基于动态反馈的多元参数回归空调负荷预测控制方法

A Predictive Control Method of Air-conditioning Load Based on Dynamic Feedback and Multiple Parameters Regression

针对目前常用的空调负荷预测算法中精度与实用性的问题,提出多重负荷预测的方法:负荷的趋势预测模型和精确预测模型。趋势预测模型用于预测24 h内各时刻的负荷,建立基于气象、历史和时间参数长期和短期多元参数回归模型,并引入预测控制方法中反馈校正和滚动优化的方法,并对误差采用一次平滑法。采用遍历搜索法,寻找最优误差,反馈给模型进行修正。每滚动一次,舍去旧值,引入新值,并重新寻优一次误差,直至完成预测。精确预测用于下一个时刻的负荷,利用相似日的历史负荷建立二阶ARX模型,对气象负荷进行第一次修正,再利用前一日的负荷建立一阶ARX模型,对预测负荷进行第二次修正,利用滚动优化进行动态反馈修正。利用某小区的实际数据进行测试,预测结果满足精度要求。

Aiming at the problems of accuracy and practicality in current commonly used air conditioning load forecasting algorithms,multi-load forecasting methods:load trend forecasting model and accurate forecasting model were put forward.Trend forecasting model wasused to forecast the load in24hours.Long-term and short-term multi-parameter regression models based on meteorological,historicaland time parameters were established.Feedback correction and rolling optimization methods were introduced in predictive controlmethod,and the first-order smoothing method was used to estimate the error.The ergodic search method was used to find the optimal errorand feedback to the model for correction.Once scrolling each time,the old value was recanaged,new values were introduced,and thefirst error was researched until the prediction was completed.Accurate forecasting was used for the next time load.The second-order ARXmodel was established by using similar day's historical load.The first-order ARX model was established by using the previous day's load.The second revision was made to the forecasted load,and the dynamic feedback revision was made by rolling optimization.The actual dataof a certain district were tested,and the prediction results met the accuracy requirements.