针对气田集输管道的腐蚀问题,提出了一种基于GRA-IFA-LSSVM组合模型的内腐蚀速率预测算法。对GRA (灰色关联分析)模型、IFA (改进萤火虫)模型以及LSSVM (最小二乘支持向量机)模型理论进行了介绍,提出了组合模型的组合流程以及组合模型的评价指标;以我国某气田集输管道为例,对GRA-IFA-LSSVM组合模型的预测精度进行验证,同时,将其预测精度与其他常见预测模型的精度进行了对比。结果表明:温度、H₂S含量、CO₂含量、pH以及流速属于影响气田集输管道腐蚀的重要因素;使用GRA-IFA-LSSVM组合模型对气田集输管道内腐蚀速率进行预测时,其平均绝对误差为1.946%,均方根误差为1.496%,可决系数为97.53%,该组合模型的三项评价指标均小于其他常见预测模型。GRA-IFA-LSSVM组合模型对气田集输管道进行内腐蚀速率预测具有很强的准确性、鲁棒性及先进性,可以为气田集输管道的保护提供数据支持。

Aiming at the corrosion problem of gas field gathering pipelines, a prediction algorithm of internal corrosion rate based on the GRA-IFA-LSSVM combined model was proposed. The theories of GRA (Gray Relational Analysis) model, IFA (Improved Firefly) model and LSSVM (Least Squares Support Vector Machine) model were introduced, and the combined process and evaluation indexes of the combined model were proposed. The prediction accuracy of the GRA-IFA-LSSVM combined model was verified, taking a domestic pipeline as an example, and was compared with those of the other common prediction models. The results show that temperature, H₂S content, CO₂ content, pH value and flow rate were important factors affecting the corrosion of gas field gathering pipelines. When the GRA-IFA-LSSVM combined model was used to predict the internal corrosion rate of gas field gathering pipelines, the average absolute error was 1.946%, the root mean square error was 1.496%, and the absolute coefficient was 97.53%. The three evaluation indexes of the combined model were all smaller than those of the other common prediction models. The GRA-IFA-LSSVM combined model had strong accuracy, robustness and advancement in the prediction of internal corrosion rate of gas field gathering pipelines, and could provide data support for the protection of gas field gathering pipelines.