本文關(guān)鍵詞：基于小波變換和ARMA-LSSVM的忙時話務(wù)量預(yù)測　出處：《計算機工程與設(shè)計》2014年12期 　論文類型：期刊論文

  更多相關(guān)文章： 話務(wù)量 小波變換 自回歸滑動平均模型 最小二乘支持向量機 組合預(yù)測

【摘要】：為提高受多種因素影響的話務(wù)量數(shù)據(jù)的預(yù)測精度和穩(wěn)定性,提出一種考慮多因素影響的基于小波變換和自回歸滑動平均(ARMA)-最小二乘支持向量機(LSSVM)的話務(wù)量組合預(yù)測模型。對忙時話務(wù)量數(shù)據(jù)進行相關(guān)性分析,得出影響話務(wù)量的重要因子;利用小波變換對數(shù)據(jù)進行分解和重構(gòu),得到低頻分量和高頻分量;將低頻分量輸入ARMA模型進行預(yù)測,將高頻分量和話務(wù)量重要影響因子輸入粒子群算法優(yōu)化的LSSVM模型進行預(yù)測,將兩組預(yù)測結(jié)果合成。實驗結(jié)果表明,該模型進一步提高了預(yù)測精度和穩(wěn)定性。
[Abstract]:In order to improve the prediction accuracy and stability of traffic data affected by many factors. A new method based on wavelet transform and autoregressive moving average (ARMA-LS-LSSVM) based on wavelet transform and least squares support vector machine (LSSVM) is proposed. The correlation analysis of busy time traffic data is carried out. The important factors affecting traffic are obtained. The wavelet transform is used to decompose and reconstruct the data to obtain the low-frequency and high-frequency components. The low frequency components are input into the ARMA model for prediction, and the high frequency components and traffic important factors are input to the LSSVM model optimized by particle swarm optimization. The experimental results show that the prediction accuracy and stability of the model are further improved.
【作者單位】： 新疆大學(xué)信息科學(xué)與工程學(xué)院;中國移動通信集團新疆有限公司;
【基金】：中國移動通信集團新疆有限公司研究發(fā)展基金項目(XJM2013-2788)
【分類號】：TP393.06
【正文快照】： 0引言準確的話務(wù)量預(yù)測,能夠為網(wǎng)絡(luò)管理、規(guī)劃與設(shè)計提供重要依據(jù),為網(wǎng)絡(luò)擁塞、覆蓋和干擾等提供決策支持,因此提高話務(wù)量預(yù)測的精度是當(dāng)前研究的熱點。目前預(yù)測的模型主要有差分自回歸移動平均(ARI-MA)模型[1,2]、灰色模型[3,4]、神經(jīng)網(wǎng)絡(luò)預(yù)測模型[5,6]、馬爾科夫預(yù)測模型[7,，

本文編號：1415886