计算机工程与应用
計算機工程與應用
계산궤공정여응용
Computer Engineering and Applications
2015年
21期
213-218
,共6页
造价估算%大数据%神经网络%深度学习%电力工程
造價估算%大數據%神經網絡%深度學習%電力工程
조개고산%대수거%신경망락%심도학습%전력공정
cost estimation%big data%neural network%deep learning%power engineering
为了实现对电力工程造价高效、精确的估算，提出了一种电力工程造价的随机权深度神经学习估算算法（Random Weighted Deep Neural Learning，RWDNL）。通过构建外权随机的带有小中间层的多隐层神经网络模型，利用神经网络深度学习实现了对海量数据有效特征的提取以及电力工程项目造价估算。数值仿真实验结果表明该方法使工程造价估算精度和速度大大提高，可获得令人满意的泛化能力。
為瞭實現對電力工程造價高效、精確的估算，提齣瞭一種電力工程造價的隨機權深度神經學習估算算法（Random Weighted Deep Neural Learning，RWDNL）。通過構建外權隨機的帶有小中間層的多隱層神經網絡模型，利用神經網絡深度學習實現瞭對海量數據有效特徵的提取以及電力工程項目造價估算。數值倣真實驗結果錶明該方法使工程造價估算精度和速度大大提高，可穫得令人滿意的汎化能力。
위료실현대전력공정조개고효、정학적고산，제출료일충전력공정조개적수궤권심도신경학습고산산법（Random Weighted Deep Neural Learning，RWDNL）。통과구건외권수궤적대유소중간층적다은층신경망락모형，이용신경망락심도학습실현료대해량수거유효특정적제취이급전력공정항목조개고산。수치방진실험결과표명해방법사공정조개고산정도화속도대대제고，가획득령인만의적범화능력。
In order to realize an efficient and accurate cost estimation of power engineering project, a cost estimation method for power engineering project with big data is proposed based on a so-called Random Weighted Deep Neural Learning(RWDNL)algorithm. By means of building a multi-layer random weighted neural network with a small central layer, effective features are extracted from mass engineering data and power engineering project cost estimation is also real-ized by neural network deep learning. The experimental results demonstrate the outstanding performances of the proposed RWDNL method on estimation time consume and estimation accuracy, as well as its satisfactory generalization ability.
