改进神经网络在预测评判油气田设备腐蚀分析中的应用

主要针对一般BP神经网络易陷入局部最小值、收敛速度慢、引起扬荡效应的缺点,提出用一种改进遗传算法对BP网络的权值、阈值进行训练,构建优化的混合算法神经网络模型。在华北油田某管道的腐蚀情况分析中,证明了该方法的正确性和优越性。

The Application of Improved Neural Network to the Prediction and Judging of the Oil and Gas Field Equipment Erosion

The BP shortcoming of apting to fall into some minimum,slowing to converge and causing the effect of shaking influence its application to predict and judge the erosion state of tank.According to the Characteristic of the improved adaptive GA-IAGA in extensive space search and converging to the optimum goal as soon as possible in the optimum direction,the text have proposed that we should use the improved adaptive GA-IAGA to optimize the BP neural network and structure the optimized mix Algorithm neural network model.The application of the optimized model to the tank erosion prediction and judging of a oil depot in Huabei Oilfield shows that its use will prove greatly the learning efficiency and the accurate rate in prediction and judging.