In the oil industry, it is common to use submersible electric pumps (ESP) operating with multiphase fluid flow "gas-liquid". The presence of large amounts of gas in the pump generates instabilities and degradation in performance of the pump. In this paper it is presented a method to create a...

In the oil industry, it is common to use submersible electric pumps (ESP) operating with multiphase fluid flow "gas-liquid". The presence of large amounts of gas in the pump generates instabilities and degradation in performance of the pump. In this paper it is presented a method to create a nonlinear model, which interpolates the behavior of an ESP operating with multiphase fluid flow, using artificial intelligence from experimental data.The method is based on support vector machines for regression (SVMr), which is a robust tools capable of solving nonlinear and high dimensional problems. Once the SVMr have internal parameters that influence the performance of it, this paper used a genetic algorithm as optimizer of the SVMr parameters to reach a good solution to the problem. The results obtained with SVMr and the data set shows that it is possible to obtain representative models with artificial intelligence and a limited number of training data

Fechado