personnel occupancy load control factors, predict the indoor temperature change trend caused by load changes, provide compensation for the control system adjustment amount, improve the indoor environment quality problems caused by large lag under conventional control methods, adjust the cooling energy supply of air conditioning systems, and reduce energy consumption. Finally, the simulation experiment involving a student activity center and a small office building is carried out and the results are analyzed. Simulation results show that: the predictive control strategy proposed in this paper can better maintain environmental stability in a building, can respond more rapidly and has greater energy-saving potential.