Neural network predictive control of thermal environment at the end of air conditioning system

In this paper, BP neural network is used to design a thermal environment predictive control scheme. Using BP neural network as controller, Regarding comfort and energy saving optimization as optimal object of the neural network. nonlinear system predictive control scheme is achieved by means of solving the optimal control sequence, and combined with the rolling optimization and feedback correction, The network can carry on the calibration error of the past time and forecast error of the future time, fully adapting to the model precision loss and interference factors, which make the control scheme has high stability and robustness. Meanwhile, The simulation experiment based on TensorFlow platform is designed. The experimental results show that the predictive control algorithm designed in this paper not only meets the dual requirements of energy saving and comfort, but also can effectively suppress interference. it has the significant advantages of short response time, small overshoot and no oscillation.