Study On Defect Quantification Method Of Oil And Gas Pipeline Based On Magnetic Eddy Current

Accurate damage degree prediction is important to ensure the safety,stability operation of the oil and gas pipeline in service,especially those in long-term service.In order to solve the problem of quantification of oil and gas pipeline defects,a series of quantitative analysis methods for magnetic eddy current detection signals are studied in this dissertation.The main research contents are as follows:1.In order to obtain the core characteristics of magnetic eddy current defect signals,a low-dimensional feature space construction method was proposed,which was based on local decomposition smoothing pseudo-Wigner-Ville distribution.Compared with the traditional time-frequency analysis method,the MBCSC index of feature space has improved from 73% to 93%,which solve the problem of feature redundancy.The high-efficiency and low-dimensional dual-objective optimization of magnetic eddy current signal feature extraction is realized,which can be used as the data basis for the defect quantification model.2.Aiming at the complex relationship between defect depth and magnetic eddy current signal features,a classification-based defect depth quantification method is proposed.Defect depth is divided into 10 grades.An intelligent classification and recognition model based on PSO-ELM is established to achieve semi-quantification of defect depth.The results of single-point defect verification experiment and statistical performance test show that the quantitative accuracy of this method meets the requirements,which tolerance of ?10%T,reliability of 80%,and confidence of 95%.3.To solve the engineering optimization problem in the process of defect signal quantification prediction,an improved sparrow search algorithm based on reverse learning strategy and adaptive T distribution variation is proposed.In the numerical simulation experiment,the optimization algorithm has achieved the highest score in the three criteria indicators including optimal value,average value,and standard deviation.Experiments have verified the potential of its engineering optimization application,which can serve the intelligent quantitative parameter optimization problem of magnetic eddy current detection defect signals.4.Based on the defect magnetic eddy current data of natural gas stations,an intelligent quantitative prediction model for defects based on improved sparrow search algorithm and deep extreme learning machine is proposed.Compared with the DELM model,MAPE of the regression prediction of the improved model decreased from19.7% to 6.4%.The model can realize intelligent quantification of natural gas pipeline defects,provide decision-making basis for natural gas pipeline maintenance,and has certain engineering application significance for ensuring the safe and stable operation of pipelines.5.In order to eliminate the interference factors affecting the quantization accuracy,the saturated distortion signal and the overlapping defect cluster signal in the magnetic eddy current detection signal were reconstructed and restored.Verified by engineering examples,the quantization accuracy of GA-MOD reconstructed signals is improved by23.4%～29.4% compared with the original signals.It is proved that the GA-MOD method can improve the detection rate of defects and the quantitative accuracy of oil and gas pipeline defects.