摘要
研究了变风量空调系统神经网络预测优化控制方法,优化指标考虑了舒适性和耗能量,舒适性指标取PMV指标,耗能量包括风机和冷水泵能耗。系统的控制量为送风风速和冷水流量,被控参数为空调区域的温湿度,采用预测滚动优化控制算法训练多层前向神经网络,然后将其作为优化反馈控制器来求解变风量暖通空调系统的优化解,并在运行中实时预测空调区域的负荷。仿真结果表明,采用此方法,在模型环境、负荷参数变化的情况下,既可以达到节能的要求,又可以使空调区域的温湿度保持在舒适范围内。
Studies the optimizing control method of artificial neural network prediction. The optimizing indices include comfort index and energy consumption index, the former adopting PMV, and the latter being energy consumption of fans and chilled water pumps. Controlling targets are the supply air velocity and the chilled water flow rate, and the controlled parameters are temperature and humidity in air conditioned zones. Adopting predictive rolling optimization algorithm to train a multi-layer forward neural network, the network acts as an optimizing feedback controller to obtain the optimal solution of VAV air conditioning systems, and the network predicts real-time load of air conditioned zones in operation. Simulation results show that adopting such method can save energy and maintain temperature and humidity in air conditioned zones within comfortable scope under the condition of variable model environment and load parameters.
出处
《暖通空调》
北大核心
2005年第4期112-116,59,共6页
Heating Ventilating & Air Conditioning
