| 引用本文: | 樊红梅,刘晓民,王文娟, 等.基于多种方法的“以电折水”系数研究[J].灌溉排水学报,2021,(11):98-105. |
| FAN Hongmei,LIU Xiaomin,WANG Wenjuan, et al..基于多种方法的“以电折水”系数研究[J].灌溉排水学报,2021,(11):98-105. |
|
| 摘要: |
| 【目的】针对目前我国农业用水占比高、计量监测难度大等问题，引入“以电折水”方法计量农业用水，推进解决农业用水的计量和核算问题。【方法】以内蒙古通辽市科尔沁区为例，采用平均值预测、多元线性回归模型、BP神经网络模型3种方法对“以电折水”系数进行预测分析，并选用平均相对误差（MRE）、均方根误差（RMSE）、决定性系数（R2）对上述3种模型比较优选。【结果】“以电折水”系数平均值预测误差平均为7.40%，预测值与实测值拟合效果较差；多元线性回归模型预测误差平均为2.40%，预测值与实测值拟合效果较好；BP神经网络模型预测误差平均值为1.65%，预测值与实测值拟合效果最好。多元线性回归模型和BP神经网络模型的MRE、RMSE、R2分别为0.024、0.175、0.923和0.018、0.131、0.957。【结论】多元线性回归模型预测精度较好，BP神经网络模型预测精度最好，平均值预测模型预测精度最差。 |
| 关键词: “以电折水”系数；平均值预测；多元线性回归；BP神经网络 |
| DOI：10.13522/j.cnki.ggps.2021202 |
| 分类号: |
| 基金项目: |
|
| Estimating Water Consumption from Electricity Consumption: How to Calculate the Conversion Coefficient |
|
FAN Hongmei, LIU Xiaomin, WANG Wenjuan, et al.
|
|
1. School of Water Conservancy and Civil Engineering, Inner Mongolia Agricultural University, Hohhot 010018, China;
2. Inner Mongolia Jinhuayuan Environmental Resources Engineering Consulting Co., Ltd, Hohhot 010020, China
|
| Abstract: |
| 【Background】Agriculture is the largest water-consuming sector in China, but how to estimate agricultural water use is an issue that remains extremely difficult despite its importance in improving water measurement. Considering that moving water for irrigation needs power, an alternative method emerging over the past decades is to estimate it based on the electricity reading on the national grid. The purpose of this paper is to investigate how to estimate the conversion coefficient of this method.【Method】We took Horqin District at Tongliao City in Inner Mongolia as an example. We calculated the conversion coefficient using the multiple linear regression model and the BP neural network model respectively, with the standard average method taken as the control (CK). For each method, the average relative error (MRE), root mean square error (RMSE) and decisive coefficient (R2) were selected as indicators to compare the above models.【Result】The error of the standard average method was 7.40%. In contrast, the relative error of the multiple linear regression model and the BP neural network models was 2.40% and 1.65%, respectively. These results indicated that it is feasible to estimate agricultural water consumption using the electricity consumption read from the national grid. Among the three methods we compared, the BP neutral network was most accurate for estimating the conversion coefficient.【Conclusion】The MRE, RMSE and R2 of the multiple linear regression model were 0.024, 0.175 and 0.923 respectively, compared with their associated values in the BP neural network model, which were 0.018, 0.131 and 0.957. The BP neural network can thus be used as a robust method to estimate the conversion coefficient in converting electricity consumption to agricultural water consumption. |
| Key words: converting electricity to water; conversion coefficient; multiple linear regression; BP neural network |
