【摘要】 本文建立了提前1-4小时的空调冷负荷预测模型。使用相关性分析选择模型输入参数,使用ARX和ANN建立冷负荷预测模型,并对两种模型的预测精度做出对比。预测结果表明,当数据量大、变量维度小时,ANN模型无法发挥其优势,预测效果和ARX模型基本相同。考虑到ARX模型的简单性,选取ARX对未来时刻的冷负荷进行预测。提前1-4小时的ARX预测模型的CV-RMSE分别是11.6%、15.8%、18.9%和20.7%,各时间尺度下的预测精度都能够满足工程需求。更多还原

【Abstract】 In this paper, a cooling load forecasting model for air conditioning is established 1 to 4 hours in advance.Correlation analysis is used to select model input parameters, ARX and ANN are used to build the cooling load prediction model, and the prediction precision of the two models is compared. The prediction results show that when the data is large and the variable dimension is small, the ANN model can not exert its advantages, and the prediction effect is basically the same as the ARX model. Considering the simplicity of the ARX model, ARX is selected to predict the cooling load in the future.The CV-RMSE of the ARX prediction model 1-4 hours in advance was 11.6 %, 15.8 %, 18.9 % and 20.7 %, respectively. The prediction accuracy at each time scale can meet the engineering needs.更多还原