摘要:為了減少煤炭過(guò)度開(kāi)采造成的浪費(fèi)問(wèn)題和傳統(tǒng)的煤炭需求預(yù)測(cè)方法無(wú)法處理好小樣本、局部極小點(diǎn)的問(wèn)題,文章提出了一種支持向量機(jī)( SVM) 預(yù)測(cè)模型。此種方法的合適是把煤炭預(yù)測(cè)需求的問(wèn)題轉(zhuǎn)化成了序列預(yù)測(cè)的問(wèn)題。文章論述了支持向量機(jī)的基本原理,同時(shí)還構(gòu)造了多輸入、單輸出的SVM預(yù)測(cè)模型。測(cè)試結(jié)果表明,和傳統(tǒng)的煤炭需求預(yù)測(cè)方法相比優(yōu)越性非常明顯,具有很好的應(yīng)用空間和市場(chǎng)價(jià)值。
關(guān)鍵詞:支持向量機(jī);預(yù)測(cè);煤炭需求量;序列預(yù)測(cè)
中圖分類號(hào):TP311 文獻(xiàn)標(biāo)識(shí)碼:A 文章編號(hào):1009-3044(2014)20-4906-03
Coal Demand Prediction Based on a Support Vector Machine Model
XU Xiao-bing1,SHI Xing-yan2
(1.Zhengzhou Railway Vocational and Technical College, Zhengzhou 450052, China;2.Henan Vocational College of Agriculture, Zhengzhou 451450,China)
Abstract: In order to reduce the waste caused by excessive exploitation of coal issues and traditional coal demand forecasting method can not handle the small sample local minima problem, the article presents a support vector machine (SVM) forecasting model. This method is suitable to forecast demand of coal is transformed into series forecasting. This article discusses the basic principles of support vector machines, and also construct a multi-input, single-output SVM prediction model. Test results show that the traditional methods of coal demand forecast is very obvious advantages, with good use of space and market value.
Key words: support vector machine; prediction;coal demand; Series Prediction
國(guó)內(nèi)能源消費(fèi)的近四分之三是煤炭,所以說(shuō)我國(guó)最重要的基礎(chǔ)能源就是煤炭。煤炭產(chǎn)業(yè)在我國(guó)經(jīng)濟(jì)與社會(huì)發(fā)展中扮演著非常重要的角色。但是經(jīng)常會(huì)出現(xiàn)過(guò)度開(kāi)采,導(dǎo)致煤炭過(guò)剩的現(xiàn)象,這就造成了煤炭資源的浪費(fèi)[1]。為了避免這種情況的出現(xiàn),同時(shí)保證經(jīng)濟(jì)發(fā)展的需求,有必要對(duì)煤炭的需求進(jìn)行科學(xué)正確的預(yù)測(cè)。目前預(yù)測(cè)的方法很多,常用的有神經(jīng)網(wǎng)絡(luò)、灰色系統(tǒng)模型預(yù)測(cè)法。但是這些預(yù)測(cè)方法都存在一些問(wèn)題,比如神經(jīng)網(wǎng)絡(luò)預(yù)測(cè)法存在局部極小值問(wèn)題,灰色系統(tǒng)模型預(yù)測(cè)法無(wú)法很好的處理好小樣本、局部極小點(diǎn)的問(wèn)題[2]。
為了解決傳統(tǒng)煤炭需求預(yù)測(cè)方法存在的問(wèn)題,文章提出了支持向量機(jī)的方法,此方法是基于系統(tǒng)學(xué)的VC維( vapnik-chervonenks dimension)理論的一種機(jī)器學(xué)習(xí)方法。其最基本的原則是結(jié)構(gòu)化風(fēng)險(xiǎn)最小化。支持向量機(jī)核心思想是通過(guò)求解線性約束的二次規(guī)劃問(wèn)題,從而能很好地得到全局最優(yōu)解。……