Pipe fitting defect detection method and system

Abstract

The invention provides a pipe fitting defect detection method and system. The method comprises the following steps: step A1, collecting an eddy current signal generated by a pipe fitting after a flaw detector sends out a detection signal; a2, performing feature extraction on the eddy current signal to obtain feature data; step A3, processing the feature data based on a pre-trained defect identification model to perform defect identification of the pipe fitting to obtain a defect identification result, the defect identification result comprising defects and defect size data; and step A4, performing grade classification on the defects based on the size data of the defects to obtain a defect grade classification result. Through artificial intelligence, the identification precision of the pipe fitting is improved, the identification efficiency is improved, quantitative and qualitative analysis is beneficial to subsequent decision making, and the method has important significance for ensuring the quality of the high-energy-efficiency internal-thread copper pipe and preventing safety accidents.

Description

Technical Field

[0001] The present invention relates to the technical field of non-destructive testing of metal pipelines, and in particular to a method and system for detecting defects in pipe fittings. Background Art

[0002] Eddy current testing, a widely used nondestructive testing method, leverages the principles of electromagnetic induction to identify and evaluate surface and near-surface defects in metal materials. This method analyzes eddy current signals generated within the material being tested to infer the presence and nature of defects. In practical applications, eddy current testing can effectively identify defects such as cracks, corrosion, and wear, playing an irreplaceable role in ensuring product quality and safety.

[0003] A key step in eddy current testing technology is the preliminary analysis of the collected eddy current signal. Defects are identified by meticulously analyzing the eddy current signal's amplitude, phase, waveform, and other characteristics, combined with a library of known standard defect samples. Larger surface defects often cause significant changes in the eddy current signal because they alter the eddy current flow path, thereby affecting the impedance of the detection coil. In contrast, smaller defects may only cause minor changes in the signal. Existing criteria, for example, define a small defect as a minor flaw if the filtered eddy current signal amplitude is ≤40%, resulting in a smaller amplitude and narrower waveform, while a large defect as a major flaw with a larger amplitude and wider waveform if the amplitude is ≥60%. This analysis method is inefficient, fails to accurately determine the defect's location and quantitative size, and fails to provide users with intuitive defect observation. Furthermore, it requires the testing equipment to possess high sensitivity and precision to capture these subtle differences. Accurately locating and quantitatively assessing surface flaws remains an unresolved issue. Summary of the Invention

[0004] Based on the above problems, the present invention provides a pipe defect detection method and system, aiming to solve the technical problems of low accuracy and insufficient quantification of pipe defect detection in the prior art.

[0005] The present invention provides a pipe defect detection method, comprising:

[0006] Step A1, collecting the eddy current signal generated by the pipe after the flaw detector sends a detection signal;

[0007] Step A2, extracting features from the eddy current signal to obtain feature data;

[0008] Step A3: Processing the feature data based on the pre-trained defect recognition model to perform defect recognition on the pipe fitting to obtain a defect recognition result, which includes the defect and its size data;

[0009] Step A4: classify the defects based on their size data to obtain defect classification results.

[0010] Furthermore, in step A3, a defect probability estimation value of each inspection point in the pipe is identified from the feature data based on the defect recognition model;

[0011] In step A4, a defect identification map of the pipe fitting is formed based on the defect probability estimation value of each detection point and the preset color coding mapping rule.

[0012] Furthermore, in step A3,

[0013] The defect identification results also include the location data of the defect;

[0014] In step A4, the defect location data is corrected, and then a defect identification map of the pipe is formed based on the corrected defect location data, the defect probability estimation value of each detection point and the preset color coding mapping rule.

[0015] Furthermore, the defect size data includes the defect area;

[0016] In step A4, based on the defect size data, the defects are classified according to a preset classification rule to obtain a defect classification result;

[0017] The preset classification rules include:

[0018] When the defect area is not less than the first area threshold and less than the second area threshold, the defect is determined to be a minor damage level;

[0019] When the defect area is not less than the second area threshold and less than the third area threshold, the defect is determined to be of a moderate level;

[0020] When the defect area is not less than the third area threshold, the defect is determined to be a major damage level.

[0021] Furthermore, the defect size data includes defect depth;

[0022] In step A4, based on the defect size data, the defects are classified according to a preset classification rule to obtain a defect classification result;

[0023] The preset classification rules include:

[0024] When the defect depth is not less than the first depth threshold and less than the second depth threshold, the defect is determined to be a minor damage level;

[0025] When the defect depth is not less than the second depth threshold and less than the third depth threshold, the defect is determined to be of a moderate level;

[0026] When the defect depth is not less than the third depth threshold, the defect is determined to be a major damage level.

[0027] Furthermore, in step A2, before extracting features from the eddy current signal, the eddy current signal is subjected to signal enhancement processing based on a wavelet transform algorithm and / or temperature compensation processing based on a temperature compensation algorithm.

[0028] Furthermore, in step A1, eddy current signals corresponding to the pipe fittings under the detection signals of each preset frequency emitted by the flaw detector are respectively collected;

[0029] In step A2, before extracting the features of the eddy current signal, the eddy current signal collected under the detection signal of each preset frequency is subjected to multi-frequency signal fusion processing in the frequency domain to obtain a fused eddy current signal, and then the feature extraction of the fused eddy current signal is performed to obtain feature data.

[0030] Furthermore, in step A1, eddy current signals corresponding to the pipe fittings under the detection signals of each preset frequency emitted by the flaw detector are respectively collected;

[0031] In step A2, feature extraction is performed on the eddy current signals collected under the detection signals of each preset frequency in the frequency domain, and the extracted feature data are fused to obtain final feature data.

[0032] Another aspect of the present invention provides a pipe defect detection system for executing the aforementioned pipe defect detection method, comprising:

[0033] The signal acquisition module is used to collect the eddy current signal generated by the pipe after the flaw detector sends out the detection signal;

[0034] The feature extraction module is connected to the signal acquisition module and is used to extract the features of the eddy current signal to obtain feature data;

[0035] A defect recognition module, connected to the feature extraction module, is used to process feature data based on a pre-trained defect recognition model to perform defect recognition on the pipe fitting and obtain defect recognition results, which include defect and defect size data;

[0036] The defect classification module is connected to the defect recognition module and is used to classify the defects based on their size data to obtain defect grade classification results.

[0037] Furthermore, the defect recognition module identifies the defect probability estimation value of each inspection point in the pipe from the feature data based on the defect recognition model;

[0038] The pipe defect detection system also includes:

[0039] The defect identification module is connected to the defect recognition module and is used to form a defect identification map of the pipe fitting based on the defect probability estimation value of each detection point and the preset color coding mapping rule.

[0040] The beneficial technical effect of the present invention is that: the present invention uses a defect recognition algorithm constructed based on artificial intelligence to identify defects on features extracted from eddy current signals, effectively identifies defects, and obtains quantitative information such as defect size data, distinguishes the severity of the identified defects according to the size, improves the recognition accuracy of pipe fittings through artificial intelligence, and improves the recognition efficiency. Quantitative and qualitative analysis helps in subsequent decision-making, which is of great significance for ensuring the quality of high-efficiency internal threaded copper pipes and preventing safety accidents. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] Figure 1 This is a flow chart of the steps of a pipe defect detection method of the present invention;

[0042] Figure 2 This is a module schematic diagram of a pipe defect detection system of the present invention. DETAILED DESCRIPTION

[0043] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making any creative efforts shall fall within the scope of protection of the present invention.

[0044] It should be noted that, in the absence of conflict, the embodiments of the present invention and the features in the embodiments may be combined with each other.

[0045] The present invention will be further described below with reference to the accompanying drawings and specific embodiments, but they are not intended to limit the present invention.

[0046] See also Figure 1 The present invention provides a pipe defect detection method, comprising:

[0047] Step A1, collecting the eddy current signal generated by the pipe after the flaw detector sends a detection signal;

[0048] Step A2, extracting features from the eddy current signal to obtain feature data;

[0049] Step A3: Processing the feature data based on a pre-trained defect recognition model to perform defect recognition on the pipe fitting, and obtaining a defect recognition result, which includes the defect and its size data;

[0050] Step A4: classify the defects based on their size data to obtain defect classification results.

[0051] Specifically, the pipe is made of metal, preferably a copper pipe, and preferably an internally threaded pipe.

[0052] Specifically, the defect recognition model is constructed based on a support vector machine (SVM) algorithm or a lightweight neural network.

[0053] The present invention uses a defect recognition algorithm based on artificial intelligence to identify defects based on features extracted from eddy current signals, effectively identifying defects and obtaining quantitative information such as defect size data. The severity of the identified defects is differentiated according to the size. Artificial intelligence is used to improve the recognition accuracy of pipe fittings and the recognition efficiency. Quantitative and qualitative analysis helps in subsequent decision-making, which is of great significance for ensuring the quality of high-efficiency pipe fittings and preventing safety accidents.

[0054] Furthermore, in step A3, a defect probability estimation value of each inspection point in the pipe is identified from the feature data based on the defect recognition model;

[0055] In step A4, a defect identification map of the pipe fitting is formed based on the defect probability estimation value of each detection point and the preset color coding mapping rule.

[0056] The present invention uses artificial intelligence algorithms to not only quickly and accurately identify defects and defect quantitative information, but also to create defect identification maps based on the defect identification results, realize the function of distinguishing and displaying each defect, and provide users with an intuitive defect experience.

[0057] Specifically, different colors can be used to represent different areas in the defect identification map. For example, green can represent defect-free or qualified areas, yellow can warn of areas with minor scratches or flaws, and red can clearly indicate areas identified as severely defective. This allows operators to quickly identify key issues that require priority attention and implement corresponding quality control improvements in relevant processes for different levels of defects.

[0058] Specifically, the preset color coding mapping rules are as follows:

[0059]

[0060] Where P(defect) represents the estimated probability of defect;

[0061] #00FF00 represents green; #FFFF00 represents yellow; and #FF0000 represents red. This is a color representation in the hexadecimal color coding system.

[0062] Furthermore, in step A3

[0063] The defect identification results also include the location data of the defect;

[0064] In step A4, the defect location data is corrected, and then a defect identification map of the pipe is formed based on the corrected defect location data, the defect probability estimation value of each detection point and the preset color coding mapping rule.

[0065] More accurate defect location information can be obtained through defect location correction.

[0066] Specifically, the correction method for the defect position data is to perform coordinate offset correction based on the gradient direction of the eddy current signal. The calculation formula is as follows:

[0067] (x',y')=AffineTransform(x,y,θ)

[0068] Among them, Affine Transformation is affine transformation;

[0069]

[0070] Here, x', y' represent the coordinates after the defect is corrected, and x, y represent the coordinates before the defect is corrected.

[0071] S represents the eddy current signal.

[0072] Furthermore, the defect size data includes the defect area;

[0073] In step A4, based on the defect size data, the defects are classified according to a preset classification rule to obtain a defect classification result;

[0074] The preset classification rules include:

[0075] When the defect area is not less than the first area threshold and less than the second area threshold, the defect is determined to be a minor damage level;

[0076] When the defect area is not less than the second area threshold and less than the third area threshold, the defect is determined to be of a moderate level;

[0077] When the defect area is not less than the third area threshold, the defect is determined to be a major damage level.

[0078] The defect level is determined based on the defect area, and operators can take corresponding measures to improve quality control points in related processes for defects of different levels.

[0079] Furthermore, the defect size data includes defect depth;

[0080] In step A4, based on the defect size data, the defects are classified according to a preset classification rule to obtain a defect classification result;

[0081] The preset classification rules include:

[0082] When the defect depth is not less than the first depth threshold and less than the second depth threshold, the defect is determined to be a minor damage level;

[0083] When the defect depth is not less than the second depth threshold and less than the third depth threshold, the defect is determined to be of a moderate level;

[0084] When the defect depth is not less than the third depth threshold, the defect is determined to be a major damage level.

[0085] The defect level is determined based on the defect depth, and operators can take corresponding measures to improve quality control points in related processes for defects of different levels.

[0086] Specifically, the defect classification formula is as follows:

[0087]

[0088] Wherein, d represents the defect size, which is the defect area or defect depth.

[0089] When d is the defect area, min represents the first area threshold, d mid represents the second area threshold, d max Indicates the third area threshold.

[0090] When d is the defect depth, min represents the first depth threshold, d mid represents the second depth threshold, d max Represents the third depth threshold.

[0091] Furthermore, the characteristic data includes energy characteristics and kurtosis characteristics.

[0092] Specifically, feature data extraction is performed in the frequency domain, and the energy feature extraction formula is as follows:

[0093]

[0094] Among them, f energy Indicates energy characteristics, reaction signal intensity, S i represents the eddy current signal at the i-th sampling point, and N represents the number of sampling points;

[0095] The extraction formula of kurtosis feature is as follows:

[0096]

[0097] Among them, f kurtosis Indicates the kurtosis characteristic, reflecting the sharpness of the signal, Si represents the eddy current signal at the i-th sampling point, N represents the number of sampling points, σ represents the standard deviation of the entire sampling point signal, and μ represents the mean of the entire sampling point signal.

[0098] Furthermore, in step A2, before extracting features from the eddy current signal, the eddy current signal is subjected to signal enhancement processing based on a wavelet transform algorithm and / or temperature compensation processing based on a temperature compensation algorithm.

[0099] Specifically, the calculation formula for signal enhancement processing of eddy current signals based on the wavelet transformation algorithm is as follows:

[0100] S enhanced =WaveletDenoise(S raw ,mother_wavelet,threshold)

[0101] Among them, S enhanced represents the eddy current signal after signal enhancement processing;

[0102] WaveletDenoise represents the wavelet denoising algorithm;

[0103] S raw represents the eddy current signal before signal enhancement processing;

[0104] Specifically, mother_wavelet represents a basic function for generating a wavelet basis function; preferably, the wavelet transformation algorithm uses Daubechieshuo wavelet basis function or Morlet wavelet odd function.

[0105] Specifically, "threshold" represents the denoising threshold. Either a hard threshold or an adaptive threshold can be used for shrinkage denoising. Adaptive thresholding removes noise signals and enhances defect signals, enabling better differentiation between real defects and noise. This provides higher-quality input features for subsequent defect recognition models, reducing the probability of false positives and missed detections.

[0106] Specifically, the calculation formula of the adaptive threshold is as follows:

[0107] d threshold =μ defect +k·σ defect

[0108] in,

[0109] d threshold represents the adaptive threshold;

[0110] μ defect represents the mean value of the signal features extracted from the eddy current signal;

[0111] σ defect represents the standard deviation of the signal features extracted from the eddy current signal;

[0112] k represents the safety factor, which is usually 2-3.

[0113] The purpose of temperature compensation is to eliminate the impact of temperature changes on the signal. Specifically, the calculation formula of the temperature compensation algorithm is as follows:

[0114] S corrected =S raw0 [1+β(T-T0)] -1

[0115] S raw0 represents the eddy current signal before temperature compensation, S corrected It represents the eddy current signal after temperature compensation, T represents the temperature under the current environment or measurement conditions, and T0 represents the reference temperature value.

[0116] Specifically, the signal enhancement processing of the eddy current signal based on the wavelet transformation algorithm can be performed in the time-frequency domain. The temperature compensation processing of the eddy current signal based on the temperature compensation algorithm can be performed in the time domain or the frequency domain.

[0117] Specifically, the signal enhancement processing of the eddy current signal based on the wavelet transformation algorithm can be performed after the temperature compensation processing of the eddy current signal based on the temperature compensation algorithm, or can be performed before the temperature compensation processing of the eddy current signal based on the temperature compensation algorithm.

[0118] Furthermore, in step A1, eddy current signals corresponding to the pipe fittings under the detection signals of each preset frequency emitted by the flaw detector are respectively collected;

[0119] In step A2, before extracting the features of the eddy current signal, the eddy current signal collected under the detection signal of each preset frequency is subjected to multi-frequency signal fusion processing in the frequency domain to obtain a fused eddy current signal, and then the feature extraction of the fused eddy current signal is performed to obtain feature data.

[0120] Through multi-frequency signal fusion processing, signal information at different frequencies is integrated to obtain more comprehensive signal characteristics.

[0121] As an embodiment of the present invention, the originally collected eddy current signal is subjected to signal enhancement processing, temperature compensation processing, and then multi-frequency signal fusion processing.

[0122] As another embodiment of the present invention, the originally collected eddy current signal is first subjected to multi-frequency signal fusion processing, and then subjected to signal enhancement processing and temperature compensation processing.

[0123] Multi-frequency signal fusion processing can be performed using a weighted fusion method.

[0124] The calculation formula of the weighted fusion method is as follows:

[0125]

[0126] in,

[0127] S fused represents the signal after weighted fusion;

[0128] α k Represents the weight corresponding to the k-th signal before fusion;

[0129] represents the square of the standard deviation of the signal before the kth fusion;

[0130] S k (f k ) represents the signal before fusion obtained under the detection signal of the k-th preset frequency;

[0131] f k A value representing a specific preset frequency of a detection signal of the kth preset frequency;

[0132] K represents the number of signals before fusion.

[0133] Furthermore, in step A1, eddy current signals corresponding to the pipe fittings under the detection signals of each preset frequency emitted by the flaw detector are respectively collected;

[0134] In step A2, feature extraction is performed on the eddy current signals collected under the detection signals of each preset frequency in the frequency domain, and the extracted feature data are fused to obtain final feature data.

[0135] See also Figure 2 The present invention further provides a pipe defect detection system for executing the aforementioned pipe defect detection method, comprising:

[0136] A signal acquisition module (1) is used to acquire eddy current signals generated by the pipe after the flaw detector sends a detection signal;

[0137] A feature extraction module (2) is connected to the signal acquisition module (1) and is used to extract features from the eddy current signal to obtain feature data;

[0138] A defect recognition module (3) is connected to the feature extraction module (2) and is used to process feature data based on a pre-trained defect recognition model to perform defect recognition of the pipe fitting and obtain a defect recognition result, wherein the defect recognition result includes defect and defect size data;

[0139] The defect classification module (4) is connected to the defect recognition module (3) and is used to classify the defects according to preset classification rules based on the size data of the defects to obtain a defect grade classification result.

[0140] The present invention uses a defect recognition algorithm based on artificial intelligence to identify defects based on features extracted from eddy current signals, effectively identifies defects, and obtains quantitative information such as defect size data. The severity of the identified defects is distinguished according to the size. The recognition accuracy of pipe fittings is improved through artificial intelligence, and the recognition efficiency is improved. Quantitative and qualitative analysis helps in subsequent decision-making, which is of great significance for ensuring the quality of high-efficiency internally threaded copper pipes and preventing safety accidents.

[0141] Furthermore, the defect recognition module (3) identifies the defect probability estimation value of each inspection point in the pipe from the feature data based on the defect recognition model;

[0142] The pipe defect detection system also includes:

[0143] The defect identification module (5) is connected to the defect recognition module (3) and is used to form a defect identification map of the pipe fitting based on the defect probability estimation value of each detection point and a preset color coding mapping rule.

[0144] The above are only preferred embodiments of the present invention and do not limit the implementation mode and protection scope of the present invention. For those skilled in the art, it should be aware that all solutions obtained by equivalent substitutions and obvious changes made using the description and illustrations of the present invention should be included in the protection scope of the present invention.

Claims

1. A pipe defect detection method, characterized in that: include: Step A1, collecting an eddy current signal generated by the pipe after the flaw detector sends a detection signal; Step A2, extracting features from the eddy current signal to obtain feature data; Step A3, processing the feature data based on a pre-trained defect recognition model to perform defect recognition on the pipe fitting to obtain a defect recognition result, wherein the defect recognition result includes defect and defect size data; Step A4: classify the defects based on the size data of the defects to obtain a defect grade classification result.

2. A pipe defect detection method as claimed in claim 1, characterized in that: In the step A3, a defect probability estimation value of each inspection point in the pipe is identified from the feature data based on the defect recognition model; pipe fittings In the step A4, a defect identification map of the pipe is formed based on the defect probability estimation value of each detection point and a preset color coding mapping rule.

3. A pipe defect detection method as claimed in claim 2, characterized in that: In step A3, The defect identification result also includes the location data of the defect; In step A4, the position data of the defect is corrected, and then a defect identification map of the pipe is formed based on the corrected position data of the defect, the defect probability estimation value of each detection point and the preset color coding mapping rule.

4. A pipe defect detection method as claimed in claim 1, characterized in that: The size data of the defect includes the defect area; In the step A4, based on the size data of the defect, the defect is classified according to a preset classification rule to obtain the defect grade classification result; The preset classification rules include: When the defect area is not less than the first area threshold and less than the second area threshold, determining that the defect is a minor damage level; When the defect area is not less than the second area threshold and less than the third area threshold, determining that the defect is of a moderate level; When the defect area is not less than the third area threshold, the defect is determined to be a major damage level.

5. A pipe defect detection method as claimed in claim 1, characterized in that: The size data of the defect includes the depth of the defect; In step A4, based on the size data of the defect, the defect is classified according to a preset classification rule to obtain a defect classification result; The preset classification rules include: When the depth of the defect is not less than the first depth threshold and less than the second depth threshold, determining that the defect is a minor damage level; When the defect depth is not less than the second depth threshold and less than the third depth threshold, determining that the defect is of a moderate level; When the defect depth is not less than the third depth threshold, the defect is determined to be a major damage level.

6. A pipe defect detection method as claimed in claim 1, characterized in that: In step A2, before extracting features from the eddy current signal, the eddy current signal is subjected to signal enhancement processing based on a wavelet transform algorithm and / or temperature compensation processing based on a temperature compensation algorithm.

7. A pipe defect detection method as claimed in claim 1, characterized in that: In the step A1, eddy current signals corresponding to the pipe fitting under the detection signals of each preset frequency emitted by the flaw detector are respectively collected; In step A2, before extracting features from the eddy current signal, multi-frequency signal fusion processing is performed on the eddy current signal collected under the detection signal of each preset frequency in the frequency domain to obtain a fused eddy current signal, and then feature extraction is performed on the fused eddy current signal to obtain the feature data.

8. A pipe defect detection method as claimed in claim 1, characterized in that: In the step A1, eddy current signals corresponding to the pipe fitting under the detection signals of each preset frequency emitted by the flaw detector are respectively collected; In step A2, feature extraction is performed on the eddy current signals collected under the detection signals of each preset frequency in the frequency domain, and the extracted feature data are fused to obtain final feature data.

9. A pipe defect detection system, characterized in that: A method for detecting pipe defects according to any one of claims 1 to 8, comprising: A signal acquisition module, used for acquiring an eddy current signal generated by the pipe after the flaw detector sends a detection signal; A feature extraction module, connected to the signal acquisition module, for extracting features from the eddy current signal to obtain feature data; a defect recognition module, connected to the feature extraction module, for processing the feature data based on a pre-trained defect recognition model to perform defect recognition on the pipe fitting and obtain a defect recognition result, wherein the defect recognition result includes defect and defect size data; The defect classification module is connected to the defect recognition module and is used to classify the defects based on the size data of the defects to obtain a defect grade classification result.

10. A pipe defect detection system as claimed in claim 9, characterized in that: A defect recognition module identifies a defect probability estimation value of each inspection point in the pipe from the feature data based on the defect recognition model; The pipe fitting defect detection system further includes: A defect identification module is connected to the defect recognition module and is used to form a defect identification map of the pipe fitting based on the defect probability estimation value of each detection point and a preset color coding mapping rule.

Images