A hybrid modeling method for interval time prediction of the intermittent pumping well based on IBSO-KELM

Abstract

In actual oilfield production, the intermittent pumping well’s interval time is mainly determined by the production experience of managers, which makes it hard for the oil well to reach the best production efficiency with stable production and lower energy consumption. To solve this problem, a hybrid model based on kernel extreme learning machine optimized by an improved brain storm optimization (IBSO-KELM) method is presented in this paper. The analytical model is first built to get the experiential interval time using experiential values of many production parameters; then, the IBSO-KELM model is used to compensate the calculation error of the analytical model. In the KELM model, the values of two model parameters are optimally selected by IBSO algorithm which has a self-adaptive ability between the global searching and the local searching and also has lower computation complexity. Two evaluation indexes are used to improve the predictive performance of the IBSO-KELM model, in which, the compensation evaluation index is used to start the error compensation calculation for the experiential time and the model evaluation index is used to judge whether the established off-line model is suitable to the current working condition. Case studies using production data of one oil well in China are conducted to illustrate the effectiveness of the proposed method.