平均绝对百分比误差
粒子群优化
均方误差
遗传算法
支持向量机
体积热力学
管道(软件)
超参数优化
结构工程
算法
材料科学
工程类
计算机科学
数学
人工智能
统计
机器学习
机械工程
物理
量子力学
作者
Kangjian Yang,Hongyuan Fang,Hongjin Liu,Bin Li,Xijun Zhang,Yangyang Xia,Kejie Zhai
标识
DOI:10.1016/j.ijmecsci.2023.108665
摘要
Corrosion and cracks are common issues in drainage pipelines. To investigate the mechanical properties of pipes with defects, a series of pipeline bearing capacity tests were carried out. In addition, a prediction model using a combination of self-organizing maps, genetic algorithms, and support vector machines (SOM-GA-SVM) was developed to predict the bearing capacity and circumferential strain of the pipeline. The prediction results obtained using this model were compared with those obtained using three other optimization algorithms. Furthermore, the influence of loading speed and defect volume on the prediction accuracy of the model was analyzed. The results indicated that the MAPE of the prediction results was less than 7%, the RMSE was less than 8, and the R2 was greater than 0.98; Additionally, the prediction accuracy of the SOM-GA algorithm was significantly higher than that of the genetic algorithm, particle swarm optimization algorithm, and grid search method; It was found that removing the loading speed from the variables and changing the defect depth and width to defect volume can improve the prediction accuracy of the model. After removing the loading speed, the average MAPE and average RMSE of the prediction model were reduced by 7.357% and 8.385%, respectively. After changing the defect depth and width to defect volume, the average MAPE and average RMSE of the prediction model were reduced by 4.925% and 5.054%, respectively.
科研通智能强力驱动
Strongly Powered by AbleSci AI