Missing interpolation method for submarine pipeline magnetic flux leakage data based on knn-svr

Abstract

The KNN-SVR-based missing interpolation method for submarine pipeline magnetic flux leakage data of the present invention includes step 1: segment characteristic data blocks from the original magnetic flux leakage data without missing points to form a complete data set, and construct a KD tree of the complete data set ; Step 2: Normalize the magnetic flux leakage data containing missing points, obtain a data set to be interpolated consisting of data blocks to be interpolated, and perform zero padding on the data set to be interpolated; Step 3: Search in the complete data set K nearest neighbors of the data block to be interpolated, and K complete data blocks are obtained; Step 4: Based on the K complete data blocks, construct a training set and normalize the training set; Step 5: Use the support vector regression machine to Carry out training; Step 6: Predict the missing feature values ​​in the data block to be interpolated. The present invention combines the KNN algorithm based on Euclidean distance with the SVR algorithm, which improves the prediction accuracy, reduces the over-fitting problem, and has good robustness to signal noise.

Description

technical field [0001] The invention relates to the fields of data processing and artificial intelligence, in particular to a KNN-SVR-based interpolation method for missing magnetic flux leakage data of submarine pipelines. Background technique [0002] Usually, the working conditions of the oil pipeline are very harsh, and it is easy to cause damage such as corrosion cracks, so it is necessary to carry out non-destructive testing on it. Magnetic flux leakage detection is a commonly used non-destructive testing method, which is to judge the damage of the pipeline through comprehensive analysis of the magnetic flux leakage data collected by the internal detector. During the operation of the internal detector, it is inevitable that due to the abnormal behavior of the sensor, some sampled data will be abnormal or missing. Preprocessing the data before analyzing the magnetic flux leakage data, an important part of which is to interpolate the missing data to ensure the integrity...

AI Technical Summary

Problems solved by technology

[0006] The embodiment of the present invention provides a KNN-SVR-based interpolation method for missing magnetic flux leakage data of submarine pipelines, which solves the problems of long calculation time, a large number of overfitting, and complex data missing points in the existing anomaly judgment and filling methods.

Images