基于KPCA-ALO-WLSSVM的埋地管道外腐蚀速率预测  

Prediction of external corrosion rate of buried pipeline based on KPCA-ALO-WLSSVM

作　　者：张新生[1] 张莹莹 ZHANG Xin-sheng;ZHANG Ying-ying(School of Management,Xi'an University of Architecture and Technology,Xi'an 710055,China)

机构地区：[1]西安建筑科技大学管理学院,西安710055

出　　处：《安全与环境学报》2022年第4期1804-1812,共9页Journal of Safety and Environment

基　　金：国家自然科学基金项目(41877527);陕西省社会科学基金项目(2018S34)。

摘　　要：为提高埋地油气管道外腐蚀速率预测精度,建立了一种基于KPCA-ALO-WLSSVM的埋地管道外腐蚀速率预测模型。以沿川气东送管线所做埋片试验获取的数据为例,首先利用核主成分分析(KPCA)对管道外腐蚀影响因素进行处理,以重构的综合指标作为模型的输入值;然后利用加权最小二乘支持向量机(WLSSVM)对外腐蚀因素和速率进行仿真建模,并利用蚁狮优化算法(ALO)对WLSSVM建模中的参数进行寻优。结果表明:KPCA提取了累计贡献率为97.84%的3个主元,减化了建模过程的复杂性;所构建的ALO-WLSSVM外腐蚀速率预测模型的平均相对误差为4.390%,均方根误差为0.276,各项指标均优于其对比模型,证明了本模型具有更好的学习性和更高的拟合效果。To improve the prediction accuracy of external corrosion rate of buried oil and gas pipelines, a weighted least squares support vector machine(WLSSVM) prediction model of external corrosion rate of buried pipeline based on kernel principal component analysis(KPCA) and ant lion optimization algorithm(ALO) was established. Firstly, KPCA was used to reduce the dimension of the index system of corrosion influencing factors, and the reconstructed comprehensive index was used as the input value of the model;Then, WLSSVM was used to simulate the complex functional relationship between external corrosion factors and corrosion rate of buried pipeline, and ALO was used to iteratively optimize the penalty factors and core parameters in the WLSSVM model to avoid the influence of randomness of parameter selection on model performance;Finally, taking the data obtained from the embedded slice test along the Sichuan East gas transmission pipeline as an example, the modeling and simulation were carried out by using MATLAB software to verify the performance of the model. The results showed that: KPCA extracted three principal components with a 97.84% cumulative contribution rate to reconstruct the original data, which reduced the interaction between various factors and the complexity of the modeling process;WLSSVM overcame the disadvantage of low robustness of LSSVM by assigning values to the errors in the modeling process of LSSVM, and the maximum relative error of the constructed KPCA-WLSSVM was 5.326% lower than that of the KPCA-LSSVM model. The average relative error of KPCA-ALO-WLSSVM model optimized by ALO was 4.390%, the root mean square error was 0.276, and the sum of squares of residual error was 0.458, which were better than the KPCA-PSO-WLSSVM model optimized by particle swarm optimization(PSO). It is proved that the model has better learning and better fitting effect, and provides a new way for the prediction of corrosion rate outside buried pipelines.

关 键 词：安全工程 埋地管道 外腐蚀速率 核主成分分析(KPCA) 蚁狮优化算法(ALO) 加权最小二乘支持向量机(WLSSVM) 

分 类 号：X937[环境科学与工程—安全科学]