【作者】 李国辉；

【导师】 朱建良；

【作者基本信息】 哈尔滨理工大学 ， 计算机应用技术， 2005， 硕士

【摘要】 本文介绍了电力负荷预测的原理、起源以及当前国内外电力负荷预测领域的概况,并概述了目前常用的几种负荷预测方法,对新近引入电力系统的灰色预测理论及其相应的各种模型的基本原理做了较为详细的探讨。本文在对传统的灰色预测模型分析的基础上,经过理论推导,提出了微偏灰色预测模型;并结合最优维数灰色预测模型,又提出了最优维数微偏灰色预测模型,经实例预测表明,本文提出的两种新的灰色预测模型在预测精确度上都较传统的灰色预测模型有某种程度的提高。本文提出了基于灰色理论与BP 神经网络的电力负荷预测方法。该方法首先用不同灰色预测模型进行预测,然后从中选出最优值作为训练样本对神经网络进行训练,最后利用训练好的神经网络进行预测。本预测模型由于采用了灰色理论与BP 神经网络相结合的方法,吸取了二者的优点,避免了单一预测模型所存在的预测风险。从对哈尔滨市市郊农电局所属电网2003年各月及全年的电量负荷预测实例中可以看到,其预测精确度和稳定性与上述其它灰色预测模型比较都有明显的改进。本文提出的三种预测方法得到了哈尔滨市市郊农电局有关领导和生产营业部门的认同,对电力企业进行计划拟定和生产运行均具有较高的指导和应用意义。更多还原

【Abstract】 This article introduces the principle、origins and electric power load forecasting and forecasting abstract of domestic and trend predict the realm of electric power load. Introduces current in common use and a few estimative method, to recently lead into the gray estimative theories of the electric power system and the basic principle of the one of its models does the detailed research. This article has proposed leaning towards and predicting grey models and optimum dimension predicting grey models on the basis of the analysis of the traditional predicting models. On the basis of the thing that simply predict the model little greyly, combining and using the optimum dimension predicting grey models, the author has proposed again that the optimum dimension leaning towards the grey models of predicting a little. Predication showed by the instance , there is some improvement in predicting the accuracy comparing with the traditional grey prediction models in two kinds of new prediction models which this article has put forward. On the basis of the thing that the theory has analyzed the grey theory and neural network theory, This article offer the method after putting forward the electric power load based on grey theory and BP neural network. This method uses different grey models to predict the selected neural network separately at first, and chooses the optimum value as the training samples to train the neural network, then utilizes and trains the good neural network to predict. Originally the predicative model adopts the method that the grey theory combining with BP artificial neural network, have drawn the advantage of the two, the prediction risk of avoiding the single predicative model. The instance of prediction from electric consumption load of one month and the whole year of 2003 in rural and suburbs of Harbin can see that the accuracy and the stability have a great improvement comparing to other predicative models. The three means of prediction brought forward in the article have obtained certification by some leading cadres concerned and business departments which have a great instruction and applied meaning to electric enterprises to process working out plans as well as the production revolved.更多还原