Forecast Model for Gas Well Productivity Based on GA and SVM

Abstract:

The accurate prediction of gas well productivity is an important task in gas reservoir engineering research. According to the global optimization ability of the genetic algorithm (GA) and the superior regression performance of the support vector machine (SVM), this paper proposed a method based on GA and SVM to improve the prediction accuracy. As the proposed model can reduce the dimensionality of data space and preserve features of gas well productivity, compared with BP neural network model, the proposed GA-SVM model for gas well productivity in practical engineering has higher accuracy and speed, the maximum error is 1.5%. Thus, it provided a new method for the forecast of gas well productivity.

• Periodical: Applied Mechanics and Materials (Volumes 71-78)
• Pages: 4958-4962
• DOI: https://doi.org/10.4028/www.scientific.net/AMM.71-78.4958
• Online since: July 2011
• Authors: Chun Bi Xu, Jing Cheng Liu, Jun Li
• Keywords: Forecast Model, GA, Gas Well Productivity, Support Vector Machine (SVM)

References

[1] Yuanqian Chen. Numerical Computation of Gas/Oil Reservoir (Oil Industry Press, Beijing 1990), in Chinese.

[2] Weishu Fu, Shunli He and Yingzhi Ran. Offshore Oil Vol. 24(2004), pp.62-68, in Chinese.

[3] Min Xu, Fenxia Zhao. Natural Gas Industry Vol. 20(2000), pp.73-75, in Chinese.

[4] Vapnik V N. The nature of statistical learning theory (Springer, NewYork 1995).

[5] Cristianini N, Taylor J S. An introduction to support vector machines and other kernel-based learning methods(Cambridge University Press, London 2000).

DOI: 10.1017/cbo9780511801389

[6] Jie Shang, Zhanli Zhu. Corrosion Science and protection technology Vol. 19(2004), pp.225-228.

[7] Ming Zhou, Shudong Sun. Theories and Application of Genetic Algorithms (National Defense Industry Press, Beijing 1999).

[8] Keqi Wang, shaochun Yang, Tianhong Dai et. al. Computer Applications and Software Vol. 26(2009), pp.109-111, in Chinese.