A pipeline magnetic flux leakage defect detection method based on extreme learning machine

Abstract

The invention relates to a method for detecting magnetic flux leakage defects in pipelines based on an extreme learning machine. It uses the length, width, depth data of the known magnetic leakage defects in the pipeline and the waveform characteristic values ​​of the magnetic flux leakage signals to establish an extreme learning machine model and train the parameters in the sample data. The length, width, and depth data of the pipeline magnetic leakage defect are known as the input of the model. The trial and error method is used to select the number of hidden layer nodes, and the hidden layer output matrix and output weight are calculated. The magnetic flux leakage signal waveform characteristic value is used as the model. The output of the model. When magnetic flux leakage occurs in the pipeline, the magnetic flux leakage signal waveform of the unknown magnetic flux leakage defect shape is obtained, and the extreme learning machine model is used to detect the magnetic flux leakage defect in the pipeline. This invention uses an extreme learning machine model to perform intelligent inversion of pipeline defect shapes, which has the advantages of fast learning speed and good generalization performance. It can quickly and accurately construct the shape of the defect using the detected defect waveform, thereby knowing the severity of the defect. It can predict pipeline risks and prevent pipeline leakage.

Description

technical field [0001] The invention belongs to the technical field of pipeline detection, and in particular relates to a pipeline magnetic flux leakage defect detection method based on an extreme learning machine. Background technique [0002] In the field of pipeline defect research, there is a large amount of pipeline defect shape data, which requires fast detection speed. For these difficulties, the finite element algorithm is widely used at present. The finite element algorithm is a high-efficiency and commonly used calculation method. It can discretize the continuous discretization into a collection of several finite-sized unit bodies, which can be applied to any differential equation. in various physical fields. However, the finite element algorithm faces the characteristics of a large amount of data, the speed is relatively slow, and the time consumption is long, which cannot meet the requirements. In addition, there are support vector machines. Support vector mach...

AI Technical Summary

Problems solved by technology

Traditional neural network learning algorithms (such as BP algorithm) need to artificially set a large number of network training parameters, and it is easy to generate local optimal solutions

Images