改进BP神经网络在管道腐蚀速率预测中的应用  

The Application of the Improved BP Neural Network in Prediction of Pipeline Corrosion Rate,

作　　者：刘刚[1] 周志龙[2] 王正涛[3] 张小平[1] 刘楠[4] 

机构地区：[1]中国人民解放军92426部队 [2]中国人民解放军63680部队 [3]中国人民解放军75232部队 [4]中国人民解放军后勤工程学院

出　　处：《油气储运》2006年第4期34-37,58,共5页Oil & Gas Storage and Transportation

摘　　要：针对影响管道腐蚀速率预测精度出现的BP算法易陷入局部最小值、收敛速度慢和引起振荡效应等问题,根据改进自适应遗传算法在广泛的空间搜索和向最优解的方向尽快收敛于最优目标的特点,提出了使用改进自适应遗传算法优化BP神经网络,构建了优化的混合算法神经网络模型。实际应用表明,该模型将大大提高网络的学习效率和预测评判的准确率。The BP shortcomings are apt to fall into some minimum,slowing to converge and causing the effect of shaking, influencing its application to the prediction of the pipeline corrosion rate. According to the characteristic of the improved adaptive GA—EIAGA in extensive space search and converging to the optimum goal as soon as possible in the optimum direction, this paper points out that the improved adaptive GA—IAGA should be used to optimize the BP neural network and structure the optimized mix algorithm neural network model. The application of the optimized model to the pipeline corrosion rate prediction shows that it will prove greatly the learning effectively and the accuracy in prediction and judging.

关 键 词：管道 BP神经网络 优化 腐蚀速率 预测 应用 

分 类 号：TP18[自动化与计算机技术—控制理论与控制工程] TU311.41[自动化与计算机技术—控制科学与工程]