CONTACT US: Contact info
|
Citation |
Lu H, Xu ZD, Iseley T, Matthews JC. J. Pipeline Syst. Eng. Pract. 2021; 12(4): e587. |
|
Copyright |
(Copyright © 2021, American Society of Civil Engineers) |
|
DOI |
|
PMID |
unavailable |
|
Abstract |
For the residual strength prediction of corroded pipelines, the existing standard has a small application range, and the finite-element method has too many assumptions. This paper proposes a new data-driven prediction framework. Firstly, principal component analysis (PCA) is used to reduce the dimensions of the existing data to determine the input-output structure of the prediction model. Secondly, support vector machine (SVM) based on multiobjective optimization is employed to predict the pipeline's residual strength. Compared with the traditional estimation methods, the model proposed in this paper is data-driven and combines data dimension reduction, multiobjective optimization, and a machine learning model. In addition, the accuracy and stability of the model are considered in the multiobjective optimization. The proposed framework is tested in a pipeline burst pressure data set. The results indicate that the mean absolute percentage error of the proposed models ranges from 1.353% to 3.220%, which has good prediction accuracy and stability. This paper also discusses the influence of the multiobjective optimization algorithm and dimension reduction on the prediction model. The following primary conclusions are drawn: (1) SVM optimized by multiobjective optimizer performs better than SVM optimized by the single-objective optimizer, and the original SVM performs worst, and (2) reducing the raw data dimensions can improve the residual strength prediction performance for corroded pipelines reduce the complexity of the model, and shorten the calculation time. © 2021 American Society of Civil Engineers. Language: en |
|
Keywords |
Forecasting; Data reduction; Pipeline corrosion; Pipelines; Support vector machines; Systems engineering; Predictive analytics; Dimensionality reduction; Multiobjective optimization; Turing machines |