建筑空调系统最优停机时间预测与控制

Predicting the Optimal Shut-Down Time of Heat,Ventilation,Air-Conditioning and Cooling System in Building

针对建筑空调系统在停机阶段联合优化控制自然通风和室内空调系统,以在满足室内人员舒适的同时尽量减小建筑总能耗的问题,研究了自然通风量、风机盘管和新风送风量为控制输入,室内环境,包括空气温度、空气含湿量为系统状态,建筑总能耗为目标,室内人员舒适度为约束,且控制变量和状态变量受约束的最优控制问题。研究中通过采用神经网络来逼近和预测天气、室内初始环境参数与空调最优停机时间、最优联合控制策略之间的非线性映射关系,找到了一条简单可行、性能较好的非最优策略,极大地提高了联合控制策略的普适性和推广能力;经神经网络结构参数的优化配置后,建筑空调系统停机时间预测和控制的可靠性和精准性有所提高。室内环境状态和建筑能耗的数值比较表明:所提策略在有效降低问题复杂度和计算成本的同时,仿真时间在1h的尺度上可进一步节能约20%。

The problem of optimal joint control of heating,ventilation,air-conditioning and cooling （HVAC） as well as natural ventilation during the HAVC shut-down stage is studied for energy saving while maintaining indoor thermal comfort.The problem is formulated as a constrained optimization problem by taking the volume of natural ventilation and the statuses of the supplied air of both FCU and FAU as system inputs,the indoor air temperature and the indoor air moisture content as system states,the total building energy consumption as the optimization objective,and the comfort requirements of indoor personnel as constraints.Results show that a good-enough policy instead of the best policy can be obtained by using ANN to approximate and to predict the relationship among the climate parameters,the initial conditions of indoor environment,the optimal shut-down time of HVAC,and the joint control policy of natural ventilation and HVAC.The policy has higher adaptability and generalization ability,and the predicting accuracy and reliability of ANN can be further improved by optimizing its configuration.Comparisons of both the indoor environment and the building energy consumption show that the good-enough policy further achieves 20 ％ reduction in building energy consumption in one hour simulation.