摘要
腐蚀速率是反映管道腐蚀动力学过程的重要特征参数,为实现对管道长期运行可靠性和剩余寿命的精准评估,对腐蚀速率的预测显得尤为重要。最小二乘支持向量机(LSSVM)是一种基于机器学习的方法,常用于分类和预测研究,惩罚参数γ与核参数σ2是LSSVM的2个重要参数,在进行计算时只能经验取值,对计算结果影响较大。通过利用遗传算法(GA)对参数进行寻优,构建了GA-LSSVM预测模型,并将模型应用到管道腐蚀速率的预测。通过与其他预测模型的结果进行比较表明,GA-LSSVM模型精度和预测结果精度较高,可实现对管道腐蚀速率的预测。
Corrosion rate is an important characteristic parameter to reflect the corrosion dynamics process of pipeline. In order to accurately evaluate the long-term operation reliability and remaining life of pipeline,the prediction of corrosion rate is particularly important. Least squares support vector machine( LSSVM) is a method based on machine learning,which is often used in classification and prediction research.Since penalty parameters γ and kernel parameters σ2 are two important parameters of LSSVM,the value of these two parameters can only be obtained by experience in calculation,causing a great impact on the calculation results. In this paper,the genetic algorithm( GA) was used to optimize the parameters,the GA-LSSVM prediction model was built and the model was applied to the prediction of pipeline corrosion rate.Compared with the results of other prediction models,the results showed that the accuracy of GA-LSSVM model and prediction results were relatively higher,which could realize the prediction of pipeline corrosion rate.
作者
陈永红
苏永生
胡平
CHEN Yong-hong;SU Yong-sheng;HU Ping(Collge of Mechanical and Electrical Engineering,Wuhan Donghu University,Wuhan 430212,China;College of Power Engineering,Naval Univeristy of Engineering,Wuhan 430030,China;College of Ships and Oceangraphy,Naval Univeristy of Engineering,Wuhan 430030,China)
出处
《材料保护》
CAS
CSCD
2021年第1期63-67,共5页
Materials Protection
基金
武汉东湖学院青年基金(2020DHZK002)资助。
