城市燃气负荷的短期预测

A short-term prediction model for urban gas load:A case study of Dalian

城市燃气负荷的短期预测对保证城市供气的相对稳定尤为重要。为此,通过对各类气象因子与燃气日负荷间的相关性分析,以有效温度为主导因素,结合大连地区温度分布规律,推导出燃气日负荷预测模型与月负荷预测模型间的关系,建立了不同预测步长下基于双曲正弦函数的燃气负荷短期预测模型,计算出日负荷预测模型在最不利工况下(传统节日集中的月份)的平均绝对百分比误差(MAPE)和均方百分比误差(RMSPE)值分别为3.67%和5.03%,月负荷预测模型的MAPE和RMSPE值分别为1.02%和1.32%,均小于10%,日负荷预测模型与月负荷预测模型的预测能力均达到较高精确度。利用该转换关系不但能够实现预测模型预测步长的改变,而且所获得的模型预测效果也较理想。对于不同的城市,只要根据该地区气象及燃气负荷特点,选择合适的模型参数值,利用该预测方法即可实现不同预测步长下燃气负荷的短期预测。

With the rapid development of the gas industry,a higher demand is required for the city gas supply system.To keep the city's relatively stable gas supply and to protect the economic interests of the gas industry,the short-term prediction of city gas load is particularly important.Taking the city of Dalian as a case of study,we made the correlation analysis of different meteorologic factors with gas load,in combination with the outside temperature distribution in this city,to derive the relationship between daily and monthly gas load prediction models.Then we built a short-term prediction model based on the hyperbolic sine function with the effective temperature as the dominant factor in different prediction steps.Finally,we used this model to achieve the forecast results of the daily gas load in February (with many traditional holidays) and March and the monthly gas load in 2006 in this city.In light of the accuracy evaluation method of prediction models suggested by the literature[12],the mean absolute percentage error (MAPE) and the root mean absolute percentage error (RMSPE) of the daily gas load prediction model in February of 2006 are 3.67% and 5.03% respectively and the MAPE and RMSPE of the monthly gas load prediction model in 2006 are 1.02% and 1.32% respectively,the values of which are all less than 10%.This proves that both daily and monthly gas load prediction models reach a highly accuracy in prediction.As a result,their relationship could be applied to forecast the change of the predictive step and the achieved results are satisfactory.Above all,this proposed method will be useful in the short-term prediction of gas load for different cities,by choosing suitable model parameters according to the weather and the gas load characteristics of the city.