摘要
本文在室内实验数据库构建的基础上,选用支持向量机递归特征消除算法(SVM-RFE)对积液的影响因素进行量化排序,并采用粒子群优化算法(PSO)对SVM-RFE算法的惩罚因子c和核函数参数g寻优过程进行优化。结果显示算法预测得到的重要度排名为:气相折算速度、液相折算速度、气相黏度、液相密度、温度、管径、倾角、界面张力、压力、液相黏度、气相密度。对重要度排序前六的参数进行了敏感性分析,结果表明:气相折算速度越大、液相折算速度越小、气相黏度越小、液体密度越小、管径越大,积液越严重。本文建立的积液影响因素重要度排名模型具有预测精度高、稳定性强的优点,可为输气管线积液风险识别及管控提供理论支撑。
Based on the construction of the laboratory experimental database,the support vector machine recursive feature elimination algorithm(SVM-RFE)is used to quantify the influencing factors of fluid accumulation,and the particle swarm optimization algorithm(PSO)is used to optimize the optimization process of the penalty factor c and kernel function parameter g of the SVM-RFE algorithm.The results show that the importance ranking sequences predicted by the algorithm are as follows:gas phase conversion speed,liquid phase conversion speed,gas phase viscosity,liquid phase density,temperature,pipe diameter,inclination angle,interfacial tension,pressure,liquid phase viscosity and gas phase density.The sensitivity analysis of the top 6 parameters in the importance ranking was carried out,and the results showed that the larger the gas phase conversion speed,the smaller the liquid phase conversion speed,the smaller the gas phase viscosity,the smaller the liquid density,and the larger the pipe diameter,the more serious the liquid accumulation.The importance ranking model of the influencing factors of fluid accumulation established in this paper has the advantages of high prediction accuracy and strong stability,which can provide theoretical support for the identification and control of fluid accumulation risk in gas pipelines.
作者
郭永杰
洪祥
杨绍军
Guo Yongjie;Hong Xiang;Yang Shaojun(China Petrochemical Corporation,Dazhou,Sichuan 635002)
出处
《西部特种设备》
2025年第4期42-50,共9页
Western Special Equipment
