改进SSA-LSSVM模型在埋地管道点蚀深度预测中的应用  

Application of improved SSA-LSSVM model in prediction of pitting depth of buried pipelines

作　　者：骆正山[1] 徐龙寅 骆济豪 王小完[1] LUO Zhengshan;XU Longyin;LUO Jihao;WANG Xiaowan(School of Management,Xi'an University of Architecture and Technology,Shaanxi 710055,China;Ruixin Institute of Beijing Institute of Technology,Beijing 102488,China)

机构地区：[1]西安建筑科技大学管理学院,西安710055 [2]北京理工大学睿信学院,北京102488

出　　处：《安全与环境学报》2023年第9期3115-3122,共8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：埋地管道点蚀深度受土壤环境、运输物质、管道材质等多种因素的影响,因此腐蚀数据存在不稳定性,会导致精确预测其点蚀深度存在较大难度,故提出RS结合MSSA-LSSVM预测模型。首先利用RS对腐蚀影响因素实现降维,提取关键影响因素;其次融合三步改进策略解决麻雀搜索算法已陷入局部最优等问题,利用时间复杂度分析对算法改进后性能进行验证;然后利用MSSA求解出LSSVM中核函数参数σ2和惩罚因子C的最优解,同时选取RBF核函数,使其预测性能达到最优,最终构建RS-MSSA-LSSVM的埋地管道点蚀深度预测模型。结果表明:优化后模型精度得到了极大的提升,且均优于其他模型,证明该模型鲁棒性较好。To accurately predict the pitting depth of buried pipelines,it is proposed to combine RS with MSSA-LSSVM to construct a prediction model for the pitting depth of buried pipelines.The pitting corrosion of buried pipelines is affected by various factors such as soil environment,transportation materials,and pipeline materials,which can easily lead to data instability.Therefore,it is necessary to eliminate the influencing factors with a low contribution rate and reconstruct the input sample data.Firstly,the rough set is used to reduce the dimension of the main influencing factors affecting pipeline corrosion,and reduce the influence of redundant factors on the prediction accuracy.Secondly,to solve the problem that the sparrow search algorithm has fallen into local optimum and has slow convergence,a three-step improvementst trategy is proposed.In the first step,the population initialization introduces the Cubic chaotic mapping strategy to enhance its ergodicity.In the second step,a sine and cosine search strategy is introduced,that is,after the position of the follower is updated,the optimal feasible solution is obtained by using the sine and cosine search.The third step is to introduce the Cauchy Gaussian mutation strategy in the later stage of the algorithm to perturb the current optimal individual.Thirdly,the improved sparrow search algorithm is used to optimize the parameters of LSSVM.Finally,the MSSA-LSSSVM buried pipeline pitting depth prediction model is established.Combining the actual pitting depth data of Tarim onshore oil and gas pipelines for verification,the results show that the root mean square error of the model(E_(RMSE))is 0.0567,the average absolute eror value(E_(MAE))is 0.0212,the Hill inequality coefficient(E_(TheilIC))is 0.00356 and the determination coefficient(R^(2))is 0.999464,and the performance is greatly improved.It is proven that the model has good robustness,can accurately predict the development trend of pitting corrosion of buried pipelines,and provide a reference for safe pipeline opera

关 键 词：安全工程 埋地管道 点蚀深度 粗糙集(RS) 改进麻雀搜索算法(MSSA) 最小二乘支持向量机(LSSVM) 

分 类 号：X937[环境科学与工程—安全科学]