Quantification method of tubing defects based on machine learning algorithm and magnetic flux leakage signals.

Tubing is the pipeline that transports crude oil and natural gas from the oil and gas layer to the surface of the earth. Due to the harsh operating environment, the tubing will suffer from etch pits, scratches, cracks, perforations, and even direct fractures of different degrees of defective conditions. If tubing defects are not detected and quantified in a timely manner, the continued use of tubing will result in tubing leakage and failure. Magnetic flux leakage (MFL) testing as a nondestructive testing method enables the identification and quantitative analysis of defects in metal tubing. To improve the quantification accuracy of defects in the wellhead MFL testing of tubing defects during workover operations, this paper proposes a multi-output least-squares support vector regression machine (MLSSVR) model optimized based on the simulated annealing algorithm. The size of tubing defects can be quantified by establishing the mapping between the characteristic quantity of MFL signals and the defect size. The experimental results of MFL testing of tubing defects show that the root mean square error (RMSE) of the diameter of tubing defects of the simulated annealing algorithm optimized multi-output least-squares support vector regression (SA-MLSSVR) machine model proposed in this paper is 0.4562 mm, and the RMSE of the depth of tubing defects is 0.1504 mm. Compared with the non-optimized MLSSVR model, the overall RMSE of tubing defects is reduced by 36.48%. The SA-MLSSVR model only needs one-ninth of the time to achieve the same quantification accuracy as the particle swarm optimized multi-output least-squares support vector regression machine model.

Signal extraction using complementary ensemble empirical mode in pipeline magnetic flux leakage nondestructive evaluation.

The magnetic flux leakage (MFL) evaluation is often used for the overhauling of oil extracting operation in the oil field to realize the real-time damage assessment of the pipeline. Since the MFL signal is affected by various noise sources in the field, this paper introduces the complementary ensemble empirical mode decomposition with adaptive noise (CEEMDAN). On the basis of this, a particle swarm optimization wavelet threshold (PSO-WT) method is proposed, and the signal reconstruction option is improved to extract the leakage magnetic flux signal of tubing defects. First, CEEMDAN is used to add pairs of positive and negative white noise to the MFL signal, and then the signal is decomposed into several intrinsic mode functions (IMFs). Second, the correlation coefficient selection limit is defined. Taking into account the characteristics of the decomposed signal, the useless IMFs and useful IMFs are selected from the IMF components, where some of the useful IMF components contain less noise. Third, the PSO-WT algorithm is combined to further filter the noisy and useful IMF components. Finally, the filtered IMF components and the pure useful IMF components are selected to reconstruct the signal. In the experiment, the ensemble empirical mode decomposition (EEMD) method and CEEMDAN are used to decompose the noisy MFL signals ensemble in the field. The MFL signal is reconstructed under the correlation coefficient selection. It can be seen from the comparison of EEMD that the MFL signal is reconstructed under the same conditions after CEEMDAN decomposition, and its signal-to-noise ratio is increased by 8%. At the same time, after CEEMDAN decomposition, the selected noisy useful IMFs are further filtered by the wavelet threshold (WT) method and the PSO-WT method. Also, it indicates that the reconstructed signal processed by PSO-WT is 17% higher than the reconstructed signal after WT processing.