基于Elman神经网络的蓄能空调需求响应策略研究  

Research on Demand Response Strategy of Thermal Energy Storage Air-conditioning System Based on Elman Neural Network

作　　者：奚源 孟庆龙[1] 任效效 刘佳莉 XI Yuan;MENG Qinglong;REN Xiaoxiao;LIU Jiali(Chang’an University,Xi’an 710061,Shaanxi,China;Xi’an Jiaotong University,Xi’an 710047,Shaanxi,China)

机构地区：[1]长安大学,陕西西安710061 [2]西安交通大学,陕西西安710049

出　　处：《建筑科学》2022年第4期190-197,204,共9页Building Science

基　　金：陕西省重点研发计划项目(2020NY-204);山东省可再生能源建筑应用技术重点实验室开放课题(JDZDS02)。

摘　　要：需求响应是实现电网削峰填谷,缓解电网供需平衡的新手段。蓄能空调是需求响应中的有效弹性资源,可极大提升建筑用能灵活性。以西安市某蓄能空调实验平台为对象,开发预测模型,并进一步实施需求响应策略。为获取历史负荷数据,搭建TRNSYS仿真模型,基于负荷预测精度最高的Elman神经网络算法,建立蓄能罐的储、释能时长预测模型。结果显示,Elman负荷预测模型拟合度R;可达到0.91,储能、释能时长预测模型的R;值分别为0.97、0.94。此外,与传统4种需求策略对比,基于Elman神经网络的需求响应策略在需求响应时段内的耗电量最少,且全天运行费用最低。Demand response is a new means to achieve grid peak-shaving and valley-filling and realize the balance of grid supply and demand. Thermal energy storage(TES) can be utilized as an effective component in demand response(DR), which can significantly increase building energy ?exibility. This study developed an Elman neural network(ENN) prediction model for a TES air condition system in Xi’an. Based on this prediction model, a control strategy for DR was determined. To obtain the historical data, a TRNSYS simulation model was established. Based on the Elman neural network algorithm with the highest load forecast accuracy, a prediction model for storage-release time of the energy storage tank was established. Compared with experimental data, the R~2 value of load forecasting reached 0.91;for storage-release time of the energy storage tank, it reached 0.97 and 0.94 respectively. The test results indicate that the proposed demand response strategy consumed the least power during the demand response period and had the lowest operating cost throughout the day compared with the other four traditional demand response strategies.

关 键 词：空调需求响应 TRNSYS ELMAN神经网络 主动储能 

分 类 号：TU831.3[建筑科学—供热、供燃气、通风及空调工程]