Research of Single Well Production Prediction Based on Improved Extreme Learning Machine

摘要

In order to improve the precision of oilfield single well production prediction, a single well production prediction model based on improved extreme learning machine (RWELM) is proposed. Substituting wavelet function for common activation function, structural risk minimization principle is integrated into the model in order to avoid the local minimum and over-fitting problem commonly faced by traditional extreme learning machine (ELM) in single well production forecasting. Dynamic data of an oil well production is simulated of Lun Nan oilfield. Experimental results show that the forecasting model is better than ELM, LM-BP neural networks, BP network with delay time sequence in both generalization performance and predictive accuracy.