本文選題：滑坡概率 + 邏輯回歸模型�。� 參考：《江西理工大學(xué)》2017年碩士論文

【摘要】：本文以贛南離子型稀土礦區(qū)邊坡為例,研究滑坡概率在邊坡穩(wěn)定性分析中的應(yīng)用。論文的主要研究工作及結(jié)論如下:(1)滑坡概率模型及應(yīng)用。研究根據(jù)贛南離子型稀土礦區(qū)邊坡的實(shí)際概況,選定滑坡關(guān)鍵影響因素,通過邊坡現(xiàn)場(chǎng)調(diào)查、巖土參數(shù)測(cè)定、數(shù)據(jù)監(jiān)測(cè)等3種方式獲取了21個(gè)邊坡樣本,采用邏輯回歸模型結(jié)合確定性系數(shù)CF建立滑坡概率模型,由滑坡概率模型和邊坡樣本進(jìn)行實(shí)例應(yīng)用。選用龍南縣東江鄉(xiāng)足洞試驗(yàn)礦某邊坡作為工程實(shí)例,進(jìn)行滑坡概率求解,包括確定子集區(qū)間分類及各子集區(qū)間確定性系數(shù)CF、邊坡樣本各參數(shù)轉(zhuǎn)化為確定性系數(shù)CF值、應(yīng)用SPSS軟件進(jìn)行邏輯回歸等,得出滑坡概率求解方程,繼而對(duì)目標(biāo)邊坡滑坡概率進(jìn)行預(yù)測(cè),結(jié)果表明滑坡概率適用于邊坡穩(wěn)定性分析。(2)滑坡概率模型優(yōu)化。滑坡概率分析邊坡穩(wěn)定性其精度和準(zhǔn)確性存在瑕疵,因此對(duì)其進(jìn)行了3種優(yōu)化,包括滑坡影響因素子集區(qū)間選取、邊坡狀態(tài)取值以及降低滑坡影響因素間的相關(guān)性。研究選取了兩種不同的子集區(qū)間分類方法對(duì)子集區(qū)間選取進(jìn)行優(yōu)化,結(jié)果表明采用I類方法即步長(zhǎng)分類法較為適用。比較不同邊坡狀態(tài)取值進(jìn)行擬合分析,研究結(jié)果表明模型的準(zhǔn)確性不受取值影響,但預(yù)測(cè)結(jié)果存在的偏差和精度受取值影響,取值越大,精度越高,但偏差存在的可能性越大,實(shí)際應(yīng)用時(shí),應(yīng)根據(jù)實(shí)際需求和邊坡樣本精度選取邊坡狀態(tài)取值。降低滑坡影響因素間的相關(guān)性可以降低滑坡因子間相關(guān)性,可以對(duì)模型進(jìn)行優(yōu)化,研究采用安全系數(shù)法,假定單一影響因素為特定變量進(jìn)行滑坡影響因素優(yōu)化,結(jié)果表明對(duì)原回歸項(xiàng)中顯著性不明顯的回歸項(xiàng)進(jìn)行優(yōu)化,其結(jié)果得到明顯的優(yōu)化;而原回歸項(xiàng)中顯著性非常明顯的回歸項(xiàng),對(duì)其進(jìn)行優(yōu)化,則優(yōu)化結(jié)果不明顯;模型整體優(yōu)化效果明顯。(3)滑坡概率與安全系數(shù)應(yīng)用比較和關(guān)系分析。分別求解樣本邊坡的滑坡概率和安全系數(shù),并進(jìn)行等級(jí)劃分,結(jié)果表明滑坡概率和安全系數(shù)評(píng)價(jià)邊坡穩(wěn)定性都有較好的適用性和準(zhǔn)確性,且其工程等級(jí)劃分接近一致。同時(shí)對(duì)滑坡概率和安全系數(shù)進(jìn)行函數(shù)關(guān)系推導(dǎo),并進(jìn)行擬合和驗(yàn)證,可以推導(dǎo)出二者存在函數(shù)關(guān)系,其結(jié)果表示為:lnP/(1-P)=f(F_s)=20.2-16.0F_s。
[Abstract]:In this paper, the application of landslide probability in slope stability analysis is studied by taking the slope of ion type rare earth mining area in Gannan as an example. The main research work and conclusions are as follows: (1) landslide probability model and its application. According to the actual situation of the slope of ion type rare earth mining area in south Jiangxi province, the key influencing factors of landslide were selected, and 21 slope samples were obtained through slope field investigation, geotechnical parameter measurement and data monitoring. The probability model of landslide is established by means of logical regression model combined with deterministic coefficient CF. The landslide probability model and slope sample are used for practical application. A slope in Dongjiang Township Test Mine of Longnan County was selected as an engineering example to solve the landslide probability, including determining the classification of subsets and deterministic coefficient CFCs of each subset, and converting each parameter of slope sample into deterministic coefficient CF value. Using SPSS software to solve the equation of landslide probability, the landslide probability is predicted. The results show that the landslide probability is suitable for slope stability analysis. (2) landslide probability model optimization. The probability analysis of slope stability has defects in accuracy and accuracy. Therefore, three kinds of optimization are carried out, including the selection of subsets of landslide influencing factors, the selection of slope state values and the reduction of the correlation among landslide influencing factors. Two different subset interval classification methods are selected to optimize the subset interval selection. The results show that the class I method, i.e. step size classification, is more suitable. The results show that the accuracy of the model is not affected by the values, but the deviation and accuracy of the prediction results are affected by the values. The greater the value, the higher the accuracy, but the greater the possibility of deviation. In practical application, the value of slope state should be selected according to actual demand and slope sample precision. Reducing the correlation between landslide factors can reduce the correlation between landslide factors, and the model can be optimized. The safety factor method is adopted to optimize the landslide influencing factors, assuming that a single influencing factor is a specific variable. The results show that the results of optimization of the regression terms which are not significant in the original regression term are obviously optimized, but the results are not obvious if the regression terms are very significant in the original regression term. The overall optimization effect of the model is obvious. (3) the application of landslide probability and safety factor is compared and the relationship between landslide probability and safety factor is analyzed. The landslide probability and safety factor of the sample slope are calculated and classified respectively. The results show that the landslide probability and safety factor have good applicability and accuracy in evaluating the slope stability, and the engineering grade classification is close to the same. At the same time, the function relation between landslide probability and safety factor is deduced, and the functional relationship between them can be deduced by fitting and verifying. The result is expressed as 20. 2-16. 0 FSD / (1-P) f (FSP) and 20. 2-16. 0 FSP.
【學(xué)位授予單位】：江西理工大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2017
【分類號(hào)】：TD854.6;TD865

【參考文獻(xiàn)】

相關(guān)期刊論文 前10條

1 黃凱龍;張學(xué)焱;黃永剛;王丹;饒運(yùn)章;;邊坡滑坡概率求解優(yōu)化分析[J];有色金屬科學(xué)與工程;2016年02期

2 饒運(yùn)章;張學(xué)焱;李雪珍;黃永剛;張永勝;;滑坡概率在邊坡穩(wěn)定性分析中的應(yīng)用[J];災(zāi)害學(xué);2016年01期

3 饒運(yùn)章;張學(xué)焱;;基于Logistic回歸模型確定權(quán)重的模糊綜合評(píng)判法在邊坡穩(wěn)定性分析中的應(yīng)用[J];有色金屬科學(xué)與工程;2015年06期

4 亓星;許強(qiáng);鄭光;胡澤銘;;降雨誘發(fā)順層巖質(zhì)及土質(zhì)滑坡動(dòng)態(tài)預(yù)警力學(xué)模型[J];災(zāi)害學(xué);2015年03期

5 孫世國;趙雪芳;王群;郭s，

本文編號(hào)：2073585