Micro-magnetic detection method for simultaneously identifying material and surface hard spots of steel pipe

Abstract

The invention discloses a micro-magnetic detection method for simultaneously identifying the material and surface hard spots of a steel pipe, which adopts a marshalling serial magnetization method to measure multi-dimensional micro-magnetic characteristic parameters of the steel pipe under different excitation frequencies and amplitude conditions. Steel pipe material identification and hard spot identification are carried out in sequence by adopting mode identification and quantitative regression intelligent models, and multi-dimensional micro-magnetic characteristic parameters input into the two intelligent models are selected according to a set grouping rule. The marshalling rule is determined by a calibration experiment under sweep amplitude and sweep frequency magnetization conditions. A multi-channel micro-magnetic sensor is carried on an annular scanner, scanning is carried out in the axial direction of a steel pipe, the micro-magnetic sensor of each channel carries out multi-dimensional micro-magnetic characteristic parameter detection according to a set marshalling serial magnetization method, and the multi-dimensional micro-magnetic characteristic parameters are input into a mode recognition and quantitative regression intelligent model according to a marshalling rule; and steel pipe material identification and three-dimensional imaging of hard spots on the surface of the steel pipe are realized. The invention provides an on-line detection way for automatic sorting and performance evaluation in the steel pipe manufacturing stage.

Description

Technical Field

[0001] The present invention relates to the technical field of nondestructive testing, and particularly relates to a micro-magnetic testing method for simultaneously identifying the material of a steel pipe and surface hard spots, which can provide an on-line testing approach for automatic sorting and performance evaluation during the manufacturing stage of steel pipes. Background Art

[0002] Unevenness defects such as hard spots (with a hardness higher than that of the surrounding base material by 30 HV and any dimension greater than 50 mm) generally exist in steel plates and pipes produced by major steel mills. There are large hardness differences in these local areas, resulting in stress concentration during plastic deformation or service of the steel, and thus cracks are generated on the surface. Carbon steel is commonly used for oil pipelines to withstand the pressure and corrosion of petroleum products. Hydrogen has high permeability, which can cause embrittlement of some conventional materials, and alloy steel is commonly used for hydrogen pipelines. If steel mixing occurs, serious safety problems may occur. Currently, the material type and surface hardness value are mainly tested based on destructive experiments, and it cannot be detected 100%, and it is easy to miss detections.

[0003] Internationally, the method of indirectly standardizing the material and surface hardness using the magnetic Barkhausen noise signal has been fully verified. For example

[0004] "TAMPL, HAMMERSBERG P, PERSSON G, et al. Case depth evaluation of induction-hardened camshaft by using magnetic Barkhausen noise (MBN) method [J]. Nondestructive Testing and Evaluation, 2021, 36(5): 494-514. WANG Y, MELIKHOV Y, MEYDAN T. Multifunctional induction coil sensor for evaluation of carbon content in carbon steel [J]. IEEE Transactions on Magnetics, 2022, 59(2): 1-5. DING S, TIAN G, SUTTHAWEEKUL R. Non-destructive hardness prediction for 18CrNiMo7-6 steel based on feature selection and fusion of Magnetic Barkhausen Noise [J]. Ndt & E International, 2019, 107: 102138. ARANAS J R C, HEY M, PODLESNY M. Magnetic Barkhausen noise characterization of two pipeline steels with unknown history [J]. Materials Characterization, 2018, 146: 243-257." The samples with different microstructures and the surface hardness values of the steel plates were respectively identified and characterized. However, there are few reports on the multi-channel micro-magnetic detection for online simultaneous identification of the steel pipe material and surface hard spots. Therefore, the simultaneous detection of the steel pipe material and surface hardness value is of great significance. Summary of the Invention

[0005] Aiming at the defects of the prior art, the present invention provides a micro-magnetic detection method for simultaneous identification of the steel pipe material and surface hard spots, which fuses two types of micro-magnetic signals (magnetic Barkhausen noise, tangential magnetic field strength), uses an annular scanner to attach a multi-channel micro-magnetic sensor to the surface of the steel pipe, and scans along the axial direction of the steel pipe. The micro-magnetic sensors of each channel perform multi-dimensional micro-magnetic characteristic parameter detection according to the set grouped serial magnetization method, and input the multi-dimensional micro-magnetic characteristic parameters into the pattern recognition and quantitative regression intelligent model according to the grouping rules to realize the identification of the steel pipe material and the three-dimensional imaging of the surface hard spots of the steel pipe.

[0006] To achieve the above object, the technical solution adopted by the present invention is as follows:

[0007] A micro-magnetic detection method for simultaneously identifying the material of a steel pipe and surface hard spots, which adopts a grouped serial magnetization method. An annular scanner is used to attach multi-channel micro-magnetic sensors to the surface of the steel pipe and scan along the axial direction of the steel pipe, measuring the multi-dimensional micro-magnetic characteristic parameters of the steel pipe under different excitation frequencies and amplitudes. According to the set grouping rules, the measured multi-dimensional micro-magnetic characteristic parameters are extracted and input into a pattern recognition intelligent model and a quantitative regression intelligent model respectively, to predict the material of the steel pipe and the surface hardness value, and identify hard spots from the surface hardness scan imaging diagram based on the determination threshold.

[0008] The specific implementation steps are as follows:

[0009] 1) Serialize the magnetization of the micro-magnetic sensors carried in the annular scanner according to the grouping rules. Specifically, the steel pipe is magnetized multiple times by sequentially exciting a magnetic field with a preset frequency and amplitude in time segments, and the excited magnetic fields are arranged in a serial time sequence from low to high in amplitude.

[0010] 2) Test the micro-magnetic signals (including magnetic Barkhausen noise and tangential magnetic field) during the grouped serial magnetization process, extract the multi-dimensional micro-magnetic characteristic parameters, and according to the set grouping rules, extract and classify the multi-dimensional micro-magnetic characteristic parameters, and input them into the pattern recognition intelligent model and the quantitative regression intelligent model respectively.

[0011] 3) Move the annular scanner along the axial direction of the steel pipe at a set step length, and repeat steps (1), (2), and (3) until the scanning task of the set path is completed. Draw the surface hardness predicted at different positions to form a surface hardness scan imaging diagram, and mark the area above the threshold from the surface hardness scan imaging diagram according to the surface hardness threshold for hard spot determination, which is determined as a hard spot.

[0012] The grouped serial magnetization method is characterized in that the grouping rules are determined by calibration experiments under the conditions of scanning amplitude and scanning frequency. The specific process is as follows:

[0013] 1) Prepare steel pipe specimens of different materials, and each specimen contains hard spots and the surface hardness in the central area of the hard spots is higher than the threshold.

[0014] 2) Set the test ranges of the excitation magnetic field frequency and amplitude, divide the test frequency and amplitude ranges according to an equal interval scanning step length to form an excitation parameter matrix.

[0015] 3) Use the micro-magnetic sensors to perform point-by-point movement detection on the path across the hard spots. At each detection position, according to the excitation parameter matrix, test the multi-dimensional micro-magnetic characteristic parameters of the steel pipe under different excitation frequencies and amplitudes one by one, and test the surface hardness at the detection point position.

[0016] 4) Use the multi-dimensional micro-magnetic characteristic parameter data measured in the non-hard spot areas of different materials and the material labels as the training set of the pattern recognition intelligent model, train to obtain an intelligent steel pipe material recognition model, and mark the multi-dimensional micro-magnetic characteristic parameter S1 input thereto and its corresponding excitation condition H1;

[0017] 5) For specimens of the same material, use the surface hardness data and multi-dimensional micro-magnetic characteristic parameter data of all test points as the training set of the quantitative regression intelligent model, train to obtain an intelligent prediction model for the surface hardness of the steel pipe, and mark the multi-dimensional micro-magnetic characteristic parameter S2 input thereto and its corresponding excitation condition H2;

[0018] 6) Perform a summation process on the excitation condition H1 and the excitation condition H2, and serially arrange all the excitation conditions in ascending order of amplitude, which is the magnetization grouping rule.

[0019] Through the grouped serial magnetization method, measure the multi-dimensional micro-magnetic characteristic parameters of the steel pipe under different excitation frequencies and amplitude conditions; select the multi-dimensional micro-magnetic characteristic parameters according to the set grouping rule and input them into the pattern recognition and quantitative regression intelligent models to achieve steel pipe material identification and imaging of hard spots on the steel pipe surface.

[0020] The annular scanner attaches the multi-channel micro-magnetic sensor to the surface of the steel pipe and scans along the axial direction of the steel pipe.

[0021] Advantageous effects: The method of the present invention utilizes the non-destructive characteristics of the micro-magnetic method, and can not only achieve steel pipe material identification and imaging of hard spots on the steel pipe surface, but also provide an on-line detection method for automatic sorting and performance evaluation in the steel pipe manufacturing stage. Description of the Drawings

[0022] Figure 1 : Micro-magnetic detection system for hard spots on the steel pipe surface;

[0023] Figure 2 : Prediction process of the intelligent steel pipe material recognition model and the quantitative regression intelligent model;

[0024] Figure 3 : Surface hardness scanning imaging;

[0025] Figure 4 : Binary image results of surface hardness scanning imaging under different thresholds;

[0026] Figure 5 : Grouping rule establishment process;

[0027] In the figure: 1 - multi-channel sensor fixture 2 - steel pipe annular device 3 - steel pipe Specific Embodiments

[0028] Embodiments of the present invention will be described in detail below. Examples of the embodiments are shown in the accompanying drawings. The embodiments described below by referring to the drawings are exemplary and are intended to explain the present invention, and should not be construed as a limitation to the present invention.

[0029] The present invention proposes a micro-magnetic detection method for simultaneously identifying the material of a steel pipe and surface hard spots, which integrates two types of micro-magnetic signals (magnetic Barkhausen noise, tangential magnetic field intensity). An annular scanner is used to attach multi-channel micro-magnetic sensors to the surface of the steel pipe and scan along the axial direction of the steel pipe. Each channel of the micro-magnetic sensor performs multi-dimensional micro-magnetic characteristic parameter detection according to the set grouping serial magnetization method. The multi-dimensional micro-magnetic characteristic parameters are input into the pattern recognition and quantitative regression intelligent models according to the grouping rules, so as to realize the identification of the steel pipe material and the three-dimensional imaging of the surface hard spots of the steel pipe.

[0030] Specifically, the method adopts the grouping serial magnetization method. An annular scanner is used to attach multi-channel micro-magnetic sensors to the surface of the steel pipe and scan along the axial direction of the steel pipe. The multi-dimensional micro-magnetic characteristic parameters of the steel pipe under different excitation frequencies and amplitudes are measured. According to the set grouping rules, the measured multi-dimensional micro-magnetic characteristic parameters are extracted and input into the pattern recognition intelligent model and the quantitative regression intelligent model respectively, and the material of the steel pipe and the surface hardness value are predicted. Hard spots are identified from the surface hardness scan imaging map according to the judgment threshold. It includes:

[0031] (1) Material identification: According to the set grouping rules, the measured multi-dimensional micro-magnetic characteristic parameters are extracted and input into the pattern recognition intelligent model, and the material of the steel pipe is predicted.

[0032] (2) Hardness prediction: According to the set grouping rules, the measured multi-dimensional micro-magnetic characteristic parameters are extracted and input into the quantitative regression intelligent model corresponding to the material, and the surface hardness of the steel plate is predicted.

[0033] (3) Hard spot identification: Hard spots are identified by thresholding the surface hardness scan imaging Figure 2 value.

[0034] The grouping rules are determined by calibration experiments under the scanning amplitude and scanning frequency magnetization conditions. The specific process is as follows:

[0035] 1) Prepare steel pipe specimens of different materials. Each material specimen contains hard spots and the surface hardness of the central area of the hard spots is higher than the threshold;

[0036] 2) Set the test ranges of the excitation magnetic field frequency and amplitude, and divide the test frequency and amplitude ranges according to the equal interval scanning step size to form an excitation parameter matrix;

[0037] 3) Use a micro-magnetic sensor to perform point-by-point movement detection along the path across the hard spot. At each detection position, test the multi-dimensional micro-magnetic characteristic parameters of the steel pipe under different excitation frequencies and amplitudes one by one according to the excitation parameter matrix, and test the surface hardness at the detection point position.

[0038] 4) Use the multi-dimensional micro-magnetic characteristic parameter data and material labels measured in the non-hard spot areas of different materials as the training set of the pattern recognition intelligent model, train to obtain the intelligent recognition model of the steel pipe material, and mark the input multi-dimensional micro-magnetic characteristic parameter S1 and its corresponding excitation condition H1.

[0039] 5) For specimens of the same material, use the surface hardness data and multi-dimensional micro-magnetic characteristic parameter data of all test points as the training set of the quantitative regression intelligent model, train to obtain the intelligent prediction model of the steel pipe surface hardness, and mark the input multi-dimensional micro-magnetic characteristic parameter S2 and its corresponding excitation condition H2.

[0040] 6) Perform a summation process on the excitation condition H1 and the excitation condition H2, and arrange all the excitation conditions in a serial time sequence from low to high in amplitude, which is the magnetization grouping rule. 2.1) Select steel pipe specimens of the same size but different materials.

[0041] The device used in the detection method of the present invention includes a micro-magnetic detection instrument and an annular scanner. As Figure 1 shown, it is the annular scanner. This device mainly consists of a multi-channel sensor fixture 1 and a movable device 2. 8 micro-magnetic sensors can be installed on the multi-channel sensor fixture 1. By carrying the multi-channel sensor fixture 1 with the movable device 2, the micro-magnetic sensors can be attached to the surface of the steel pipe 3 for axial scanning.

[0042] During specific implementation, first, mount the multi-channel micro-magnetic sensors on the annular scanner, attach them to the surface of the steel pipe, and perform axial scanning along the steel pipe. Each channel of the micro-magnetic sensor detects the multi-dimensional micro-magnetic characteristic parameters according to the set grouped serial magnetization method, and input the multi-dimensional micro-magnetic characteristic parameters into the pattern recognition and quantitative regression intelligent models according to the grouping rule to achieve steel pipe material identification and imaging of the hard spots on the steel pipe surface.

[0043] As Figure 2 shown, it is the prediction process of the steel pipe material identification intelligent model and the quantitative regression intelligent model. First, magnetize the steel pipe multiple times by sequentially exciting a magnetic field with a preset frequency and amplitude in a time sequence. The excited magnetic field is arranged in a serial time sequence from low to high in amplitude, extract the multi-dimensional magnetic characteristic parameters, and according to the set grouping rule, extract and classify the multi-dimensional micro-magnetic characteristic parameters, and input them into the pattern recognition intelligent model and the quantitative regression intelligent model corresponding to the material respectively to obtain the steel pipe material and the surface hardness value.

[0044] As Figure 3The surface hardness scanning imaging is shown. According to the hardness values, it is plotted in the form of a cloud map to evaluate the uniformity of the surface hardness of the steel pipe.

[0045] As Figure 4 shown is the binaryzation result of the surface hardness scanning imaging under different thresholds. The surface hardness scanning imaging Figure 2 is binarized, and the hard spot area can be obtained according to the threshold.

[0046] As Figure 5 shown is the process of establishing the grouping rule. The grouping rule is determined by the calibration experiment under the scanning amplitude and scanning frequency magnetization conditions. Set the range of excitation parameters, divide it according to the test frequency and amplitude range to form an excitation parameter matrix; use the multi-dimensional magnetic characteristic parameters and material labels measured in the non-hard spot areas of different materials as the training set of the pattern recognition intelligent model, and train to obtain the intelligent identification model of the steel pipe material, and mark the multi-dimensional magnetic characteristic parameter S1 and its corresponding excitation condition H1 input when the model is optimal; for the specimens of the same material, use the surface hardness data and multi-dimensional magnetic characteristic parameter data as the training set of the quantitative regression intelligent model, and train to obtain the intelligent prediction model of the surface hardness of the steel pipe, and mark the multi-dimensional micro-magnetic characteristic parameter S2 and its corresponding excitation condition H2 input when the model is optimal; perform a summation process on H1 and the excitation condition H2, and arrange all the excitation conditions in a serial time sequence from low to high in amplitude, which is the magnetization grouping rule.

Claims

1. A micromagnetic detection method for simultaneously identifying the material of a steel pipe and hard spots on its surface, characterized in that: The multi-dimensional micro-magnetic characteristic parameters of the steel pipe under different excitation frequencies and amplitudes were measured by using the grouping serial magnetization method. According to the set grouping rules, the measured multi-dimensional micro-magnetic characteristic parameters were extracted and input into the pattern recognition intelligent model and the quantitative regression intelligent model respectively to predict the material and surface hardness value of the steel pipe. The hard spots were identified from the surface hardness scanning imaging map according to the judgment threshold. The implementation steps are as follows: (1) The micromagnetic sensors carried in the annular scanner are serially magnetized according to the grouping rules. Specifically, the steel pipe is magnetized multiple times by exciting the magnetic field with a preset frequency and amplitude in a time sequence, and the excited magnetic field is arranged in a serial time sequence from low to high according to the amplitude; (2) Testing the micromagnetic signals during the serial magnetization process of the marshaling, which contain magnetic Barkhausen noise, incremental magnetic permeability and tangential magnetic field, extracts multidimensional micromagnetic characteristic parameters, extracts and classifies the multidimensional micromagnetic characteristic parameters according to the set marshaling rules, and inputs them into the pattern recognition intelligent model and the quantitative regression intelligent model respectively; (3) Use the pattern recognition intelligent model to determine the material of the steel pipe, select the corresponding quantitative regression intelligent model according to the material, and predict the surface hardness value; (4) Move the annular scanner along the axial direction of the steel pipe according to the set step length, and repeat steps (1), (2), and (3) until the scanning task of the set path is completed. The surface hardness predicted at different positions is plotted to form a surface hardness scanning image. According to the surface hardness threshold for hard spot determination, the area above the threshold is marked in the surface hardness scanning image, and it is determined to be a hard spot.

2. The micromagnetic detection method for simultaneously identifying the steel pipe material and surface hard spots according to claim 1 is characterized in that: The grouping rules are determined by calibration experiments under sweep amplitude and sweep frequency magnetization conditions. The specific process is as follows: (1) Steel pipe specimens of different materials are prepared, each of which contains hard spots and the surface hardness of the central area of ​​the hard spots is higher than a threshold value; (2) setting the test range of the excitation field frequency and amplitude, dividing the test frequency and amplitude range according to the equally spaced scanning steps to form an excitation parameter matrix; (3) Use a micromagnetic sensor to perform point-by-point mobile detection on the path across the hard spot. At each detection position, according to the excitation parameter matrix, the multi-dimensional micromagnetic characteristic parameters of the steel pipe under different excitation frequencies and amplitudes are tested one by one, and the surface hardness of the detection point is tested; (4) The multi-dimensional micro-magnetic characteristic parameter data and material labels measured in the non-hard spot area of ​​different materials are used as the training set of the pattern recognition intelligent model, and the steel pipe material intelligent recognition model is trained to mark its input multi-dimensional micro-magnetic characteristic parameter S1 and its corresponding excitation condition H1; (5) For samples of the same material, the surface hardness data and multi-dimensional micro-magnetic characteristic parameter data of all test points are used as the training set of the quantitative regression intelligent model, and the steel pipe surface hardness intelligent prediction model is trained to mark its input multi-dimensional micro-magnetic characteristic parameter S2 and its corresponding excitation condition H2; (6) The excitation conditions H1 and H2 are summed, and all the excitation conditions are arranged in serial order from low to high according to the amplitude, which is the magnetization grouping rule.

3. The micromagnetic detection method for simultaneously identifying the steel pipe material and surface hard spots according to claim 1 is characterized in that: The device for implementing the method includes a micromagnetic detection instrument and an annular scanner; the annular scanner includes a multi-channel sensor fixture and a movable device; 8 micromagnetic sensors are installed on the multi-channel sensor fixture, and the multi-channel sensor fixture is carried by the movable device, so that the micromagnetic sensor can be attached to the surface of the steel pipe for axial scanning.

4. The micromagnetic detection method for simultaneously identifying the steel pipe material and surface hard spots according to claim 3 is characterized in that: Firstly, the multi-channel micromagnetic sensor is mounted on a circular scanner so that it fits the surface of the steel pipe and scans along the axial direction of the steel pipe. The micromagnetic sensor of each channel performs multi-dimensional micromagnetic characteristic parameter detection according to the set grouping serial magnetization method. The multi-dimensional micromagnetic characteristic parameters are input into the pattern recognition and quantitative regression intelligent model according to the grouping rules to realize the recognition of steel pipe material and the imaging of hard spots on the steel pipe surface.

5. The micromagnetic detection method for simultaneously identifying the steel pipe material and surface hard spots according to claim 3 is characterized in that: It is the prediction process of the intelligent model for material identification of steel pipes and the quantitative regression intelligent model. Firstly, the steel pipe is magnetized multiple times by stimulating the magnetic field with preset frequency and amplitude in time sequence. The excited magnetic field is arranged in serial time sequence from low to high according to the amplitude, and the multi-dimensional magnetic characteristic parameters are extracted. According to the set grouping rules, the multi-dimensional micro-magnetic characteristic parameters are extracted and classified, and respectively input into the pattern recognition intelligent model and the quantitative regression intelligent model of the corresponding material to obtain the material and surface hardness value of the steel pipe.

6. The micromagnetic detection method for simultaneously identifying the steel pipe material and surface hard spots according to claim 5 is characterized in that: The surface hardness is scanned and imaged, and according to the hardness value, it is drawn into a cloud map to evaluate the uniformity of the surface hardness of the steel pipe.

7. The micromagnetic detection method for simultaneously identifying the steel pipe material and surface hard spots according to claim 5 is characterized in that: Binarization results of surface hardness scanning imaging under different thresholds. The surface hardness scanning imaging image is binarized, and the hard spot area can be obtained according to the threshold.

8. The micromagnetic detection method for simultaneously identifying the steel pipe material and surface hard spots according to claim 5, characterized in that: The grouping rule establishment process is that the grouping rule is determined by the calibration experiment under the conditions of sweep amplitude and sweep frequency magnetization; the excitation parameter range is set, divided according to the test frequency and amplitude range, and an excitation parameter matrix is ​​formed; the multidimensional magnetic characteristic parameters and material labels measured in the non-hard spot area of ​​different materials are used as the training set of the pattern recognition intelligent model, and the steel pipe material intelligent recognition model is trained to obtain the multidimensional magnetic characteristic parameter S1 and its corresponding excitation condition H1 input when the model is optimal; the surface hardness data and multidimensional magnetic characteristic parameter data of the sample of the same material are used as the training set of the quantitative regression intelligent model, and the intelligent prediction model of the steel pipe surface hardness is trained to obtain the multidimensional micromagnetic characteristic parameter S2 and its corresponding excitation condition H2 input when the model is optimal; H1 and the excitation condition H2 are summed, and all the excitation conditions are arranged in serial time sequence from low to high according to the amplitude, which is the magnetization grouping rule.

Images