Search

CN-121978200-A - Wavelet noise reduction and gradient feature extraction method for buried pipeline weld joint positioning

CN121978200ACN 121978200 ACN121978200 ACN 121978200ACN-121978200-A

Abstract

The invention discloses a wavelet noise reduction and gradient feature extraction method for buried pipeline weld joint positioning, and relates to the technical field of nondestructive testing; the method comprises the steps of firstly obtaining an original magnetic field signal of a pipeline area through a magnetic field acquisition device, then carrying out wavelet noise reduction treatment, carrying out wavelet transformation on the signal, carrying out wavelet reconstruction after noise coefficient is restrained through threshold treatment, obtaining a denoising signal, finally carrying out gradient feature extraction, calculating gradient of the denoising signal, standardizing, identifying extreme points in the standardized gradient through setting a threshold value as characteristic points of a welding seam, and therefore realizing accurate positioning of the welding seam position.

Inventors

  • SONG YAQIANG

Assignees

  • 江苏省特种设备安全监督检验研究院

Dates

Publication Date
20260505
Application Date
20260122

Claims (8)

  1. 1. The wavelet noise reduction and gradient feature extraction method for positioning the welded seam of the buried pipeline is characterized by comprising the following steps of: (1) Acquiring an original magnetic field intensity sequence acquired along the axis of the buried pipeline; (2) The adaptive noise reduction step of the wavelet domain comprises the steps of carrying out discrete wavelet transformation on the original magnetic field intensity sequence and decomposing the original magnetic field intensity sequence to different scales, determining adaptive threshold values of all scales based on the statistical characteristics of wavelet coefficients under the scales, carrying out nonlinear threshold processing on the wavelet coefficients by applying the adaptive threshold values to inhibit non-periodic noise components, carrying out inverse transformation on the processed wavelet coefficients, and reconstructing to obtain a magnetic field signal sequence after noise reduction; (3) Calculating a first-order differential gradient sequence of the magnetic field signal sequence after noise reduction, calculating a mean value and a standard deviation of the gradient sequence, and carrying out standardization processing on the gradient sequence based on the mean value and the standard deviation to obtain a standardized gradient sequence; (4) The method comprises the steps of determining the characteristic positions of welding seams, namely setting a characteristic determination threshold value, comparing the absolute value of each point in the standardized gradient sequence with the characteristic determination threshold value, determining the point with the absolute value exceeding the characteristic determination threshold value as the characteristic point of the welding seams, and determining the magnetic field characteristic signal positions of the butt welding seams of the buried pipelines according to the positions of the characteristic points of the welding seams.
  2. 2. The method of claim 1, wherein in the step of adaptively denoising in the wavelet domain, an adaptive threshold for each scale is determined based on a statistical characteristic of wavelet coefficients under the scale, and specifically using the following formula t_j=σ_j×sqrt (2×log (N)), where t_j is a threshold for the j-th scale, σ_j is a noise level estimate of the wavelet coefficients of the j-th scale, and N is a signal length.
  3. 3. The method according to claim 1 or 2, wherein in the wavelet domain adaptive noise reduction step, the nonlinear thresholding is a soft or hard threshold function.
  4. 4. The method according to claim 1, wherein in the gradient statistical feature extraction step, the specific formula of the normalization process is g_norm [ n ] = (G [ n ] - μ)/σ, where G [ n ] is an original gradient value, μ is a mean value of the gradient sequence, σ is a standard deviation of the gradient sequence, and g_norm [ n ] is a normalized gradient value.
  5. 5. The method according to claim 1, wherein the setting of the feature decision threshold is based on a statistical distribution of the normalized gradient sequence, selecting a predetermined multiple standard deviation as the threshold.
  6. 6. A buried pipeline butt weld positioning system for implementing the method of any one of claims 1-5, the system comprising: the magnetic sensor array is used for moving along the trend of the pipeline and collecting the original magnetic field intensity sequence; the signal processing unit is in communication connection with the magnetic sensor array and is configured to execute the wavelet domain self-adaptive noise reduction step, the gradient statistical feature extraction step and the weld feature position judgment step; And the positioning identification unit is used for marking the position on the surface above the pipeline according to the position of the welding seam characteristic signal output by the signal processing unit.
  7. 7. The system of claim 6, wherein the signal processing unit further comprises a visualization module for displaying the raw magnetic field strength sequence, the denoised magnetic field signal sequence, the normalized gradient sequence, and the marked locations of the weld feature points.
  8. 8. A computer readable storage medium, on which a computer program is stored, characterized in that the program, when being executed by a processor, implements the steps of the method according to any of claims 1-5.

Description

Wavelet noise reduction and gradient feature extraction method for buried pipeline weld joint positioning Technical Field The invention relates to the technical field of nondestructive testing of special equipment, in particular to a magnetic field signal filtering and characteristic extraction method for positioning butt welding seams of buried pipelines. Background At present, mature accurate positioning systems for butt welds of buried pipelines are lacking in the market. In the fields of similar metal magnetic memory detection and the like, a frequency domain filtering and frequency domain feature extraction method is generally adopted for signal processing. However, in the buried pipeline welding seam detection scene, the background interference signal and the welding seam characteristic signal often have the characteristics of non-periodicity and strong uncertainty, and the traditional frequency domain analysis method is difficult to effectively separate and extract. Therefore, the prior art is difficult to meet the requirements of accurate noise reduction and effective feature extraction on magnetic field signals in the process of positioning butt welds of buried pipelines, and a signal processing method capable of adapting to aperiodic and random interference is needed. Disclosure of Invention The invention aims to overcome the defects of the background technology, and provides a wavelet noise reduction and gradient feature extraction method for positioning a buried pipeline welding seam, which aims to solve the problem that aperiodic and uncertain interference signals and feature signals cannot be effectively processed in the prior art and realize accurate filtering and feature extraction of magnetic field signals in the process of positioning the buried pipeline butt welding seam. The technical scheme of the invention is as follows: the wavelet noise reduction and gradient feature extraction method for positioning a buried pipeline weld joint is realized by a magnetic field acquisition device and a data processing device, and comprises the following steps of: (1) And a magnetic field signal acquisition step, namely, using a magnetic field acquisition device to axially move along the pipeline to acquire an original magnetic field signal sequence data_In containing a welding seam area, wherein n is the number of sampling points. (2) And wavelet noise reduction processing, namely performing wavelet transformation on the original signal, suppressing a wavelet coefficient corresponding to noise through threshold processing, and performing wavelet inverse reconstruction to obtain a denoised signal sequence data_Out [ N ], wherein n=0, 1. (3) And the gradient feature extraction step is to calculate and normalize a gradient sequence of the denoised signal, set a feature extraction threshold value, and identify points with normalized gradient values exceeding the threshold value as feature points (abnormal points) so as to determine the accurate position of the welding seam differential magnetic field signal. Compared with the prior art, the invention has the beneficial effects that: excellent non-periodic noise suppression capability, namely, the signal and the noise can be adaptively separated in a time-frequency domain by adopting a wavelet noise reduction technology, and the method is particularly good at processing non-stationary and non-periodic noise which is difficult to be handled by traditional frequency domain filtering. The method has the advantages that the significance of the magnetic field mutation characteristics at the weld joint is enhanced through gradient transformation and standardization processing, and the characteristic points can be accurately positioned by combining threshold judgment, so that a reliable basis is provided for the physical positioning of the subsequent weld joint. The algorithm has strong robustness, the method has clear flow, parameters (such as wavelet base, noise reduction threshold and characteristic extraction threshold) can be optimized according to the statistical characteristics of actual signals, and the adaptability is wide. Detailed Description The present invention will be described in detail with reference to examples. The system comprises the following components: The magnetic field acquisition device is used for acquiring a space magnetic field signal above the buried pipeline, and is usually a magnetic sensor array or a single-point scanning sensor. The data processing device is connected with the magnetic field acquisition device in a wired or wireless way, is internally embedded with a software program for running the algorithm of the invention, and is responsible for finishing the noise reduction, the feature extraction and the result display of the signals. The core algorithm flow of the method comprises two stages: Stage one wavelet noise reduction Wavelet decomposition, in which discrete wavelet transformation is carried out on the original ma