摘要
针对储气库注采管柱的内腐蚀速率预测问题,建立了基于阿基米德优化算法(Archimedes Optimization Algorithm,AOA)与核极限学习机(Kernel Extreme Learning Machine,KELM)相结合的模型提高腐蚀速率预测精度。通过引入佳点集、改进密度降低因子、采用黄金正弦算法缩小搜索空间,提高局部开发能力,利用改进阿基米德优化算法(Improved Archimedes Optimization Algorithm,IAOA)优化KELM正则化系数(C)和核函数参数(γ),进而建立IAOA-KELM储气库注采管柱内腐蚀速率预测模型;使用MATLAB软件运用该模型对某注采管柱内腐蚀数据集进行学习与预测,将IAOA-KELM模型与KELM、粒子群优化算法(Particle Swarm Optimization,PSO)-KELM、AOA-KELM结果进行预测误差对比。结果表明,IAOA-KELM模型的预测值与实际值较为拟合,其E RMSE为0.65%,E MAE为0.39%,R 2为99.83%,均优于其他模型。研究表明,IAOA-KELM模型能够更为准确地预测储气库注采管柱内腐蚀速率,为储气库注采管柱的运维及储气库的健康管理提供参考。
The underground corrosion environment of gas storage is complicated.To prevent corrosion failure of pipe string,this study established a model based on the combination of Archimedes Optimization Algorithm(AOA)and Nuclear Extreme Learning Machine(KELM)to predict the corrosion rate of gas storage injection and production pipe string.Firstly,this study constructed an Improved Archimedes Optimization Algorithm(IAOA).In AOA,individuals are initialized with random position information,and the good-point set is applied to the initial stage of AOA to enhance the global exploration capability of the algorithm.In addition,by improving the density reduction factor and integrating the golden sine algorithm with the local search phase position update,the algorithm can avoid falling into the local optimal.After analysis,the time complexity of IAOA is the same as that of AOA,indicating that the improvement strategy does not reduce the operation efficiency.Secondly,this study used IAOA to optimize the KELM regularization coefficient(C)and kernel function parameter(γ)and then established the IAOA-KELM corrosion rate prediction model.Finally,IAOA-KELM was used to learn and predict 150 sets of corrosion data in a gas storage injection and production pipe string with MATLAB software.One hundred thirty sets of data were randomly selected as training sets for the model to learn,and the remaining 20 groups were tested on the model training results.The error comparison between the prediction results of the IAOA-KELM model and those of KELM,Particle Swarm Optimization(PSO)-KELM,and AOA-KELM was carried out.The results show that the E RMSE,E MAE,and R 2 of the IAOA-KELM model are 0.65%,0.39%,and 99.83%,indicating that the prediction accuracy is better than other models.It is proved that the IAOA-KELM model can predict the corrosion rate more accurately and provide a reference for the operation and maintenance of injection-production pipe string and the health management of gas storage.
作者
骆正山
于瑶如
骆济豪
王小完
LUO Zhengshan;YU Yaoru;LUO Jihao;WANG Xiaowan(School of Management,Xi'an University of Architecture and Technology,Xi'an 710055,China;Ruixin Institute of Beijing Institute of Technology,Beijing 102488,China)
出处
《安全与环境学报》
CAS
CSCD
北大核心
2024年第3期971-977,共7页
Journal of Safety and Environment
基金
国家自然科学基金项目(41877527)
陕西省教育厅自然专项基金项目(2018S34)。
关键词
安全工程
地下储气库
注采管柱
核极限学习机
改进阿基米德优化算法
腐蚀速率
safety engineering
underground gas storage
injection and production string
Kernel Extreme Learning Machine(KELM)
Improved Archimedes Optimization Algorithm(IAOA)
corrosion rate
