摘要
针对含缺陷管道失效压力预测这一问题,使用BAS优化算法对传统BP神经网络的初始权值和阈值进行优化,并采用含缺陷管道爆破试验数据对BP神经网络和BAS-BPNN模型分别进行训练预测。BAS-BPNN模型预测结果的最小相对误差为0.09%,最大相对误差为9.32%,平均相对误差仅为6.04%,决定系数R2为0.9779,预测结果精度相较于传统BP神经网络有了较大提升。
Considering the prediction of failure pressure of the defective pipeline,making use of the BAS optimization algorithm to optimize both initial weights and thresholds of the traditional BP neural network was implemented,including having BP neural network and BAS-BPNN model trained and predicted by using the test data of defected pipeline’s bursting.The results show that,the minimum relative error and the maximum relative error from the BAS-BPNN model is 0.09%and 9.32%respectively.The average relative error is 6.04%only and coefficient of determination R2 is 0.9779.The prediction accuracy,as compared to the traditional BP neural network,is improved obviously.
作者
姜峰
马娟娟
凌晓
郭凯
JIANG Feng;MA Juan-juan;LING Xiao;GUO Kai(College of Petroleum and Chemical Engineering,Lanzhou University of Technology)
出处
《化工机械》
CAS
2022年第6期974-980,共7页
Chemical Engineering & Machinery
基金
甘肃省自然科学基金项目(20JR5RA451,21JR7RA221)。