After long-term high-intensity injection and extraction development, the fault-block reservoir has entered the extraordinarily high water-bearing period. Dispersed remaining oil makes it difficult to determine the distribution location through the production dynamics and static information. In order to improve the recovery rate of remaining oil in the late development stage. A method of qualitatively identifying the dominant potential area of remaining oil by combining field experience and machine learning based on the geological and development characteristics of the fault block oil reservoir. Through the numerical simulation of the research area, relevant data information is obtained by using the simulated divided grid as a unit. Combining the reservoir engineering relevant knowledge and field experience, the research area is classified as four categories of remaining oil potential areas using six evaluation indexes. The grid is labeled with corresponding labels. The K-neighborhood (KNN) classification prediction algorithm is selected to classify the grid into potential evaluation. The results show that, the KNN algorithm is more accurate for class domain corss-sample classification.The accuracy rate of dividing potential areas reaches 98.3%, which effectively solves the problem of identifying potential areas in extra-high water-bearing reservoirs. This method realizes the accurate positioning of remaining oil potential areas at the late stage of development and assists in developing remaining oil with high efficiency and accuracy.

Beichen Zhao, Chaoxiang Wang. Identifying the Dominant Potential Area Method of Remaining Oil in Fault Block Oil Reservoir Based on KNN Predictive Classification Algorithm. Academic Journal of Engineering and Technology Science (2023) Vol. 6, Issue 3: 1-8. https://doi.org/10.25236/AJETS.2023.060301.

[1] Siqi Ouyang. Study on Flow Unit and Remaining Oil Distribution of Chang 81 Reservoir in Baibao Area, Ordos Basin[D]. Northwest University, 2019. 

[2] Yichao Zhang. Study on the adjustment strategy of the horizontal well development in Bohai BZ oilfield [D]. China University of petroleum, Beijing, 2020. 

[3] Lijie Liu. Classification and evaluation method of remaining oil in ultra-high water cut stage[J]. Petroleum Geology and Recovery Efficiency, 2022, 29(05):83-90. 

[4] Zhenpeng Wang. Development potential classification evaluation for water-flooding in conglomerate reservoir[J]. Lithologic reservoirs, 2018, 30(5):109-115. 

[5] Shuaiwei Ding, Hanqiao Jiang, et al. Classification and evaluation of deepwater oil reservoirs by combining clustering algorithm based on fuzzy C-mean with Bayesian discrimination function[J]. Journal of Xi'an Shiyou University(Natural Science Edition), 2014, 29(02):42-49. 

[6] Fuzhen Chen, Hanqiao Jiang, et al. A quantitative description of reservoir flow fields and its application[J]. Journal of Oil and Gas Technology, 2011, 33(12):110-115. 

[7] COVER T, HART P, et al. Nearest neighbor pattem classi6cation[J]. IEEE Transactions on Information Theory, 1967, 13(1):21-27. 

[8] Bicego M, Loog M. Weighted K-nearest neighbor revisited[C]. International Conference on Pattern Recognition(ICPR), 2016:1642-1647. 

[9] Cheng Huang, Wenjin Pan, et al. Research and application of oil multi-peak model based on machine learing[J]. Journal of Southwest Petroleum University (Science & Technology Edition), 2020, 42(6):75-81. 

[10] Kui Sun. Logging Identification Method of Complex Lithology in Buried Hill Based on the Improved KNN Algorithm[J]. Special Oil & Gas Reservoirs, 2022, 29(03):18-27. 

[11] JIANG J H, CHEN Y J, MENG X Q, et al. A novel density peaks clustering algorithm based on k nearest neighbors for improving assignment process[J]. Physica A:Statistical Mechanics and its Applications, 2019, 523:702-713.