Forecasting Method of Crude Oil Output Based on Optimization of LSSVM by Particle Swarm Algorithm

摘要

The accurate prediction of crude oil output plays an important role in the development of oilfield planning. This paper proposes a least squares support vector machine model based on the optimization of particle swarm algorithm (PSO-LSSVM) to predict the crude oil output. Each pair of penalty factor and kernel function parameter was taken as a particle, which follows the optimal particle in the current solution space and adjusts the search direction and speed accordingly. The optimal penalty factor and kernel function parameter were determined by fitness function, and then the optimal LSSVM model was obtained. In this paper, we studied the relationship between crude oil production and its influencing factors by using this model. The experimental results showed that it has fast convergence speed and high prediction accuracy. This study might contribute to the development of the oilfield planning. Moreover, this model will provide a useful reference for the prediction of other dynamic production indexes in oilfield development.