The Fracture Density and Fractal Dimension Prediction Based on Support Vector Machine

Lei  Zhao; Liehui  Zhang; Yilin  Wang

doi:10.4236/ijg.2013.44062

Lei Zhao^*, Liehui Zhang, Yilin Wang

Abstract: The key of effective development for the fractured reservoir is to describe the distribution of the fracture and build the fracture geological model. To acquire more optimal exploration and development of the oilfield, objective geologic model of reservoir fractures is needed for further knowledge of the spatial distribution condition of fractures. Adopting well-logging and seismic techniques can be expensive and usually yield multiple solutions, yet resolution will disturb the seismic method, making it difficult to acquire accurate and sound fracture parameters from seismic data. In this paper, the theoretical foundation for support vector machine, fractal geometry, combined the various information in geology, logging, well core, seism and field outcrop about the fracture and calculate the parameters of the fracture (fracture density and fractal dimension), and the good foundation is established for fractured reservoir description of discrete fracture network model. Based on analysis of conventional prediction methods of development indices and factor influencing the parameters of the fracture, a support vector machine method is established to predict the parameters of the fracture. The new support vector machine method is based on time series analysis to select the kernel function. Trains and tests the support vector machine network with historical data to construct the support vector regression prediction model. A case was fit into the model to test and analyse its reliability, the results suggested that the model had a high prediction performance, and could be well applied to the prediction of fracture parameters.

Keywords: Fractured Reservoir; Support Vector Machine; Fracture Density; Fractal Dimension; MATLAB

Cite this paper: L. Zhao, L. Zhang and Y. Wang, "The Fracture Density and Fractal Dimension Prediction Based on Support Vector Machine," International Journal of Geosciences, Vol. 4 No. 4, 2013, pp. 672-679. doi: 10.4236/ijg.2013.44062.

References

[1] L. Guaiquirian and P. Gonzalez, “Use of Discrete Fracture Network ‘DFN’ to Characterize and Model a Naturally Fractured Sandstone Reservoir: A Case Study of Orocual Field,” SanJuan Formation, Venezuela, 2007, pp. 15-18

[2] R. M. Fang, “Support Vector Machine Theory and Its Application Analysis,” China Electric Power Press, Beijing, 2007.

[3] Z. H. Feng and J. M. Yang, “Practical Selection of Support Vector Machine Parameters for SVM Regression,” Mechanical Engineering & Automation, Vol. 3, 2007, pp. 17-18.

[4] K. Ito and R. Nakano, “Optimizing Support Vector Regression Hyperparameters Based on Cross-Validation,” Proceedings of the International Joint Conference on Neural Networks, Vol. 3, 2003, pp. 2077-2082.

[5] X. H. Han, S. L. Lu, “Dentification of the Lithology of Dujiatai Formation in J13 Well Block with Least Squares Support Vector Machine,” Special Oil & Gas Reservoirs, Vol. 6, 2011, pp. 18-20.

[6] H. Y. Li, S. M. Peng, “Prediction of Inter-Well Fractures with Fractal Technique,” Journal of The University of Petroleum, 2002, pp. 48-50.

[7] L. K. Dong, “Fractal Theory and Applications,” Liaoning Science and Technology Press, Shenyang, 1991, pp. 10-21.

[8] G. L. Zhang, “Assessing Hydrocarbon Accumulation of Fractured Reservoir in Qijia Buried Hill by Fractal Theory,” Special Oil & Gas Reservoirs, Vol. 2, 2002, pp. 15-16.

[9] G. Recktenwald, “Numerical Methods With MATLAB Implementation and Application,” China Machine Press, Beijing, 2004, pp. 128-186

[10] Z. H. Feng and J. M. Yang, “Practical Selection of Support Vector Machine Parameters for SVM Regression,” Mechanical Engineering & Automation, Vol. 3, 2007, pp. 17-18.