摘要
为了提高预测学习系统的预测能力,给出了Boosting提升BP神经网络的算法描述.通过一个加权的多数表决合并了全部预测,提高了BP算法的准确性和泛化能力.将该算法应用于油田水淹层识别,与BP神经网络算法相比,训练得到的模型具有更好的泛化能力,提高了水淹层识别的解释精度.测试集识别率达到73%.
It is very easy for the BP neural network learning to be trapped into the local optimization and the low generation ability.In order to improve the prediction ability of learning system,a algorithm of Boosting improving BP neural network is given in our paper,which combines all the predictions through weighting,and enhances the accuracy and generation capacity of BP algorithm.We use this algorithm in the oil field flooded layer identification and get better generation ability by the training model comparing the BP neural network algorithm,and also advance the explaining precision.The recognizing ratio of testing assemblage is up to 73%.
出处
《大庆石油学院学报》
CAS
北大核心
2006年第3期97-99,共3页
Journal of Daqing Petroleum Institute