考虑季节因素的产业用电量关联分析及预测

发布时间：2018-04-15 22:23

本文选题：产业电量 + 关联分析　；参考：《长沙理工大学》2014年硕士论文

【摘要】：用电量规律分析及预测是电网规划与建设的基础,也是电力需求侧管理的有效指南。随着近年来经济的高速发展,产业结构的复杂变化,使得电力需求内部结构发生着一定程度的变动。同时,用电量的季节波动性也导致部分地区电力需求仍存在季节性缺失。因此,需要在考虑季节因素的情况下,对不同类型的用电量进行分析与预测,以实现更加精细化的用电管理及制定更加经济化的购售电策略。本论文从产业和生活用电出发,做了如下研究：首先,从用电量自相关、互相关和气温影响因素三个方面对第一、第二、第三产业以及城、乡居民用电量进行相关性分析,即产业及生活用电量与自身历史数据的关联关系,各季节下五个用电量指标之间的内在关联关系和气温对其的外在影响关系。运用相关分析趋势图对以上三方面情况下各用电量趋势进行初步判断以后,计算其皮尔森相关系数,量化其相关密切程度,得到产业及生活用电量在以上三种情况下的对应关联关系。其次,根据得到各季节下三大产业以及城乡居民用电量相互之间的关联关系和气温对其的影响关系,构建各季节下VEC(Vector Error Correction)用电量预测模型。模型构建过程包括以下三部分：第一部分对各季节下五个用电量序列和气温序列进行平稳性检验；第二部分对于通过平稳性检验的用电量序列进行协整关系分析并对存在协整关系的各用电量建立协整方程；第三部分基于用电量之间的协整关系,构建各季节下VEC用电量预测模型。基于此模型进行了预测算例分析及对比,证明了其预测效果的准确性。最后,考虑运用从产业及生活用电量自相关关系和互相关关系两个角度分别构建的预测模型进行组合预测。即先基于产业及生活用电量指标自相关关联关系,根据各用电量平稳性检验和相关关系图构建各季节下ARIMA(Auto-Regressive Integrated Moving Average)用电量预测模型。根据ARIMA模型与VEC模型预测的相对误差,计算各月份下两模型的最优权重分配结果,得到各月的组合预测模型。对某省网2011年产业及生活月度用电量进行虚拟预测分析及对比,证明了该模型的准确性和可靠性。本文从内在关联和外在影响的角度对产业及生活用电量进行了相关性分析,并构建了基于关联关系的用电量预测模型。通过算例分析证明了所建立模型的准确性和有效性,对电力企业进行规划与调度、提高电网经济运行都具有一定的指导意义和参考价值。
[Abstract]:The analysis and prediction of power consumption law is the foundation of power network planning and construction, and it is also an effective guide for power demand side management.With the rapid development of economy and the complex changes of industrial structure in recent years, the internal structure of power demand has changed to a certain extent.At the same time, the seasonal fluctuation of electricity consumption also leads to the seasonal absence of electricity demand in some areas.Therefore, considering seasonal factors, different types of electricity consumption should be analyzed and forecasted, in order to achieve more refined management of electricity consumption and to formulate more economical power purchase and sale strategy.In this paper, the following research is done from industry and daily electricity consumption: firstly, from the three aspects of power consumption autocorrelation, cross-correlation and temperature influencing factors, the first, second, tertiary industry and urban and rural residents' electricity consumption are analyzed.That is, the relationship between industry and household electricity consumption and its own historical data, the internal relationship between five electricity consumption indexes in each season and the external influence of temperature on it.After using the correlation analysis trend map to judge the electricity consumption trends in the above three conditions, the Pearson correlation coefficient is calculated, and the correlation degree is quantified.Get the industry and living electricity consumption in the above three cases of the corresponding correlation.Secondly, according to the relationship between the three major industries and urban and rural residents' electricity consumption in different seasons and the influence of temperature on it, the VEC(Vector Error Correction (VEC(Vector Error Correction) electricity consumption prediction model in each season is constructed.The model construction process includes the following three parts: in the first part, five electricity consumption series and temperature series under each season are tested smoothly;In the second part, we analyze the cointegration relation of the electricity consumption series that pass the stationary test and establish the cointegration equation for the electricity consumption with cointegration relationship; the third part is based on the cointegration relationship between the electricity consumption.The prediction model of VEC power consumption in different seasons was constructed.Based on this model, the prediction results are analyzed and compared, and the accuracy of the prediction results is proved.Finally, the combined forecasting model which is constructed from two angles of industrial and household electricity consumption autocorrelation and cross-correlation is considered.Firstly, based on the autocorrelation relation of industry and household electricity consumption index, the forecast model of ARIMA(Auto-Regressive Integrated Moving average power consumption under different seasons is constructed according to the power consumption stationary test and correlation diagram.According to the relative error between the ARIMA model and the VEC model, the optimal weight distribution results of the two models in each month are calculated, and the combined prediction model for each month is obtained.The virtual prediction and comparison of the industry and monthly consumption of electricity in a province in 2011 proved the accuracy and reliability of the model.This paper analyzes the correlation between industry and household electricity consumption from the angle of internal relation and external influence, and constructs a model of electricity consumption prediction based on correlation relationship.The model is proved to be accurate and effective by example analysis. It has certain guiding significance and reference value for electric power enterprises to plan and dispatch, and to improve the economic operation of power network.
【学位授予单位】：长沙理工大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：F426.61

【参考文献】