Mehdi Rahmati , Ghasem Zargar , Abbas Ayatizadeh Tanha
{"title":"利用机器学习技术和启发式优化算法从测井曲线预测密度曲线:比较研究","authors":"Mehdi Rahmati , Ghasem Zargar , Abbas Ayatizadeh Tanha","doi":"10.1016/j.ptlrs.2024.01.008","DOIUrl":null,"url":null,"abstract":"<div><p>In the petroleum industry, the analysis of petrophysical parameters is critical for efficient reservoir management, production optimization, development strategies, and accurate hydrocarbon reserve estimations. Over recent years, the integration of machine learning methodologies has revolutionized the field, addressing challenges in geology, geophysics, and petroleum engineering, even when confronted with limited or imperfect data. This study focuses on the prediction of density logs, a pivotal factor in evaluating reservoir hydrocarbon volumes. It is important to note that during well logging operations, log data for specific depths of interest may be missing or incorrect, presenting a significant challenge. To tackle this issue, we employed the Adaptive Neuro-Fuzzy Inference System (ANFIS) and Artificial Neural Networks (ANN) in combination with advanced optimization algorithms, including Particle Swarm Optimization (PSO), Imperialist Competitive Algorithms (ICA), and Genetic Algorithms (GA). These methods exhibit promising performance in predicting density logs from gamma-ray, neutron, sonic, and photoelectric log data. Remarkably, our results highlight that the Genetic Algorithms-based Artificial Neural Network (GA-ANN) approach outperforms all other methods, achieving an impressive Mean Squared Error (MSE) of 0.0013. In comparison, ANFIS records an MSE of 0.0015, ICA-ANN 0.0090, PSO-ANN 0.0093, and ANN 0.0183.</p></div>","PeriodicalId":19756,"journal":{"name":"Petroleum Research","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-06-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.sciencedirect.com/science/article/pii/S2096249524000085/pdfft?md5=9c5236924d89719aebc109365066ee17&pid=1-s2.0-S2096249524000085-main.pdf","citationCount":"0","resultStr":"{\"title\":\"Predicting density log from well log using machine learning techniques and heuristic optimization algorithm: A comparative study\",\"authors\":\"Mehdi Rahmati , Ghasem Zargar , Abbas Ayatizadeh Tanha\",\"doi\":\"10.1016/j.ptlrs.2024.01.008\",\"DOIUrl\":null,\"url\":null,\"abstract\":\"<div><p>In the petroleum industry, the analysis of petrophysical parameters is critical for efficient reservoir management, production optimization, development strategies, and accurate hydrocarbon reserve estimations. Over recent years, the integration of machine learning methodologies has revolutionized the field, addressing challenges in geology, geophysics, and petroleum engineering, even when confronted with limited or imperfect data. This study focuses on the prediction of density logs, a pivotal factor in evaluating reservoir hydrocarbon volumes. It is important to note that during well logging operations, log data for specific depths of interest may be missing or incorrect, presenting a significant challenge. To tackle this issue, we employed the Adaptive Neuro-Fuzzy Inference System (ANFIS) and Artificial Neural Networks (ANN) in combination with advanced optimization algorithms, including Particle Swarm Optimization (PSO), Imperialist Competitive Algorithms (ICA), and Genetic Algorithms (GA). These methods exhibit promising performance in predicting density logs from gamma-ray, neutron, sonic, and photoelectric log data. Remarkably, our results highlight that the Genetic Algorithms-based Artificial Neural Network (GA-ANN) approach outperforms all other methods, achieving an impressive Mean Squared Error (MSE) of 0.0013. In comparison, ANFIS records an MSE of 0.0015, ICA-ANN 0.0090, PSO-ANN 0.0093, and ANN 0.0183.</p></div>\",\"PeriodicalId\":19756,\"journal\":{\"name\":\"Petroleum Research\",\"volume\":null,\"pages\":null},\"PeriodicalIF\":0.0000,\"publicationDate\":\"2024-06-01\",\"publicationTypes\":\"Journal Article\",\"fieldsOfStudy\":null,\"isOpenAccess\":false,\"openAccessPdf\":\"https://www.sciencedirect.com/science/article/pii/S2096249524000085/pdfft?md5=9c5236924d89719aebc109365066ee17&pid=1-s2.0-S2096249524000085-main.pdf\",\"citationCount\":\"0\",\"resultStr\":null,\"platform\":\"Semanticscholar\",\"paperid\":null,\"PeriodicalName\":\"Petroleum Research\",\"FirstCategoryId\":\"1087\",\"ListUrlMain\":\"https://www.sciencedirect.com/science/article/pii/S2096249524000085\",\"RegionNum\":0,\"RegionCategory\":null,\"ArticlePicture\":[],\"TitleCN\":null,\"AbstractTextCN\":null,\"PMCID\":null,\"EPubDate\":\"\",\"PubModel\":\"\",\"JCR\":\"Q1\",\"JCRName\":\"Earth and Planetary Sciences\",\"Score\":null,\"Total\":0}","platform":"Semanticscholar","paperid":null,"PeriodicalName":"Petroleum Research","FirstCategoryId":"1087","ListUrlMain":"https://www.sciencedirect.com/science/article/pii/S2096249524000085","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q1","JCRName":"Earth and Planetary Sciences","Score":null,"Total":0}
Predicting density log from well log using machine learning techniques and heuristic optimization algorithm: A comparative study
In the petroleum industry, the analysis of petrophysical parameters is critical for efficient reservoir management, production optimization, development strategies, and accurate hydrocarbon reserve estimations. Over recent years, the integration of machine learning methodologies has revolutionized the field, addressing challenges in geology, geophysics, and petroleum engineering, even when confronted with limited or imperfect data. This study focuses on the prediction of density logs, a pivotal factor in evaluating reservoir hydrocarbon volumes. It is important to note that during well logging operations, log data for specific depths of interest may be missing or incorrect, presenting a significant challenge. To tackle this issue, we employed the Adaptive Neuro-Fuzzy Inference System (ANFIS) and Artificial Neural Networks (ANN) in combination with advanced optimization algorithms, including Particle Swarm Optimization (PSO), Imperialist Competitive Algorithms (ICA), and Genetic Algorithms (GA). These methods exhibit promising performance in predicting density logs from gamma-ray, neutron, sonic, and photoelectric log data. Remarkably, our results highlight that the Genetic Algorithms-based Artificial Neural Network (GA-ANN) approach outperforms all other methods, achieving an impressive Mean Squared Error (MSE) of 0.0013. In comparison, ANFIS records an MSE of 0.0015, ICA-ANN 0.0090, PSO-ANN 0.0093, and ANN 0.0183.