#### 期刊菜单

Research and Short-Term Prediction of the Total Social Electricity Consumption in Beijing
DOI: 10.12677/MOS.2022.115129, PDF, HTML, XML, 下载: 68  浏览: 223  科研立项经费支持

Abstract: According to the whole social electricity consumption data in Beijing from 1990 to 2020, the changing trend of electricity consumption in Beijing and the proportion of electricity consumption in various industries are analyzed, and it turns out that the electricity consumption of the tertiary industry determines the electricity consumption in Beijing to a certain extent. Then, the outliers and ADF stability were tested on the electricity consumption sequence of the whole society in Beijing, and the ARIMA model was constructed. The fitting effect of the model was excellent so as to predict the electricity consumption value of the whole society in Beijing in the next 5 years. Compared with the preliminary statistics of Beijing’s electricity consumption in 2021, it shows that the prediction accuracy is high. In the face of the forecast trend of the rising electricity consumption of the whole society in Beijing, relevant suggestions are given from the three aspects of increasing the installed capacity, optimizing the power generation structure and ensuring the transportation.

1. 引言

2. 数据获取与分析

$全社会用电量\left(\text{W}\right)=第一产业+第二产业+第三产业+居民生活用电$ (1)

2.1. 数据来源

2.2. 北京全社会用电量现状

2.2.1. 北京全社会用电量

Figure 1. Line chart of total electricity consumption in Beijing (1990~2020)

2.2.2. 各产业用电量情况

Figure 2. Percentage of electricity consumption in Beijing (1990~2020)

3. 模型构建与检验

AR(I)MA模型是统计学中研究时间序列数据的重要模型，是对自回归模型(AR)和移动平均模型(MA)的有机结合，常用于长期数据的追踪分析。该模型简单易懂，只需要用内生变量就可对时间序列数据进行预测。参考杨艳等 [4] 人在研究陕西农村就业人数的方法，按流程先后顺序，主要分成数据处理与序列检验、模型方程构建、模型检验、预测等四个部分(如图3)：

3.1. 数据处理与序列检验

3.1.1. 数据异常值检验

3.1.2. 序列平稳性检验

Figure 3. Flow chart of model construction and inspection ideas

Figure 4. Box type diagram of the electricity consumption sequence of the whole society in Beijing

Table 1. Original sequence ADF stationarity test

Table 2. The ADF stationarity test of the first-order difference sequence

p值 < 0.05，结合Eviews软件绘制的相关图来看，自相关有明显的衰减趋势，虽后期有上升趋势，但均在可控范围内，因此一阶差分后的序列具有平稳性。如图5，用一阶差分序列建立模型，即差分整合移动平均自回归模型(ARIMA)。

Figure 5. First-order difference sequence autocorrelation and partial autocorrelation plots

3.2. 模型构建

ARIMA模型即差分整合移动平均自回归模型，共有p、d、q三个阶数，p表示自回归项数，d代表序列的差分阶数，q表示滑动平均项数。例如当 $p=n,\text{\hspace{0.17em}}d=1,\text{\hspace{0.17em}}q=m$ 时，模型中共有 $n+m+1$ 个参数，ARIMA(p,l,m)模型的表达式如下：

$\Delta {y}_{t}=\underset{i=1}{\overset{n}{\sum }}{\alpha }_{i}\Delta {y}_{t-i}+\underset{j=1}{\overset{m}{\sum }}{\beta }_{j}{\epsilon }_{t-j}+{\epsilon }_{t}+c$ (2)

3.2.1. 阶数确定

Table 3. AIC values under different models of different orders

3.2.2. 方程确定

ARIMA(1,1,0)模型中有 $c,\text{\hspace{0.17em}}{\alpha }_{1}$ 两个参数，利用SPSSAU软件求解方程参数，结果如下：

Table 4. ARIMA (1,1,0) parameters table

$\Delta {y}_{t}=349200.484-0.157\Delta {y}_{t-1}+{\epsilon }_{t}$ (3)

3.3. 模型检验

3.3.1 数值对比

3.3.2. 检验残差序列

Figure 6. Comparison of the fitted values and the real values

Table 5. Residue sequence statistics Q

Q6项即残差前6阶自相关系数的p值 ? 0.10 > 0.05，说明在95%显著水平下不能拒绝原假设，即残差为白噪声序列，方程拟合效果较好。

4. 北京市全社会用电量预测

Table 6. Forecast of total electricity consumption in Beijing (2021~2025)

5. 结论及建议

**.在0.01级别(双尾)，相关性显著。

**.在0.01级别(双尾)，相关性显著。

Table A1. Growth rate of electricity consumption in China and various industries in Beijing (1991-2020)

Table A2. The portion of electricity consumption (1990-2020)

Table A3. Beijing's GDP and added value of various industries (1990-2020) Unit: 100 million yuan

Figure 1. Line chart of electricity consumption of primary industry and added value of primary industry

Table A4. Results of the Fourth, Fifth, Sixth and seventh Beijing population census (excerpt)

 [1] 孙颖. 基于季节性ARIMA模型的全社会用电量预测研究[J]. 黄冈师范学院学报, 2017, 37(6): 52-56. [2] 郭松亮, 闫鹏君, 鄂浩坤. 基于ARIMA模型的北京市全社会用电量短期预测[J]. 北京信息科技大学学报(自然科学版), 2020, 35(5): 93-96. https://doi.org/10.16508/j.cnki.11-5866/n.2020.05.017 [3] 贺慧宇. 宜居城市: 北京规划新概念[N]. 中国建设报, 2004-11-18. [4] 杨艳, 雷咪咪. 农村三产融合背景下基于ARIMA模型对陕西省农村就业人数的分析与预测[J]. 山西农经, 2022(8): 9-12. https://doi.org/10.16675/j.cnki.cn14-1065/f.2022.08.003 [5] 黄杰俊, 赵楚波. 基于ARIMA模型预测黄山空气质量指数[J]. 科技视界, 2021(23): 127-129. https://doi.org/10.19694/j.cnki.issn2095-2457.2021.23.50 [6] 曹政. 北京“十四五”时期能源发展规划发布[N]. 北京日报, 2022-04-04(001). https://doi.org/10.28033/n.cnki.nbjrb.2022.001922