Inversion method of pipeline defect depth based on convolution neural network

Abstract

The invention provides an inversion method of a pipeline defect depth based on a convolution neural network. The method comprises the following steps: 1, randomly generating a pipeline defect profile: generating an n group pipeline defect depth matrix, and obtaining the axial magnetic flux leakage signal of the n group pipeline defect profile based on magnetic dipole model simulation; 2, obtaining a practically measured k group pipeline defect depth matrix, and measuring the corresponding axial magnetic flux leakage signal by using a magnetic field sensor; 3, constructing a convolution neural network model, and training the convolution neural network model by using the axial magnetic flux leakage signal of the simulated pipeline defect profile and the measured axial magnetic flux leakage signal of the practically measured pipeline defect profile to obtain a final convolution neural network model; and 4, preprocessing the axial magnetic flux leakage signal of a pipeline defect with an unknown depth, and inputting the final convolution neural network model in order to obtain the depth predication value of the pipeline defect with the unknown depth. The method effectively reduces the parameters needed by network training, and shortens the defect inversion time.

Description

technical field [0001] The invention belongs to the technical field of fault diagnosis and artificial intelligence, and in particular relates to a pipeline defect depth inversion method based on a convolutional neural network. Background technique [0002] In economic construction, pipeline transportation plays a very critical role. Because pipelines often work in harsh environments, the pipe wall will become thinner due to factors such as corrosion and external forces. Such defective pipelines can easily lead to leakage accidents. [0003] Magnetic flux leakage detection technology is currently one of the most effective pipeline defect detection methods at home and abroad. Due to its good reliability, high stability and fast detection speed, this technology is more and more used in detection of pipeline defects. Defect identification is an important part of the pipeline magnetic flux leakage detection system. Only by correct identification can the owner be provided with ...

AI Technical Summary

Problems solved by technology

[0009] The direct inversion algorithm has certain limitations: ① A large amount of sample data is required to establish the mapping relationship between the magnetic flux leakage signal features and the defect geometric size. If the number of samples is insufficient, the inversion results will have large errors; Accurate feature extraction algorithm is used as the basis. If the characteristics of the defect magnetic flux leakage signal cannot be accurately extracted, the relationship between the magnetic flux leakage signal characteristics and the geometric size of the defect cannot be accurately established; ③The inversion result is affected by signal noise and signal distortion larger

However, the iterative inversion method still has the following disadvantages: ① It takes a long time to calculate; ② It is difficult to guarantee the accuracy of the forward modeling model; ③ The defect size optimization algorithm is easy to fall into the local optimal solution; ④ The inversion results are affected by signal noise and signal The effect of distortion is large

Images