引入小種群的遺傳算法求解阿爾奇公式參數(shù)與彰武地區(qū)油井水飽和度的分析
發(fā)布時(shí)間：2018-08-08 13:31
【摘要】：測(cè)井解釋的方程中一般都會(huì)存在一些定量的參數(shù),譬如作為計(jì)算地層含水飽和度并估算油氣藏的阿爾奇公式,它的公式中的a、m、n值就很難在實(shí)際中被同時(shí)地準(zhǔn)確確定,通常情況下這些參數(shù)都會(huì)被單個(gè),或者兩個(gè)一起來(lái)測(cè)定,然后當(dāng)把獲取的參數(shù)真正地帶入實(shí)際生產(chǎn)中的計(jì)算時(shí),結(jié)果往往不如人意,而這些參數(shù)的正確選取又密切關(guān)系著最后解釋成果的精確度。遺傳算法具有全局尋優(yōu)的特點(diǎn),是一種新型的最優(yōu)化方法,它適合于求解非線性優(yōu)化問(wèn)題,近年來(lái)已經(jīng)在復(fù)雜函數(shù)的優(yōu)化求解結(jié)構(gòu)優(yōu)化設(shè)計(jì)、自適應(yīng)控制、系統(tǒng)控制、模式識(shí)別等領(lǐng)域取得成功應(yīng)用。本文主要闡述了一般遺傳算法的基本原理,概念以及運(yùn)算步驟,其中具體的步驟主要有：1,收集彰武地區(qū)九佛堂組的油井測(cè)井資料,這些資料包括有涉及阿爾奇公式中的同一地層不同井位中的孔隙度,地層水飽和度,地層電阻率,以及巖心實(shí)測(cè)得到的含水飽和度等數(shù)據(jù)；2,確定阿爾奇參數(shù)組合它們各種的大致取值范圍；3,根據(jù)參數(shù)的取值范圍,建立關(guān)于公式中參數(shù)a、m、n由十進(jìn)制到二進(jìn)制的編碼；4,確定算法的適應(yīng)度函數(shù)即建立所求含水飽和度與巖心實(shí)測(cè)含水飽和度差值的絕對(duì)值最小值函數(shù)；5,隨機(jī)生成50組由二進(jìn)制代表的阿爾奇參數(shù)組合；5,確定遺傳算子中關(guān)于交叉概率,變異概率的值；6,運(yùn)用遺傳算子對(duì)阿爾奇參數(shù)組合進(jìn)行遺傳操作,并根據(jù)事先設(shè)置好的終止進(jìn)化代數(shù)結(jié)束算法的迭代過(guò)程,從而選擇出最優(yōu)的阿爾奇參數(shù)組合；7,將選擇出的參數(shù)組合解碼為我們熟知的十進(jìn)制,完成算法擇優(yōu)過(guò)程。然而一般的遺傳算法往往存在容易陷入局部最優(yōu)解的缺陷,本文提出了對(duì)算法的改進(jìn)步驟,即在第6步中,當(dāng)完成一次遺傳算子操作后,重新計(jì)算一次種群中個(gè)體的適應(yīng)度,然后選取種群中適應(yīng)度值排名前70%的個(gè)體,用隨機(jī)新生成的30%的個(gè)體來(lái)代替種群中剩余的那些個(gè)體。這樣做的的目的可以在保證遺傳算法效率的同時(shí),及時(shí)地引入新的個(gè)體,避免因?yàn)椴糠謨?yōu)勢(shì)局部最優(yōu)解個(gè)體占據(jù)種群使算法收斂于局部最佳的情況。極大的提高了算法的準(zhǔn)確性。運(yùn)用該算法能同時(shí)求得阿爾奇公式中的a、m、n值,根據(jù)公式中其他已知參數(shù)可以計(jì)算出該地區(qū)地層的的含水飽和度,通過(guò)把使用這些參數(shù)計(jì)算出的含水飽和度與傳統(tǒng)圖版的方法進(jìn)行對(duì)比,證明該方法準(zhǔn)確性更高,另外一方面,對(duì)比傳統(tǒng)的遺傳算法,發(fā)現(xiàn)改進(jìn)的算法收斂于最佳時(shí)的遺傳代數(shù)明顯小與傳統(tǒng)遺傳算法,而相對(duì)誤差也好于傳統(tǒng)遺傳算法。提示這種改進(jìn)的遺傳算法求解阿爾奇公式的a、m、n值準(zhǔn)確性好且可行性高。適用于該地區(qū)的含水飽和度的預(yù)測(cè)。
[Abstract]:Generally, there are some quantitative parameters in the equation of log interpretation, such as the Archie formula for calculating formation water saturation and estimating oil and gas reservoir. Typically, these parameters are measured individually, or together, and when the obtained parameters are actually taken into actual production calculations, the results are often unsatisfactory. The correct selection of these parameters is closely related to the accuracy of the final interpretation. Genetic algorithm (GA) is a new optimization method with the characteristics of global optimization. It is suitable for solving nonlinear optimization problems. In recent years, genetic algorithm has been used to solve structural optimization design, adaptive control and system control in the optimization of complex functions. Pattern recognition and other fields have been successfully applied. This paper mainly describes the basic principle, concept and operation steps of the general genetic algorithm. The specific steps are: 1, collecting well logging data of Jiufutang formation in Zhangwu area. These data include porosity, formation water saturation, formation resistivity, and water saturation obtained from core measurements in different well locations of the same formation in the Archie formula. (2) determine the approximate range of values of the Archie parameter combinations, and establish the encoding of the parameter aqmn from decimal to binary according to the range of values of the parameters; 4. Determining the fitness function of the algorithm is to establish the absolute minimum value function of the difference between the calculated water saturation and the measured water saturation of the core, and randomly generate 50 sets of Archie parameter combinations represented by binary. 5. The value of crossover probability and mutation probability in genetic operator is determined. Genetic operator is used to perform genetic operation on the combination of Archie parameters, and the iterative process of the ending algorithm based on the pre-set termination of evolutionary algebra is given. Thus, the optimal Archive parameter combination is selected, and the selected parameter combination is decoded into the well-known decimal system to complete the optimization process of the algorithm. However, the general genetic algorithm often has the defect that it is easy to fall into the local optimal solution. In this paper, the improvement steps of the algorithm are put forward, that is, in step 6, the fitness of the individual in a population is re-calculated after the completion of a genetic operator operation. Then, the individuals with the first 70% fitness in the population were selected, and the remaining individuals in the population were replaced by the randomly generated 30% individuals. The purpose of this method is to ensure the efficiency of genetic algorithm and to introduce new individuals in time, so as to avoid the situation that the local optimal solution occupies the population and the algorithm converges to the local best. The accuracy of the algorithm is greatly improved. By using this algorithm, we can simultaneously get the value of Amim ~ n in Archie's formula. According to the other known parameters in the formula, we can calculate the water saturation of the strata in this area. By comparing the water saturation calculated by these parameters with the traditional chart plate method, it is proved that the method is more accurate. On the other hand, compared with the traditional genetic algorithm, It is found that the improved algorithm converges to the optimal genetic algebra obviously smaller than the traditional genetic algorithm, and the relative error is better than the traditional genetic algorithm. It is suggested that the improved genetic algorithm has good accuracy and feasibility in solving the Archie formula. It is suitable for the prediction of water saturation in this area.
【學(xué)位授予單位】：長(zhǎng)江大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：P618.13;P631.81
本文編號(hào)：2171929
[Abstract]:Generally, there are some quantitative parameters in the equation of log interpretation, such as the Archie formula for calculating formation water saturation and estimating oil and gas reservoir. Typically, these parameters are measured individually, or together, and when the obtained parameters are actually taken into actual production calculations, the results are often unsatisfactory. The correct selection of these parameters is closely related to the accuracy of the final interpretation. Genetic algorithm (GA) is a new optimization method with the characteristics of global optimization. It is suitable for solving nonlinear optimization problems. In recent years, genetic algorithm has been used to solve structural optimization design, adaptive control and system control in the optimization of complex functions. Pattern recognition and other fields have been successfully applied. This paper mainly describes the basic principle, concept and operation steps of the general genetic algorithm. The specific steps are: 1, collecting well logging data of Jiufutang formation in Zhangwu area. These data include porosity, formation water saturation, formation resistivity, and water saturation obtained from core measurements in different well locations of the same formation in the Archie formula. (2) determine the approximate range of values of the Archie parameter combinations, and establish the encoding of the parameter aqmn from decimal to binary according to the range of values of the parameters; 4. Determining the fitness function of the algorithm is to establish the absolute minimum value function of the difference between the calculated water saturation and the measured water saturation of the core, and randomly generate 50 sets of Archie parameter combinations represented by binary. 5. The value of crossover probability and mutation probability in genetic operator is determined. Genetic operator is used to perform genetic operation on the combination of Archie parameters, and the iterative process of the ending algorithm based on the pre-set termination of evolutionary algebra is given. Thus, the optimal Archive parameter combination is selected, and the selected parameter combination is decoded into the well-known decimal system to complete the optimization process of the algorithm. However, the general genetic algorithm often has the defect that it is easy to fall into the local optimal solution. In this paper, the improvement steps of the algorithm are put forward, that is, in step 6, the fitness of the individual in a population is re-calculated after the completion of a genetic operator operation. Then, the individuals with the first 70% fitness in the population were selected, and the remaining individuals in the population were replaced by the randomly generated 30% individuals. The purpose of this method is to ensure the efficiency of genetic algorithm and to introduce new individuals in time, so as to avoid the situation that the local optimal solution occupies the population and the algorithm converges to the local best. The accuracy of the algorithm is greatly improved. By using this algorithm, we can simultaneously get the value of Amim ~ n in Archie's formula. According to the other known parameters in the formula, we can calculate the water saturation of the strata in this area. By comparing the water saturation calculated by these parameters with the traditional chart plate method, it is proved that the method is more accurate. On the other hand, compared with the traditional genetic algorithm, It is found that the improved algorithm converges to the optimal genetic algebra obviously smaller than the traditional genetic algorithm, and the relative error is better than the traditional genetic algorithm. It is suggested that the improved genetic algorithm has good accuracy and feasibility in solving the Archie formula. It is suitable for the prediction of water saturation in this area.
【學(xué)位授予單位】：長(zhǎng)江大學(xué)
【學(xué)位級(jí)別】：碩士
【學(xué)位授予年份】：2015
【分類號(hào)】：P618.13;P631.81
【參考文獻(xiàn)】
相關(guān)期刊論文 前8條
1 潘保芝,薛林福,李舟波,陶宏根,王鵬;地層的自動(dòng)分級(jí)對(duì)比技術(shù)[J];長(zhǎng)春科技大學(xué)學(xué)報(bào);2001年04期
2 王天波,董春旭,徐麗萍,梁先忠;用分形理論確定m、n值的方法及其應(yīng)用[J];測(cè)井技術(shù);1998年01期
3 文浩;;遺傳模擬退火算法在阿爾奇公式參數(shù)估計(jì)中的應(yīng)用[J];斷塊油氣田;2008年01期
4 劉英明;鄒長(zhǎng)春;;一種基于遺傳算法的測(cè)井曲線高分辨率處理方法[J];地球物理學(xué)進(jìn)展;2006年04期
5 徐宗本,高勇;遺傳算法過(guò)早收斂現(xiàn)象的特征分析及其預(yù)防[J];中國(guó)科學(xué)E輯:技術(shù)科學(xué);1996年04期
6 沈愛(ài)新,王黎,陳守軍;油層低電阻率及阿爾奇公式中各參數(shù)的巖電實(shí)驗(yàn)研究[J];江漢石油學(xué)院學(xué)報(bào);2003年S1期
7 張振城,孫建孟,馬建海,霍玉雁;阿爾奇公式中a,m值對(duì)飽和度計(jì)算結(jié)果的影響[J];石油大學(xué)學(xué)報(bào)(自然科學(xué)版);2004年06期
8 馮國(guó)慶,陳軍,張烈輝,馮異勇;最優(yōu)化測(cè)井解釋的遺傳算法實(shí)現(xiàn)[J];天然氣工業(yè);2002年06期
，本文編號(hào)：2171929
本文鏈接：http://sikaile.net/kejilunwen/diqiudizhi/2171929.html
最近更新
教材專著
