基于强化学习的中央空调系统温度控制与节能

Temperature Control and Energy Saving of HVAC System Based on Reinforcement Learning

针对中央空调系统非线性、大滞后的特点以及难以建立精确模型的问题,提出基于无模型强化学习的中央空调系统温度控制方法。针对Energyplus和Matlab通信问题,应用MLE+工具实现两者的联合仿真。将强化学习算法与基准启停策略和模型预测控制策略进行比较,方法能够在保证舒适度的前提下使得空调系统能耗最低。得出结论强化学习方法能取得较好的控制效果。

Aiming at the nonlinear and large hysteresis characteristics of HVAC system and the difficulty of establishing accurate models, a temperature control method for HVAC system based on model-free reinforcement learning is proposed. For the Energyplus and Matlab communication problems, the MLE+tool was used to implement the joint simulation of the two. The reinforcement learning algorithm was compared with the benchmark start-stop strategy and the model predictive control strategy, and the method can minimize the energy consumption of the air-conditioning system under the premise of ensuring comfort. It was concluded that the reinforcement learning method can achieve better control effects.