Nondestructive testing method for metal plate

Abstract

The invention discloses a nondestructive testing method for a metal plate, which comprises the following steps of: 1, heating the metal plate by a pulse excitation signal generation system, and acquiring the surface temperature of the metal plate to obtain a temperature image of the surface of the metal plate; 2, identifying a temperature sudden change area in the temperature image of the surface of the metal plate to obtain a crack position; 3, collecting changes of the primary magnetic field and the secondary induced magnetic field, and obtaining a rising edge peak potential difference and a falling edge peak potential difference; and 4, constructing a crack quantitative detection model, and inputting the rising edge peak potential difference and the falling edge peak potential difference into the crack quantitative detection model to obtain the crack depth. Wherein the crack quantitative detection model comprises an input layer, a first convolution layer, a first batch of standardization layers, a second convolution layer, a second batch of standardization layers, a discarding layer, a linear rectification layer, a full connection layer and an output layer. The device and the method have the characteristics of accurately positioning the crack and accurately detecting the depth of the crack.

Description

Technical Field

[0001] The present invention relates to the technical field of nondestructive testing, and more particularly to a nondestructive testing method for a metal plate. Background Art

[0002] Due to the harsh working environment and complex working conditions, cracks may appear inside or on the surface of metal components. If the cracks or defects are not dealt with in time and allowed to worsen, they may lead to the breakage of metal components and failure of mechanical equipment, which will cause significant losses. Therefore, it is of great significance to conduct daily crack detection on metal components to achieve early identification of cracks.

[0003] Nondestructive testing refers to the method of inspecting and testing the structure, state and type, quantity, shape, nature, position, size, distribution and changes of defects inside and on the surface of the test piece by using physical or chemical methods, with the help of modern technology and equipment, without damaging or affecting the performance of the tested object or the internal organization of the tested object. At present, the commonly used nondestructive testing methods are radiographic testing (RT), ultrasonic testing (UT), liquid penetrant testing (PT), magnetic particle testing (MT) and eddy current testing (ECT). The principles based on different methods are different, and each method has its own characteristics and limitations. Basically, all kinds of testing methods have the problems of large calculation amount and low prediction accuracy, and they are all simulation predictions of artificial defects, and there is no corresponding prediction method for naturally formed cracks. Summary of the invention

[0004] The purpose of the present invention is to design and develop a non-destructive testing method for metal plates, which can accurately locate cracks in metal plates and accurately detect the depth of cracks, thereby improving the efficiency of identification and detection while ensuring accuracy.

[0005] The technical solution provided by the present invention is:

[0006] A nondestructive testing method for a metal plate comprises the following steps:

[0007] Step 1: The pulse excitation signal generating system heats the metal plate, collects the surface temperature of the metal plate, and obtains a temperature image of the surface of the metal plate;

[0008] Step 2: Identify the area with sudden temperature change in the temperature image on the surface of the metal plate to obtain the crack position;

[0009] Step 3: Collect the changes of the primary magnetic field and the secondary induced magnetic field to obtain the peak potential difference along the rising edge and the peak potential difference along the falling edge;

[0010] Step 4: construct a crack quantitative detection model, input the rising edge peak potential difference and the falling edge peak potential difference into the crack quantitative detection model, and obtain the crack depth;

[0011] The crack quantitative detection model includes an input layer, a first convolutional layer, a first batch of normalization layers, a second convolutional layer, a second batch of normalization layers, a discard layer, a linear rectification layer, a fully connected layer and an output layer.

[0012] Preferably, it also includes:

[0013] The temperature image of the metal plate surface is corrected, and the corrected temperature image of the metal plate surface satisfies:

[0014]

[0015] Where, T i, ' j Represents the corrected temperature of the pixel with coordinates (i, j), T i,j represents the measured temperature of the pixel with coordinates (i, j), Δ q is the qth difference.

[0016] Preferably, the measured temperature of the pixel point with coordinates (i, j) satisfies:

[0017]

[0018] In the formula, β q is the measured temperature of the qth adjacent pixel.

[0019] Preferably, the adjacent pixel points are 8 pixel points with a step length of 2 from the pixel point with coordinates (i, j), and one adjacent pixel point is fixed as the first adjacent pixel point, and the rest are arranged clockwise as the second to eighth adjacent pixel points.

[0020] Preferably, in step three, the changes of the primary magnetic field and the secondary induced magnetic field are collected by a Hall sensor.

[0021] Preferably, the magnetic flux densities of the primary magnetic field and the secondary induced magnetic field both satisfy:

[0022] B=μ0μ r H;

[0023] Where B is the magnetic flux density, μ0 is the vacuum magnetic permeability, μ r is the relative magnetic permeability, and H is the electric field strength.

[0024] Preferably, the Hall sensor satisfies:

[0025]

[0026] Where U v is the Hall voltage, R H is the Hall coefficient, I is the loading current intensity, and t is the thickness of the metal plate.

[0027] Preferably, the rising edge peak potential difference satisfies:

[0028]

[0029] In the formula, It is the minimum difference of Hall voltage between the primary magnetic field and the secondary induced magnetic field.

[0030] Preferably, the falling edge peak potential difference satisfies:

[0031]

[0032] In the formula, It is the maximum difference of Hall voltage between the primary magnetic field and the secondary induced magnetic field.

[0033] Preferably, the loss function of the crack quantitative detection model satisfies:

[0034]

[0035] In the formula, is the actual crack depth of the y-th pixel, is the predicted crack depth of the y-th pixel.

[0036] The beneficial effects of the present invention are:

[0037] The present invention designs and develops a nondestructive testing method for metal plates, which realizes precise positioning of cracks in metal plates through pulse excitation, and accurately detects the depth of cracks through the pressure difference of eddy current magnetic field combined with convolutional neural network. The recognition rate of crack positioning can reach more than 97%, and the maximum value of crack depth detection error is no more than 2%, which improves the efficiency of identification and detection on the basis of ensuring accuracy. The nondestructive testing method for metal plates described in the present invention is mainly suitable for nondestructive testing of non-ferromagnetic metal plates. DETAILED DESCRIPTION

[0038] The present invention is further described in detail below so that those skilled in the art can implement it according to the description.

[0039] The present invention provides a nondestructive testing method for a metal plate, which specifically comprises the following steps:

[0040] Step 1: The pulse excitation signal generating system heats the metal plate, and the surface temperature field of the metal plate is collected by an infrared camera and converted into the transient temperature of each pixel point to obtain a temperature image of the metal plate surface;

[0041] Among them, the function generator generates a pulse signal, which generates a pulse current after passing through the induction heating source and is loaded into the excitation coil. The excitation coil applies the excitation current. According to the law of electromagnetic induction, a changing magnetic field is generated around the excitation coil. This magnetic field is the primary magnetic field. Under the induction action of the primary magnetic field, eddy currents are induced in the metal plate, and the eddy currents generate a secondary induced magnetic field. After the metal plate generates eddy current thermal effect, the eddy current heat is detected by the infrared camera to obtain a temperature image of the metal plate surface.

[0042] In this embodiment, due to the excitation coil, the obtained temperature image is not accurate, so the temperature image needs to be corrected. The corrected temperature image of the metal plate surface satisfies:

[0043]

[0044] Where, T i, ' j Represents the corrected temperature of the pixel with coordinates (i, j), T i,j represents the measured temperature of the pixel with coordinates (i, j), Δ q is the qth difference;

[0045] Among them, the pixel point with coordinates (i, j) is defined as the original pixel point, and the 8 pixel points with a step size of 2 from the original pixel point are defined as the adjacent pixel points of the original pixel point. After a certain adjacent pixel point is taken as the first adjacent pixel point, the rest are arranged clockwise as the second to eighth adjacent pixel points. The difference is the difference between the measured temperature of the original pixel point and the measured temperature of the adjacent pixel point, that is, Δ1 is the measured temperature difference between the original pixel point and the first adjacent pixel point, and the remaining definitions are similar.

[0046] The measured temperature of the pixel with coordinates (i, j)

[0047]

[0048] In the formula, β q is the measured temperature of the qth adjacent pixel.

[0049] Step 2: Identify the area with sudden temperature change in the temperature image on the surface of the metal plate, which is the crack location;

[0050] Step 3: Collect the changes of the primary magnetic field and the secondary induced magnetic field through the Hall sensor, and then convert the magnetic signal into a voltage signal to obtain the peak potential difference of the rising edge and the peak potential difference of the falling edge;

[0051] Wherein, the primary magnetic field and the secondary induced magnetic field both satisfy:

[0052] B=μ0μ r H;

[0053] Where B is the magnetic flux density, μ0 is the vacuum magnetic permeability, μ r is the relative magnetic permeability, H is the electric field strength;

[0054] The Hall sensor satisfies:

[0055]

[0056] Where U v is the Hall voltage, R H is the Hall coefficient, I is the loading current intensity, and t is the thickness of the metal plate;

[0057] In this embodiment, since the actual crack may be relatively fine and contain noise, a filter circuit and an amplifier circuit are connected after the Hall sensor to filter the noise and amplify the magnetic signal at the same time.

[0058] The peak potential difference along the rising edge satisfies:

[0059]

[0060] The peak potential difference on the falling edge satisfies:

[0061]

[0062] In the formula, is the minimum difference between the Hall voltage of the primary magnetic field and the secondary induced magnetic field, It is the maximum difference of Hall voltage between the primary magnetic field and the secondary induced magnetic field.

[0063] Step 4: Construct a crack quantitative detection model, including an input layer, a first convolutional layer, a first batch of normalization layers, a second convolutional layer, a second batch of normalization layers, a discard layer, a linear rectification layer, a fully connected layer, and an output layer;

[0064] The input layer is the rising edge peak potential difference and the falling edge peak potential difference, and the output layer is the crack depth.

[0065] The loss function satisfies:

[0066]

[0067] In the formula, is the actual crack depth of the y-th pixel, is the predicted crack depth of the y-th pixel.

[0068] The quantitative crack detection model is trained using simulation data or artificial crack data, and the model parameters are updated by combining forward propagation and back propagation until the loss function converges or the number of iterations is reached.

[0069] In this embodiment, the number of iterations is 500.

[0070] The present invention designs and develops a nondestructive testing method for metal plates, which realizes precise positioning of cracks in metal plates through pulse excitation, and accurately detects the depth of cracks through the pressure difference of eddy current magnetic field combined with convolutional neural network. The recognition rate of crack positioning can reach more than 97%, and the maximum value of crack depth detection error is no more than 2%, which improves the efficiency of identification and detection on the basis of ensuring accuracy. The nondestructive testing method for metal plates described in the present invention is mainly suitable for nondestructive testing of non-ferromagnetic metal plates.

[0071] Although the embodiments of the present invention have been disclosed as above, they are not limited to the applications listed in the specification and the implementation modes. They can be fully applied to various fields suitable for the present invention. For those familiar with the art, additional modifications can be easily implemented. Therefore, without departing from the general concept defined by the claims and the scope of equivalents, the present invention is not limited to the specific details and the embodiments shown and described herein.

Claims

1. A nondestructive testing method for a metal plate, characterized in that: The steps include: Step 1: The pulse excitation signal generating system heats the metal plate, collects the surface temperature of the metal plate, and obtains a temperature image of the surface of the metal plate; Step 2: Identify the area with sudden temperature change in the temperature image on the surface of the metal plate to obtain the crack position; Step 3: Collect the changes of the primary magnetic field and the secondary induced magnetic field to obtain the peak potential difference along the rising edge and the peak potential difference along the falling edge; Step 4: construct a crack quantitative detection model, input the rising edge peak potential difference and the falling edge peak potential difference into the crack quantitative detection model, and obtain the crack depth; The crack quantitative detection model includes an input layer, a first convolutional layer, a first batch of normalization layers, a second convolutional layer, a second batch of normalization layers, a discard layer, a linear rectification layer, a fully connected layer and an output layer.

2. The nondestructive testing method for metal plates according to claim 1, characterized in that: Also includes: The temperature image of the metal plate surface is corrected, and the corrected temperature image of the metal plate surface satisfies: Where, T i, ' j Represents the corrected temperature of the pixel with coordinates (i, j), T i,j represents the measured temperature of the pixel with coordinates (i, j), Δ q is the qth difference.

3. The nondestructive testing method for metal plates according to claim 2, characterized in that: The measured temperature of the pixel point with coordinates (i, j) satisfies: In the formula, β q is the measured temperature of the qth adjacent pixel.

4. The nondestructive testing method for metal plates according to claim 3, characterized in that: The adjacent pixel points are 8 pixel points with a step length of 2 from the pixel point with coordinates (i, j), and one adjacent pixel point is fixed as the first adjacent pixel point, and the rest are arranged clockwise as the second to eighth adjacent pixel points.

5. The nondestructive testing method for metal plates according to claim 4, characterized in that: The step three collects the changes of the primary magnetic field and the secondary induced magnetic field through the Hall sensor.

6. The nondestructive testing method for metal plates according to claim 5, characterized in that: The magnetic flux densities of the primary magnetic field and the secondary induced magnetic field both satisfy: B=μ0μ r H; Where B is the magnetic flux density, μ0 is the vacuum magnetic permeability, μ r is the relative magnetic permeability, and H is the electric field strength.

7. The nondestructive testing method for metal plates according to claim 6, characterized in that: The Hall sensor satisfies: Where U v is the Hall voltage, R H is the Hall coefficient, I is the loading current intensity, and t is the thickness of the metal plate.

8. The nondestructive testing method for metal plates according to claim 7, characterized in that: The peak potential difference along the rising edge satisfies: In the formula, It is the minimum difference of Hall voltage between the primary magnetic field and the secondary induced magnetic field.

9. The nondestructive testing method for metal plates according to claim 8, characterized in that: The falling edge peak potential difference satisfies: In the formula, It is the maximum difference of Hall voltage between the primary magnetic field and the secondary induced magnetic field.

10. The nondestructive testing method for metal plates according to claim 9, characterized in that: The loss function of the crack quantitative detection model satisfies: In the formula, is the actual crack depth of the y-th pixel, is the predicted crack depth of the y-th pixel.

Images