基于灰色模型的电网负荷预测方法研究

发布时间：2018-04-26 11:34

本文选题：电网 + 负荷预测　；参考：《华北电力大学》2016年硕士论文

【摘要】：实际生活中有很多因素,例如政治、经济、重大事故等,会不同程度的影响电力系统中的电网中电压、电流等的稳定运行,这样会造成电网负荷的波动性,进而造成电网系统和电气设备的非正常运行以及其它损害等。因此,对电力系统中电网负荷的事先有效估计,是对电力系统进行合理经济调度、降低生产成本、防止电网大面积停电或者电网崩溃的迫切需求。电力系统中负荷预测是电力系统稳定运行的基础,是通过历史负荷的探究和分析,应用特定的分析方法来预测未来负荷。本文主要研究电力系统中电网负荷的较短期预测,并且针对如何建立预测精度更高和计算速度更快的预测模型进行了分析探索。本文所做的工作如下:1)本文基于电网中历史的负荷数据规律分析,采用基础灰色模型,即GM(1,1)灰色模型来预测地方电网中的负荷数值,在建模过程中提出动态新息模型建模。2)对预测结果精度检验的方法本文采用三种方法:相对误差检验法、后验差检验法和关联度检验法。3)为了提高GM(1,1)灰色模型对电网负荷的预测精度,本文提出三种方法来提高模型预测精度,第一种方法主要是改善优化GM(1,1)建模过程中的自身模型,即GM(1,1)灰色模型背景值的优化、GM(1,1)灰色模型灰色导数的优化、GM(1,1)灰色模型初始条件的合理选择等;第二种方法主要是应用融合GM(1,1)灰色模型的组合模型预测方法,将灰色预测模型分别与最小二乘法、指数平滑法、人工神经网路组合优化来预测电网负荷数据,通过弥补单一使用模型的不足的方式提高预测精度,期望达到良好的预测效果;第三种方法是基于RBF神经网络对改进GM(1,1)残差修正的负荷预测模型来对电网负荷进行分析和预测。通过最终对模型预测电网负荷的结果对比分析比较,可得出本文提出的三种提高模型预测精度的方法切实可靠,取得了很好的预测效果。
[Abstract]:There are many factors in real life, such as politics, economy, major accidents, which will affect the steady operation of voltage and current in power system to some extent, which will result in the fluctuation of power grid load. It also causes abnormal operation and other damage of power system and electrical equipment. Therefore, it is an urgent need to estimate the load of power system in advance, which is necessary for reasonable economic dispatching, reducing production cost and preventing large area power outages or power network collapse. Load forecasting in power system is the basis of stable operation of power system. Through the exploration and analysis of historical load, a specific analysis method is applied to forecast future load. This paper mainly studies the short-term forecasting of power network load, and analyzes and explores how to establish a forecasting model with higher forecasting accuracy and faster calculation speed. The work of this paper is as follows: (1) based on the analysis of the law of historical load data in the power network, this paper uses the basic grey model, that is, GM1 / 1) grey model, to predict the load value in the local power network. In the course of modeling, the dynamic innovation model modeling. 2) the method of testing the accuracy of prediction results is presented in this paper: the relative error test method, the relative error test method, the relative error test method, the relative error test method, In order to improve the forecasting accuracy of GM1 / 1) grey model, this paper presents three methods to improve the forecasting accuracy of the model. The first method is mainly to improve the self-model in the process of optimizing GM-1). That is, the optimization of the background value of the grey model, the optimization of the grey derivative of the grey model and the reasonable selection of the initial conditions of the grey model, the second method is mainly the application of the combined model prediction method of the combined grey model. The grey forecasting model is combined with the least square method, exponential smoothing method and artificial neural network to forecast the load data of the power network. The forecasting accuracy is improved by making up for the shortage of the single model, and the prediction effect is expected to be good. The third method is based on the modified RBF neural network (RBF) residual error correction load forecasting model to analyze and forecast the power grid load. Through the comparison and comparison of the results of the model forecasting power network load, it is concluded that the three methods proposed in this paper to improve the forecasting accuracy of the model are practical and reliable, and good prediction results have been obtained.
【学位授予单位】：华北电力大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TM715

【参考文献】