Reservoir Properties Prediction Based on Support Vector Regression with Optimized Parameters by Quantum Particle Swarm
X. Liu, X. Chen, J. Li, L. Zhou and C. Li
Event name: 81st EAGE Conference and Exhibition 2019
Session: Poster: Integrated Subsurface Studies - Seismic Reservoir Characterization A
Publication date: 03 June 2019
Info: Extended abstract, PDF ( 989.45Kb )
Price: € 20
Reservoir properties are the main parameters to characterize the reservoir. Therefore, it is necessary to estimate reservoir properties reasonably and effectively. Prediction of physical parameters in well-free areas using seismic data is a difficult but important work. Support vector regression can be used to predict physical parameters. However, the application is limited because of the complexity of parameter determination. We combine quantum particle swarm optimization algorithm to obtain relevant parameters for support vector regression predicting reservoir properties, which improves the efficiency and accuracy of parameter acquisition, thus improving the prediction effect. The test of model data and field data verifies that the method can predict the reservoir properties effectively and efficiently, even in the case of small samples, high accuracy is obtained. This method is eventually used to predict the reservoir properties by utilizing seismic information. In conclusion, the reservoir properties prediction method based on support vector regression with optimized parameters by quantum particle swarm has good application prospects.