A pipeline anomaly classification method based on middle-level features

Abstract

The invention provides a pipeline anomaly classification method based on middle-level features, which relates to the field of pipeline nondestructive detection and data mining. The method comprises the following steps of 1 preprocessing the pipeline magnetic flux leakage signal; 2 acquiring a magnetic flux leakage data sample set; 3 generating dominant characteristics of magnetic flux leakage datasample; 4 generating that middle layer characteristic of the magnetic flux leakage data sample based on the BOW model, calculating the difference characteristic vector for each sample circle frame; classifying the clustering difference eigenvector set by K-Means clustering to get k visual words; counting the frequency of each visual word in each sample, and obtaining the middle feature vector ofeach sample; 5 based on a feature fusion algorithm of joint sparse representation, fusing the dominant feature and middle feature of samples to form a joint sparse matrix and classifying the samples by using classification discriminant function. The method of the invention solves the technical problems of low accuracy of feature extraction and classification of pipeline abnormality and undesirabledetection effect in the prior art.

Description

technical field [0001] The invention relates to the fields of pipeline non-destructive testing and data mining, in particular to a pipeline anomaly classification method based on mid-level features. Background technique [0002] With the rapid development of the global economy, the demand for energy is increasing. As the safest and most efficient mode of transportation, pipeline transportation is known as the "artery" of the world economy. Pipelines are buried underground. Due to the harsh environment and various human factors, the pipelines are prone to wear, corrosion and leakage, resulting in serious economic losses. Non-destructive testing methods are generally used to detect the surface of oil pipelines. Commonly used non-destructive testing methods for pipelines include: ultrasonic testing, eddy current testing and magnetic flux leakage testing. Among them, the magnetic flux leakage detection technology is the most mature. Magnetic flux leakage detection has low env...

AI Technical Summary

Problems solved by technology

[0005] Aiming at the technical problem that the accuracy of pipeline abnormality classification is low and the detection effect is not ideal due to the low accuracy of pipeline magnetic flux leakage signal feature extraction existing in the above-mentioned existing pipeline abnormality classification method, the present invention provides a pipeline abnormality detection method based on middle-level features The classification method can enhance the recognition ability and save manpower and material resources

Images