Application of improved support vector machine in geochemical lithology identification

Yin, Shitao; Lin, Xiaochun; Huang, Yongjian; Zhang, Zhifeng; Li, Xiang

doi:10.1007/s12145-022-00932-2

Cited by 8 publications

(5 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Among them, ANN depends on the network topology, initial weights and thresholds (Saporetti et al 2021), thus requiring a large number of parameters during the training process, thousands of iterations leading to a slow convergence rate and a tendency to fall into local minima in case of more local minima, with subsequent uncertainty in the output results. SVM uses a kernel function to map data into high-dimensional space and builds a hyperplane based on training data to classify complex lithology, Yin et al (2023) proposed an affiliation-weighted one-toone SVM method for lithology identification, which is better for lithology classification in small samples or when the data are unbalanced. Lu et al (2022) proposed a gray wolf optimized SVM lithology identification method, which showed strong adaptability in complex reservoirs.…”

Section: Introductionmentioning

confidence: 99%

Research on Intelligent Recognition Technology in Lithology Based on Multi- parameter Fusion of Logging While Drilling

Liang,

Xiong,

Yang

et al. 2023

Preprint

View full text Add to dashboard Cite

In oil and gas drilling, timely and accurate identification of formation lithology is an important guarantee of drilling safety. Aiming at the problems of inaccurate identification of lithology in drilling by traditional methods, and low efficiency due to the fact that even modern instruments cannot respond to lithology in real time. the Categorical Boost (CatBoost) model was applied to lithology identification in this study. However, since CatBoost uses more hyperparameters in its modeling, it is difficult to optimize model prediction by manually tuning the parameters. Therefore, the introduction of Kernel Principal Component Analysis (KPCA) extracts fewer and more important features from the original data, eliminates the redundant information contained therein, and combines with Bayesian Optimization (BO) algorithm to optimize the hyperparameters during the training process, thus improving the prediction performance of CatBoost. Two experiments were designed to verify the recognition ability of the model, and the final test results of the model showed that the KPCA-BO-CatBoost model proposed in this paper had the best overall performance, and the lithology recognition accuracy reached more than 90%. The model was effective in identifying the formation lithology, realized real-time lithology identification by combining the parameters of logging while drilling, improved the efficiency and accuracy of lithology identification, and was of great significance in guiding the subsequent drilling work.

show abstract

Section: Introductionmentioning

confidence: 99%

Research on Intelligent Recognition Technology in Lithology Based on Multi- parameter Fusion of Logging While Drilling

Liang,

Xiong,

Yang

et al. 2023

Preprint

View full text Add to dashboard Cite

show abstract

“…Lithological classification information is important and basic for mineral resource exploration and geological disaster monitoring. Understanding the spatial distribution characteristics and variability of surface lithology is of great significance for regional geological mapping and mineral resource potential prediction in areas with high altitudes and poor transportation [1][2][3][4]. Traditional lithological mapping involves aerial photo examination and mapping based on interpretation keys, examination of the rocks in the field, rock sampling, and their examination in the laboratory.…”

Section: Introductionmentioning

confidence: 99%

Lithological Classification by Hyperspectral Images Based on a Two-Layer XGBoost Model, Combined with a Greedy Algorithm

Lin

Jiang

et al. 2023

Remote Sensing

View full text Add to dashboard Cite

Lithology classification is important in mineral resource exploration, engineering geological exploration, and disaster monitoring. Traditional laboratory methods for the qualitative analysis of rocks are limited by sampling conditions and analytical techniques, resulting in high costs, low efficiency, and the inability to quickly obtain large-scale geological information. Hyperspectral remote sensing technology can classify and identify lithology using the spectral characteristics of rock, and is characterized by fast detection, large coverage area, and environmental friendliness, which provide the application potential for lithological mapping at a large regional scale. In this study, ZY1-02D hyperspectral images were used as data sources to construct a new two-layer extreme gradient boosting (XGBoost) lithology classification model based on the XGBoost decision tree and an improved greedy search algorithm. A total of 153 spectral bands of the preprocessed hyperspectral images were input into the first layer of the XGBoost model. Based on the tree traversal structural characteristics of the leaf nodes in the XGBoost model, three built-in XGBoost importance indexes were split and combined. The improved greedy search algorithm was used to extract the spectral band variables, which were imported into the second layer of the XGBoost model, and the bat algorithm was used to optimize the modeling parameters of XGBoost. The extraction model of rock classification information was constructed, and the classification map of regional surface rock types was drawn. Field verification was performed for the two-layer XGBoost rock classification model, and its accuracy and reliability were evaluated based on four indexes, namely, accuracy, precision, recall, and F1 score. The results showed that the two-layer XGBoost model had a good lithological classification effect, robustness, and adaptability to small sample datasets. Compared with the traditional machine learning model, the two-layer XGBoost model shows superior performance. The accuracy, precision, recall, and F1 score of the verification set were 0.8343, 0.8406, 0.8350, and 0.8157, respectively. The variable extraction ability of the constructed two-layer XGBoost model was significantly improved. Compared with traditional feature selection methods, the GREED-GFC method, when applied to the two-layer XGBoost model, contributes to more stable rock classification performance and higher lithology prediction accuracy, and the smallest number of extracted features. The lithological distribution information identified by the model was in good agreement with the lithology information verified in the field.

show abstract

“…At present, the methodology of identifying lithology or lithofacies of carbonate rocks using well logs mainly includes the qualitative analysis of curve characteristics, crossplot method, apparent rock fabric number (ARFN) method, and relevant mathematical analysis methods [15][16][17][18][19][20][21][22][23][24]. Qualitative analyses identify lithofacies according to well log response features, curve shape, values of well logs, and imaging characteristics of different rocks [15,16].…”

Section: Introductionmentioning

confidence: 99%

“…Furthermore, there are also some lithofacies or lithology identification methods based on machine-learning techniques and mathematical algorithms, such as neural network [19], cluster analysis [20], principal component analysis [21], grey correlation [22], support vector machine [23], and Naive Bayesian method [24]. The key to these methods is to build a well-trained and suitable model which can be applied to well log measurements from wells with poor core data to predict lithofacies.…”

Section: Introductionmentioning

confidence: 99%

Lithofacies Characteristics and Methodology to Identify Lacustrine Carbonate Rocks via Log Data: A Case Study in the Yingxi Area, Qaidam Basin

Tian,

Liu,

Zhu

et al. 2023

Energies

View full text Add to dashboard Cite

Lacustrine carbonate reservoirs, extensively distributed in China, have extensive oil and gas exploration potential. However, such reservoirs are characterized by high content of terrigenous debris and complex lithofacies, and the resultant high difficulty in lithofacies identification severely restrains exploration expansion and efficient development, especially for the Upper Member of the Paleogene Lower Ganchaigou Formation (E32) of the Yingxi area in the Qaidam Basin, with burial depths generally greater than 4000 m. This research targets this area and develops a methodology for detailed lithofacies identification, after systematically investigating the characteristics of lithofacies and well log responses of lacustrine carbonate rocks, on the basis of a massive volume of data of cores, thin sections, and experiments of the study area. The analysis identified lithofacies in the Upper Member of the Paleogene Lower Ganchaigou Formation of the Yingxi area, namely, pack-wackestone, mudstone, laminated carbonate, muddy gypsum, and limy claystone. The analysis of well log response characteristics suggested that natural gamma ray, matrix density, and bulk density were sensitive to lithofacies. Then, for the first time, the rock fabric factor (RFF) method was proposed, and the lithofacies identification plot was based on the calculated RFF and high-definition spectroscopy log. The presented methodology was applied to 55 wells in the study area. The average accuracy of lithofacies interpretation in 14 cored wells reached 82.4%, indicating good application performance. This method improves the lithofacies identification accuracy of lacustrine carbonate rocks, which is of great significance for investigating the reservoir distribution law and guiding exploration and development.

show abstract

Application of improved support vector machine in geochemical lithology identification

Cited by 8 publications

References 37 publications

Research on Intelligent Recognition Technology in Lithology Based on Multi- parameter Fusion of Logging While Drilling

Research on Intelligent Recognition Technology in Lithology Based on Multi- parameter Fusion of Logging While Drilling

Lithological Classification by Hyperspectral Images Based on a Two-Layer XGBoost Model, Combined with a Greedy Algorithm

Lithofacies Characteristics and Methodology to Identify Lacustrine Carbonate Rocks via Log Data: A Case Study in the Yingxi Area, Qaidam Basin

Contact Info

Product

Resources

About