基于支持向量机的带包覆层油气管道剩余壁厚反演研究

油田现场的油气管道外壁通常带有一定厚度的包覆层,脉冲涡流检测技术可实现不拆包覆层条件下的壁厚检测,可节约大量的包覆层拆装成本。但包覆层的存在使得脉冲涡流检测传感器与管道之间存在提离距离,导致传感器的等效电阻和电感发生变化,进而对脉冲涡流检测电压信号产生影响,降低剩余壁厚反演的精度。基于此,采用支持向量机方法(Support Vector Machine,SVM)建立了提离距离、检测电压信号晚期段斜率与管道壁厚之间的定量化反演模型。利用该模型对壁厚范围5~21 mm的阶梯试件开展了0~10 cm提离距离条件下的剩余壁厚反演测试,反演结果的误差控制在7%以内,验证了所得反演模型的有效性,并具有较高的实用价值,可为带包覆层的管道剩余壁厚检测提供参考。

Research on residual wall thickness inversion of oil and gas pipelinewith cladding based on support vector machine

The oil and gas pipelines in the oil field are usually equipped with a layer of cladding in certain thickness,and the pulsed eddy current testing technology can be adopted to measure the wall thickness without removing the cladding,which saves a lot of costs for the removal and assembly of the cladding.However,as a result of cladding,distance exists between the pulsed eddy current sensor and the pipeline,which leads to the variation of the equivalent resistance and inductance of the sensor,and further affects the voltage signal of the pulsed eddy current detection,reducing the accuracy of the residual wall thickness inversion.In order to solve this problem,a quantitative inversion model of lift-off,slope of late detection voltage signal and pipe wall thickness was established with a support vector machine(SVM),and inversion test of the residual wall thickness of the step specimen with the wall thickness of 5-21 mm was carried out with this model under the condition of 0-10 cm lift-off,with the error of the inversion result controlled within 7%.Thereby,the inversion model is verified to be valid with high practical value,and certain reference could be provided for the measurement of the residual wall thickness of the pipeline with cladding.