Prediction Method of Crude Oil Production Based on FCM_GA_BP Neural Network

摘要

Accurate prediction of crude oil production is the basis for analyzing oil reservoir status and making oilfield development plans. To resolve the problem of uncertain structure and network over-fitting issue BP neural network has in predicting crude oil production, this paper proposes an optimized BP neural network method based on fuzzy clustering algorithm and generic algorithm, utilizing the global search ability of genetic algorithm and the data screening ability of fuzzy clustering algorithm. This paper selected historical data of 15 oil fields in China as training samples. The experimental prediction results show that, the optimized BP neural network prediction model based on fuzzy clustering and genetic algorithm has stable structure, which effectively prevents the over-fitting problem and has higher prediction accuracy.