基于中央空调的几种负荷预测方法对比分析

Comparison and Analysis of Several Load Forecasting Methods based on Central Air Conditioning

针对同一建筑,分别采用了多元线性回归,多元非线性回归,季节性指数平滑法,灰色预测以及神经网络方法等五种典型预测方法进行空调冷负荷预测研究。从预测精度,响应速度,建模的复杂程度,对输入数据的要求以及模型的适用性等五个方面对负荷预测的方法进行对比分析。结果表明:线性回归法精度普遍较低、灰色预测法不适用于非线性系统、神经网络预测法具有较高预测精度,而季节性指数平滑法在工程应用中具有较高价值。

Five typical forecasting methods,such as Multiple Linear Regression,Multiple Non-linear Regression,Exponential Weighted Moving Average,Gray Prediction,and Artificial Neural Network,were simulated for the same study conditions.The load forecasting methods from five aspects such as prediction accuracy,response speed,complexity of modeling,requirements for input data,and applicability of the model were compared and analyzed.The results show that the accuracy of Linear Regression method is generally low,Gray Prediction is not suitable for nonlinear systems,Artificial Neural Network has higher prediction accuracy,and Exponential Weighted Moving Average has higher value in engineering applications.