[讨论]关于预测能力的比较

hyacinth

不知道大家做没做过时间序列模型，回归模型和灰色系统模型这三种模型的预测能力的比较方面的工作？不知道这三种模型那种模型更好一些呢？

yejet

这个问题很难回答只能说各有各的优缺点和适用条件。

比如灰色理论所需的数据量小，这是它的优点，但同时也是它的缺点，对于对于数据序列较长的模型，灰色理论一般只取近期较少的数据，而舍去了很多的有用的信息。

感兴趣的话我再给大家转篇文章看一下

yejet

摘　要：
介绍了趋势分析法、回归分析法、指数平滑法、单耗法、灰色模型法、负荷密度法和弹性系数法等电力负荷预测的方法，并以预测珠海市全社会年用电量为实例，在适用条件、数据形式、计算难度和适用时间等方面对这几种预测方法进行了分析、比较。得出结论：回归分析法、趋势分析法适用于大样本，且过去、现在和未来发展模式均一致的预测，灰色模型法适用于贫信息条件下的预测；灰色系统理论采用生成数序列建模，回归分析法、趋势分析法采用原始数据建模，指数平滑法是通过对原始数据进行指数加权组合直接预测未来值的；回归分析法和趋势分析法的计算相对简单；单耗法、指数平滑法、灰色模型法较适宜近期预测，回归法、趋势分析法和改进型灰色模型较适于中、长期预测。?

　　关键词：规划；电力负荷预测；方法?

Analysis and comparison of several electric load forecasting methods?
WU Manhong, YANG Jiwang?
(Zhuhai Power Supply Branch, GPG, Zhuhai, Guangdong 519000, China)

　　Abstract: Some methods for electric load forecasting such as trend analysis, regression analysis, exponent smoothing, per unit consumption, greymodel, load density and elastic coefficient are introduced. With the load forecast of Zhuhai City as an example, the application condition, data format, calculation difficulty and suitable term of these methods are analyzed and compared. It is concluded that regression analysis and trend analysis methods apply to forecast based on large sample and a developing mode consistent in the past, at present and in the future, while grey model method applies to forecast under poor information condition. Regression analysis and trend analysis methods use raw data and grey model method uses generation data sequence for modeling; exponent smoothing method makes direct forecast using xponential weighting array of raw data. The calculation in regression analysis and trend analysis methods is easier. The methods of per unit consumption, exponent smoothing and grey model are fitfor short?term forecast, and the methods of regression analysis, trend analysis and improved grey model are suitable for medium and long term forecast.?

　　Keywords:planning; electric load forecast;method