【摘要】：傳統(tǒng)計(jì)算方法難以準(zhǔn)確地描述邊坡穩(wěn)定問(wèn)題高度復(fù)雜的非線性特征,為此提出基于和聲搜索(HS)與相關(guān)向量機(jī)(RVM)的邊坡穩(wěn)定性分析模型。由于邊坡安全系數(shù)預(yù)測(cè)存在小樣本、非線性、高維數(shù)等特征,采用RVM可以快速得出準(zhǔn)確率更高的預(yù)測(cè)值,以及預(yù)測(cè)值的后驗(yàn)概率分布,并建立置信區(qū)間。由于RVM預(yù)測(cè)效果取決于核函數(shù)類型及參數(shù)選取的合理性,引入了泛化能力、插值能力均較強(qiáng)的混合核函數(shù);并利用拉丁超立方抽樣(LHS)初始化和聲記憶庫(kù);后采用HS算法對(duì)混合核函數(shù)的參數(shù)組合進(jìn)行全局尋優(yōu)。通過(guò)編寫Matlab程序,以預(yù)測(cè)值的平均絕對(duì)誤差(MAE)最小化為尋優(yōu)目標(biāo),建立基于HS-RVM的邊坡穩(wěn)定性分析模型。應(yīng)用實(shí)例證明:HS-RVM充分挖掘了兩算法的計(jì)算能力;模型不僅精度高,而且計(jì)算速度快,簡(jiǎn)單易用;在實(shí)際工程計(jì)算中有良好的應(yīng)用前景。
[Abstract]:It is difficult to accurately describe the highly complex nonlinear characteristics of slope stability problems by traditional calculation methods. Therefore, a slope stability analysis model based on Harmony Search (HS) and Correlation Vector Machine (RVM) is proposed. Because the prediction effect of RVM depends on the reasonableness of kernel function type and parameter selection, a hybrid kernel function with strong generalization ability and interpolation ability is introduced; Latin hypercube sampling (LHS) is used to initialize the harmony memory library; and HS algorithm is used to initialize the hybrid kernel function. A slope stability analysis model based on HS-RVM is established by compiling a MATLAB program with the objective of minimizing the mean absolute error (MAE) of the predicted value. Engineering calculation has a good application prospect.
【作者單位】： 西安理工大學(xué)水利水電學(xué)院;西安理工大學(xué)西北旱區(qū)生態(tài)水利工程國(guó)家重點(diǎn)實(shí)驗(yàn)室培育基地;
【基金】：國(guó)家自然科學(xué)基金資助項(xiàng)目(51409205) 西安理工大學(xué)西北旱區(qū)生態(tài)水利工程國(guó)家重點(diǎn)實(shí)驗(yàn)室開放基金項(xiàng)目(106 221210) 水文水資源與水利工程科學(xué)國(guó)家重點(diǎn)實(shí)驗(yàn)室開放研究基金(2014491011)~~
【分類號(hào)】：TU43
，

本文編號(hào)：2221873