Preference Driven Multi-objective Optimization of Beam Pumping Process

Obtaining optimal decision parameters has significant meaning to improve the beam pumping process's inefficiency and energy-intensity. However, affected by uncertainties from mechanism, geological environment and human, it is hard to grasp the relationships among the operation parameters, the environment variables and the performances. This paper proposes a preference driven multi-objective optimization method to achieve optimal decision parameters based on neutral network model. First, using back propagation neutral network to find beam pumping system's latent rule represented by a model. Furthermore, constructing the preference function of oil yield, and finally using Non-dominated Sorting Genetic Algorithm II to optimize the preference driven multi-objective optimization problem which reaches the optimal decision parameters. The experimental results show that the optimal decision parameters can reduce the energy consumption about 15.87%, which proves the feasibility and effectiveness of the proposed method.