基于GA-BP神经网络的输油管道内腐蚀速率预测

Prediction of Corrosion Rate in Oil Pipeline By GA-BP Neural Network

针对BP(Back Propagation)算法的神经网络易陷入局部极小的缺点,尝试引入遗传算法(Genetic Algorithm)处理了BP神经网络的权值与阈值,并分别建立了BP和GA-BP两种不同算法的神经网络模型对输油管道腐蚀速率进行预测。对结果进行对比分析发现,GA-BP算法的神经网络预测精度要高于BP网络;GA-BP算法具有高的预测精度,其预测值与实验值相关系数为0.9863,表明该算法模型是合理可靠的。

In order to overcome the defect of the local minimal of BP neural network,genetic algorithm was introduced to deal with the initial weight values and threshold in BP neural network algorithm,then the BP and GA-BP neural network was established to predict corrosion rate of the oil pipeline.From the prediction results compared with two neural networks,we found that： the forecast accuracy of GA-BP neural net work was higher than that of BP neural network,and the correlation coefficient（RGA-BP） between prediction results and experimentation results was 0.9863,showing that GA-BP neural network was reasonable.