Ferromagnetic material average grain size rapid characterization method based on magnetic Barkhausen electromagnetic nondestructive testing

Abstract

The invention discloses a method for rapidly characterizing the average grain size of a ferromagnetic material based on magnetic Barkhausen electromagnetic nondestructive testing. The method comprises the following steps: preparing a sample; based on an EBSD material characterization method, carrying out quantitative characterization on the average grain size of the ferromagnetic material; based on magnetic Barkhausen noise detection equipment, electromagnetic nondestructive detection is carried out on the ferromagnetic material, and electromagnetic characteristic parameters representing the magnetic characteristics of the ferromagnetic material are obtained; and by taking the electromagnetic characteristic parameters as input and the average grain size as output, establishing a characteristic microstructure deep neural network characterization model of the magnetic Barkhausen characteristic parameters so as to realize nondestructive testing of the characteristic microstructure-average grain size of the ferromagnetic material. According to the invention, nondestructive testing of the average grain size of the ferromagnetic material is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the electromagnetic nondestructive testing technology of ferromagnetic materials, and more particularly to a method for rapidly characterizing the average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing. BACKGROUND

[0002] The grain size of steel materials in a steady state is in accordance with the hell-pitch relationship with the yield strength of the materials, that is, the smaller the grain size, the greater the strength and hardness. In addition, the grain size also has an impact on the corrosion resistance of metal materials. At present, domestic and foreign steel plants mainly use destructive optical metallography, EBSD (Electron Back Scatter Diffraction, EBSD) and other means for sampling detection of the grain size of the produced steel materials. Such destructive detection method is time-consuming, dependent on laboratories and does not conform to the concept of smart factory. On the other hand, the grain size will affect the dislocation density and magnetic domain structure and movement characteristics. Under the excitation of an external magnetic field, the electromagnetic characteristics (obtained by electromagnetic nondestructive testing technology) of ferromagnetic materials with different grain sizes are different. By analyzing the electromagnetic signals, the grain size of ferromagnetic materials can be nondestructively characterized.

[0003] Most of the current technologies focus on the mechanical property detection method of ferromagnetic materials, and the microstructure method of ferromagnetic materials is also limited to destructive testing characterization technology. Therefore, no nondestructive testing technology for directly characterizing the average grain size of the microstructure of ferromagnetic materials has been found. SUMMARY

[0004] In view of the defects in the prior art, the present application aims to provide a method for rapidly characterizing the average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing, so as to realize nondestructive testing of the average grain size of ferromagnetic materials.

[0005] To achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0006] A method for rapidly characterizing the average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing comprises the following steps:

[0007] Preparation of a sample;

[0008] Quantitative characterization of the average grain size of ferromagnetic materials based on the EBSD material characterization method;

[0009] Electromagnetic nondestructive testing of ferromagnetic materials based on a magnetic Barkhausen noise detection device to obtain electromagnetic characteristic parameters representing the magnetic properties of the ferromagnetic materials;

[0010] Using electromagnetic characteristic parameters as input and average grain size as output, a deep neural network characterization model of the characteristic microstructure of magnetic Barkhausen characteristic parameters is established to achieve non-destructive testing of the characteristic microstructure and average grain size of ferromagnetic materials.

[0011] Preferably, the rapid characterization method for the average grain size of the ferromagnetic material specifically includes the following steps:

[0012] S1, Prepare samples for EBSD material characterization methods and samples for magnetic Barkhausen noise detection;

[0013] S2, The average grain size of ferromagnetic materials is quantitatively characterized using the EBSD material characterization method;

[0014] S3. Repeat steps S1 and S2 to obtain the quantitative characterization results of the average grain size of different ferromagnetic materials.

[0015] S4. Based on step S1, electromagnetic non-destructive testing is performed on different ferromagnetic materials using a magnetic Barkhausen noise detection device, and the electromagnetic characteristic parameters of the ferromagnetic materials are extracted.

[0016] S5. Based on the quantitative characterization results in step S3, the average grain size is obtained and a correlation is established with the electromagnetic characteristic parameters of the ferromagnetic material obtained in step S4.

[0017] S6. Establish a deep neural network representation model of the characteristic microstructure of magnetic Barkhausen feature parameters;

[0018] S7. The confidence level is used to evaluate the accuracy of the deep neural network representation model of the characteristic microstructure of magnetic Barkhausen feature parameters.

[0019] S8. Following the process of steps S6 and S7, the deep neural network representation model of the characteristic microstructure of the magnetic Barkhausen feature parameters is trained ten times, and the average of the ten training results is taken as the final detection value.

[0020] Preferably, in step S1, the size of the sample used for the EBSD material characterization method is 10mm*10mm;

[0021] The sample size used for magnetic Barkhausen noise detection is 150mm*600mm.

[0022] Preferably, step S4 specifically includes:

[0023] The test direction of the magnetic Barkhausen noise detection equipment is parallel to the rolling direction of the sample. During the test, the programmable chip AD9833 outputs a sinusoidal voltage with a frequency of 10Hz and an amplitude of 600mV, which is amplified into a sinusoidal voltage with a frequency of 10Hz and an amplitude of 6V through the isolation capacitor.

[0024] Then, a sinusoidal voltage is applied to the U-shaped yoke around the coil;

[0025] During the magnetization process, an MBN sensor is used to collect MBN pulse signals.

[0026] Preferably, the MBN sensor's acquisition card has a acquisition frequency of 15–40 kHz, followed by bandpass filtering and amplification.

[0027] Preferably, the MBN sensor uses an equal-interval sampling and smoothing filtering algorithm to perform preliminary processing on the MBN pulse signal, and then uses the Hilbert-Huang transform algorithm to extract the intercepted MBN pulse signal and the smoothed envelope signal of the MBN pulse signal, and then extracts electromagnetic feature parameters from the MBN pulse signal and the smoothed envelope signal.

[0028] Preferably, the electromagnetic characteristic parameters include root mean square, ring count, half-width at half-maximum, peak value, peak time, and mean value.

[0029] Preferably, the root mean square (RMS) is the RMS value of the MBN pulse signal within one period, calculated using the following formula:

[0030]

[0031] In the formula, x is the amplitude of the MBN pulse signal, and n is the number of MBN pulse signals;

[0032] The ring count is the number of pulses whose amplitude of the MBN pulse signal exceeds a certain threshold from bottom to top, determined by analyzing the MBN pulse signal. This threshold is 30% of the peak value.

[0033] The peak time is the difference between the time corresponding to the minimum point and the time corresponding to the maximum point of the smooth envelope signal, obtained by analyzing the smooth envelope signal.

[0034] The half-height width is the distance between the intersection points of the smooth envelope signal when the vertical coordinate value is 50% of the maximum value of the smooth envelope signal, calculated by analyzing the smooth envelope signal.

[0035] Preferably, in step S6, the learning algorithm flow of the deep neural network representation model of the characteristic microstructure of the magnetic Barkhausen feature parameters is as follows:

[0036] S61, given the input vector and the output vector;

[0037] S62, Initialize weights and thresholds;

[0038] S63, calculate the output of each node in the hidden layer and the output layer;

[0039] S64, calculate the error E between the expected output and the actual output;

[0040] S65, determine whether the error E meets the requirements. If yes, end; otherwise, proceed to step S66.

[0041] S66, calculate the error of each unit in the hidden layer and the output layer;

[0042] S67, Calculate the error gradient;

[0043] S68, after updating the weights and thresholds of the hidden layer and the output layer, return to step S63.

[0044] Preferably, the confidence level in step S7 is defined as:

[0045]

[0046] In the formula, N 10 To verify the predicted value Y of the ensemble model ci Compared with the true value Y i The number of validation sets with an error ρ less than 10%, where N is the total number of validation sets, i.e.:

[0047]

[0048] The present invention provides a rapid characterization method for the average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing, which has the following advantages:

[0049] (1) To achieve non-destructive testing of the characteristic microstructure and average grain size of ferromagnetic materials, with a testing accuracy greater than 90%, an error less than 10%, and a confidence rate greater than 85%.

[0050] (2) It provides a digital and rapid evaluation method for characterizing the microstructure of ferromagnetic materials, thereby improving detection efficiency and reducing detection costs. Attached Figure Description

[0051] Figure 1 This is a flowchart illustrating the rapid characterization method for average grain size of ferromagnetic materials according to the present invention.

[0052] Figure 2 This is a schematic diagram of the EBSD characterization results of the microstructure in step S2 of the rapid characterization method for average grain size of ferromagnetic materials in this invention.

[0053] Figure 3 This is a schematic diagram of the magnetic Barkhausen noise detection device in step S3 of the rapid characterization method for average grain size of ferromagnetic materials in this invention.

[0054] Figure 4 This is a schematic diagram of a typical MBN pulse signal in step S4 of the rapid characterization method for average grain size of ferromagnetic materials in this invention. (a) is the waveform of the MBN pulse signal, and (b) is the waveform of the smooth envelope signal of the MBN pulse signal.

[0055] Figure 5 This is a schematic diagram of the ringing number in the rapid characterization method for average grain size of ferromagnetic materials of the present invention;

[0056] Figure 6 This is a schematic diagram of the half-width at half-maximum (WHM) in the rapid characterization method for average grain size of ferromagnetic materials of the present invention.

[0057] Figure 7 This is a schematic diagram of the relationship between the electromagnetic characteristic parameters and characteristic microstructure of ferromagnetic materials in step S5 of the rapid characterization method for average grain size of ferromagnetic materials in this invention.

[0058] Figure 8 This is a schematic diagram of the topology of the deep neural network characterization model of the characteristic microstructure of the magnetic Barkhausen feature parameters in the rapid characterization method of average grain size of ferromagnetic materials of the present invention.

[0059] Figure 9 This is a schematic diagram of the learning algorithm flow of the deep neural network characterization model of the characteristic microstructure of magnetic Barkhausen feature parameters in the rapid characterization method of average grain size of ferromagnetic materials of the present invention.

[0060] Figure 10 This is a schematic diagram of the evaluation results in step S8 of the rapid characterization method for average grain size of ferromagnetic materials in this invention. Detailed Implementation

[0061] To better understand the above-mentioned technical solutions of the present invention, the technical solutions of the present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0062] This invention provides a rapid characterization method for the average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing:

[0063] First, the ferromagnetic material was cut into small-sized samples of 10mm*10mm (for EBSD material characterization method) and large-sized samples of 150mm*600mm (for magnetic Barkhausen noise detection) using wire cutting.

[0064] Secondly, based on the EBSD material characterization method, the average grain size of ferromagnetic materials is quantitatively characterized.

[0065] Then, based on the magnetic Barkhausen noise detection equipment (MBN), electromagnetic non-destructive testing is performed on the ferromagnetic material to obtain the MBN pulse signal, and at the same time, the MBN electromagnetic characteristic parameters of the ferromagnetic material are defined and extracted.

[0066] Finally, by analyzing the correlation between the MBN electromagnetic characteristic parameters of ferromagnetic materials and the average grain size of ferromagnetic materials, a deep neural network characterization model of the characteristic microstructure of magnetic Barkhausen characteristic parameters is established, using the MBN electromagnetic characteristic parameters as input and the average grain size as output. The parameters of the deep neural network characterization model of the characteristic microstructure of magnetic Barkhausen characteristic parameters are then adjusted to obtain the optimal characterization model, so as to realize the non-destructive testing of the characteristic microstructure of ferromagnetic materials and the average grain size.

[0067] The rapid characterization method for average grain size of ferromagnetic materials of the present invention specifically includes the following steps:

[0068] S1 uses wire cutting to cut ferromagnetic materials into small-sized samples of 10mm*10mm (for EBSD material characterization methods) and large-sized samples of 150mm*600mm (for magnetic Barkhausen noise detection).

[0069] S2, The average grain size of the ferromagnetic material was quantitatively characterized using the EBSD material characterization method. The quantitative characterization results are as follows: Figure 2 As shown.

[0070] S3. Repeat steps S1 and S2 to obtain the quantitative characterization results of the average grain size of different ferromagnetic materials.

[0071] S4, based on step S1, uses a magnetic Barkhausen noise detection device to perform electromagnetic non-destructive testing on different ferromagnetic materials and extracts the electromagnetic characteristic parameters of the ferromagnetic materials. The magnetic Barkhausen noise detection device is as follows: Figure 3 As shown, it mainly consists of an excitation circuit, a magnetic sensor, a signal conditioning circuit, an MBN signal acquisition circuit, and a scanning mechanism drive section. The excitation section comprises a signal generator (AD9833 chip), a power amplifier, an excitation coil, and a magnetic yoke, used to drive the coil to generate a suitable alternating magnetic field to excite the Barkhausen effect and produce an external excitation magnetic field of a certain magnitude. The signal conditioning circuit uses a multi-stage amplifier circuit, including an RC filter circuit, and the amplifier chip used is the instrumentation amplifier INA111 chip. The acquisition circuit inputs to the acquisition card through a voltage follower circuit, and uses a DAQ-2010(G) acquisition card.

[0072] The test direction (the direction in which the probe generates the magnetic field) is parallel to the rolling direction of the sample. During testing, the AD9833 programmable chip outputs a sinusoidal voltage with a frequency of 10Hz and an amplitude of 600mV, which is amplified to a sinusoidal voltage with a frequency of 10Hz and an amplitude of 6V through an isolation capacitor. This sinusoidal voltage is then applied to a U-shaped yoke (800 turns of 1.5mm wire) around the coil. During magnetization, an MBN sensor (a receiving coil with a wire diameter of 0.1mm, approximately 2500 turns) is used to collect MBN pulse signals. A data acquisition card (DAQ-2010(G), sampling frequency 200kHz) acquires MBN pulse signals with a frequency of 15–40kHz, which are then amplified after bandpass filtering.

[0073] Typical MBN waveforms are as follows: Figure 4 As shown in (a), the MBN pulse signal is initially processed using an equal-interval sampling and smoothing filtering algorithm, and then the intercepted MBN pulse signal (e.g., ...) is extracted using the Hilbert-Huang Transform (HHT) algorithm. Figure 4 (as shown in blue in (a)) and the smoothed envelope signal of the MBN pulse signal (as shown in (a)). Figure 4 (as shown in blue in (b)), electromagnetic characteristic parameters are then extracted from the MBN pulse signal and the smooth envelope signal, as shown in Table 1 below.

[0074] Table 1 MBN Feature Table

[0075]

[0076] Wherein, the root mean square (RMS) is the root mean square value of the MBN pulse signal within one period, and the calculation formula is as follows:

[0077]

[0078] In the formula, x is the amplitude of the MBN pulse signal, and n is the number of MBN pulse signals.

[0079] The ring count is determined by analyzing the MBN pulse signal, specifically the number of pulses whose amplitude exceeds a certain threshold from bottom to top. This threshold is 30% of the peak value. Figure 5 As shown.

[0080] Peak time is the difference between the time corresponding to the minimum and the time corresponding to the maximum of the smooth envelope signal, obtained by analyzing the smooth envelope signal.

[0081] The half-width is calculated by analyzing the smooth envelope signal, using 50% of the maximum value of the smooth envelope signal as a basis, and determining the distance between the intersection points of the smooth envelope signal when the ordinate value is 50% of the maximum value, i.e., x2-x1. Figure 6 Place

[0082] By using a magnetic Barkhausen noise detection device to perform electromagnetic non-destructive testing on ferromagnetic material samples, the six electromagnetic characteristic parameters of MBN mentioned above can be obtained. Each sample is tested for 60 seconds, and more than 100 sets of electromagnetic characteristic data are obtained. After outlier cleaning, 100 sets of data are retained for subsequent analysis.

[0083] S5, Based on the average grain size quantified in step S3, establish a correlation between it and the electromagnetic characteristic parameters of the ferromagnetic material obtained in step S4. Plot the MBN electromagnetic characteristic parameters of the ferromagnetic material on the ordinate and the characteristic microstructure-average grain size of the ferromagnetic material on the abscissa, such as... Figure 7 As shown.

[0084] The electromagnetic characteristic parameters of MBN (peak time, ring count) and average grain size show a monotonically decreasing trend, while the RMS and mean, peak value and average grain size show a monotonically increasing trend, and the half-width at half-maximum (HWHM) trend is not significant. Therefore, except for HWHM, the electromagnetic characteristic parameters extracted by MBN nondestructive testing technology can all be used to characterize the average grain size of ferromagnetic materials.

[0085] S6. Establish a deep neural network representation model of the characteristic microstructure of magnetic Barkhausen feature parameters.

[0086] The BP (Back Propagation) neural network algorithm is a multi-layer feedforward network that learns and trains using an error backpropagation algorithm. Its network topology is as follows: Figure 8 As shown, the network structure consists of three parts: an input layer, hidden layers, and an output layer. There is only one input layer and one output layer, while there can be multiple hidden layers. Each layer consists of several neurons. Neurons within a layer are not interconnected, but neurons between layers are fully connected; that is, any neuron in the input layer is connected to all nodes in the hidden layers, and any neuron in the hidden layers is connected to all nodes in the output layer. Here, we use a BP neural network with one hidden layer as an example to introduce the basic principle of the BP algorithm.

[0087] Figure 8 In the given information, x = (x0, x1, ..., x...) n-1 ) is the input vector, x i ,i=(0,1,…,n-1) represents the input quantity of the i-th neuron in the input layer, b j j = (0, 1, ..., l-1) represents the output of the hidden layer's neuron nodes, y k k = (0, 1, ..., m-1) represents the output nodes of the output layer, n, l, and m represent the number of neurons in the input layer, hidden layer, and output layer, respectively, and v ijw represents the weights from the i-th neuron in the input layer to the j-th neuron in the hidden layer. jk Let represent the weights from the j-th node in the hidden layer to the k-th node in the output layer. Then:

[0088] The outputs of each neuron node in the hidden layer are:

[0089]

[0090] The outputs of each neuron node in the output layer are:

[0091]

[0092] Where, θ j φ k f1 and f2 represent the threshold values ​​for the hidden layer node and the output layer node, respectively; f1 and f2 represent the transfer functions for the hidden layer and the output layer, respectively.

[0093] Assume the expected output of the output layer node is O k The total error is:

[0094]

[0095] The purpose of error backpropagation is to continuously reduce the total error by adjusting the weights. Therefore, the weights should be adjusted along the negative gradient of the error, that is, the change in the weights should satisfy the following equation:

[0096]

[0097] In the two equations above, η is a constant, called the learning rate, which is usually taken as 0 < η < 1.

[0098] Combining the above equations, we can obtain the following formulas for adjusting the weights of the hidden layer and the output layer:

[0099]

[0100] After calculating the new weights and thresholds according to the weight and threshold adjustment formula, a new round of forward propagation process begins.

[0101] The above derivation process is based on a single training sample and a single-hidden-layer BP neural network. The basic idea is equally applicable to multi-hidden-layer BP neural networks. When calculating the adjustment amounts for weights and thresholds, the process proceeds recursively from the output layer through each intermediate hidden layer back to the first hidden layer. The error across multiple training samples is calculated as the cumulative sum of the errors of each individual sample.

[0102] In summary, the execution process of the BP neural network learning algorithm can be summarized into two main parts: forward propagation of the signal and backward propagation of the error. For each training sample, during forward propagation, the input vector is passed from the input layer to the output layer layer by layer; during backward propagation, the error is passed from the hidden layers back to the input layer layer by layer in some form. These two processes are repeated until the set termination condition is met. The algorithm flowchart is as follows: Figure 9 As shown.

[0103] The model uses the MBN electromagnetic characteristic parameters of ferromagnetic materials as input and the characteristic microstructure (true value) of ferromagnetic materials as output. During the training of the deep neural network representation model of the characteristic microstructure of the magnetic Barkhausen characteristic parameters, the "logsig" activation function and the "trainlm" training function are used, with the error threshold set to 0.000001, the initial learning rate set to 0.01, and the number of iterations set to 1000.

[0104] S7 uses confidence scores to evaluate the accuracy of the deep neural network representation model of the characteristic microstructure of the magnetic Barkhausen feature parameters. The confidence score is defined as:

[0105]

[0106] In the formula, N 10 To verify the predicted value Y of the ensemble model ci Compared with the true value Y i The number of validation sets with an error ρ less than 10%, where N is the total number of validation sets, i.e.:

[0107]

[0108] S8. Following steps S6 and S7, the deep neural network representation model of the characteristic microstructure of the magnetic Barkhausen feature parameters is trained ten times, and the average of the ten training results is taken as the final detection value. The evaluation accuracy of the average grain size is: under the condition that the error ρ is less than 10%, the confidence level is greater than 85% (e.g., ...). Figure 10 (As shown).

[0109] Example

[0110] This embodiment uses pickled steel from a cold-rolling production line in a steel plant as an example to illustrate a rapid characterization method for the average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing. The specific process steps are as follows: Figure 1 As shown. The quantitative characterization results of the microstructure are as follows. Figure 2 As shown, the average grain size of the pickled steel can be determined as a characteristic microstructure. The composition of the magnetic Barkhausen noise detection equipment is as follows... Figure 3 As shown. The MBN pulse signal is as follows. Figure 4As shown in Table 1, the extracted feature values ​​are also presented. The relationship between the electromagnetic feature values ​​of pickled steel MBN and its characteristic microstructure is as follows: Figure 7 As shown, the specific steps include the following:

[0111] BP model (with 9 nodes, such as...) Figure 9 A model was established, using the MBN electromagnetic characteristic parameters of pickled steel as input and the average grain size of the characteristic microstructure of pickled steel as output. The training set consisted of 189 data points, and the validation set consisted of 48 data points. During the training of the DNN model, activation functions "logsig", "tansig", and "purelin" and the training function "trainlm" were used. The error threshold was set to 0.000001, the initial learning rate was set to 0.01, and the number of iterations was set to 1000. Following the above modeling and training process, the training was repeated ten times, and the average of the ten training results was taken as the final detection value. The evaluation accuracy of the average grain size was: under the condition that the error was less than 10%, the confidence level could be greater than 85% (e.g., ...). Figure 10 (As shown).

[0112] Those skilled in the art should recognize that the above embodiments are merely illustrative of the present invention and are not intended to limit the present invention. Any variations or modifications to the above embodiments that are within the spirit and essence of the present invention will fall within the scope of the claims of the present invention.

Claims

1. A rapid characterization method for the average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing, characterized in that: Sample preparation; The average grain size of ferromagnetic materials is quantitatively characterized based on the EBSD material characterization method. Electromagnetic nondestructive testing was performed on ferromagnetic materials using a magnetic Barkhausen noise detection device to obtain electromagnetic characteristic parameters that characterize the magnetic properties of the ferromagnetic materials. Using electromagnetic characteristic parameters as input and average grain size as output, a deep neural network characterization model of the characteristic microstructure of magnetic Barkhausen characteristic parameters is established to achieve non-destructive testing of the characteristic microstructure and average grain size of ferromagnetic materials.

2. The rapid characterization method for average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing according to claim 1, characterized in that, The rapid characterization method for the average grain size of ferromagnetic materials specifically includes the following steps: S1, Prepare samples for EBSD material characterization methods and samples for magnetic Barkhausen noise detection; S2, The average grain size of ferromagnetic materials is quantitatively characterized using the EBSD material characterization method; S3. Repeat steps S1 and S2 to obtain the quantitative characterization results of the average grain size of different ferromagnetic materials. S4. Based on step S1, electromagnetic non-destructive testing is performed on different ferromagnetic materials using a magnetic Barkhausen noise detection device, and the electromagnetic characteristic parameters of the ferromagnetic materials are extracted. S5. Based on the quantitative characterization results in step S3, the average grain size is obtained and a correlation is established with the electromagnetic characteristic parameters of the ferromagnetic material obtained in step S4. S6. Establish a deep neural network representation model of the characteristic microstructure of magnetic Barkhausen feature parameters; S7. The confidence level is used to evaluate the accuracy of the deep neural network representation model of the characteristic microstructure of magnetic Barkhausen feature parameters. S8. Following the process of steps S6 and S7, the deep neural network representation model of the characteristic microstructure of the magnetic Barkhausen feature parameters is trained ten times, and the average of the ten training results is taken as the final detection value.

3. The rapid characterization method for average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing according to claim 2, characterized in that: In step S1, the sample size used for the EBSD material characterization method is 10mm*10mm. The sample size used for magnetic Barkhausen noise detection is 150mm*600mm.

4. The rapid characterization method for average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing according to claim 3, characterized in that, Step S4 specifically includes: The test direction of the magnetic Barkhausen noise detection equipment is parallel to the rolling direction of the sample. During the test, the programmable chip AD9833 outputs a sinusoidal voltage with a frequency of 10Hz and an amplitude of 600mV, which is amplified into a sinusoidal voltage with a frequency of 10Hz and an amplitude of 6V through the isolation capacitor. Then, a sinusoidal voltage is applied to the U-shaped yoke around the coil; During the magnetization process, an MBN sensor is used to collect MBN pulse signals.

5. The rapid characterization method for average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing according to claim 4, characterized in that: The MBN sensor's acquisition card has a acquisition frequency of 15–40 kHz, which is then amplified by bandpass filtering.

6. The rapid characterization method for average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing according to claim 5, characterized in that: The MBN sensor uses an equal-interval sampling and smoothing filtering algorithm to perform preliminary processing on the MBN pulse signal. Then, the Hilbert-Huang transform algorithm is used to extract the intercepted MBN pulse signal and the smoothed envelope signal of the MBN pulse signal. Finally, electromagnetic feature parameters are extracted from the MBN pulse signal and the smoothed envelope signal.

7. The rapid characterization method for average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing according to claim 6, characterized in that: The electromagnetic characteristic parameters include root mean square, ring count, half-width at half-maximum, peak value, peak time, and mean value.

8. The rapid characterization method for average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing according to claim 7, characterized in that: The root mean square (RMS) is the root mean square value of the MBN pulse signal within one period, and the calculation formula is as follows: In the formula, x is the amplitude of the MBN pulse signal, and n is the number of MBN pulse signals; The ring count is the number of pulses whose amplitude of the MBN pulse signal exceeds a certain threshold from bottom to top, determined by analyzing the MBN pulse signal. This threshold is 30% of the peak value. The peak time is the difference between the time corresponding to the minimum point and the time corresponding to the maximum point of the smooth envelope signal, obtained by analyzing the smooth envelope signal. The half-height width is the distance between the intersection points of the smooth envelope signal when the vertical coordinate value is 50% of the maximum value of the smooth envelope signal, calculated by analyzing the smooth envelope signal.

9. The rapid characterization method for average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing according to claim 2, characterized in that, In step S6, the learning algorithm flow of the deep neural network representation model of the feature microstructure of the magnetic Barkhausen feature parameters is as follows: S61, given the input vector and the output vector; S62, Initialize weights and thresholds; S63, calculate the output of each node in the hidden layer and the output layer; S64, calculate the error E between the expected output and the actual output; S65, determine whether the error E meets the requirements. If yes, end; otherwise, proceed to step S66. S66, calculate the error of each unit in the hidden layer and the output layer; S67, Calculate the error gradient; S68, after updating the weights and thresholds of the hidden layer and the output layer, return to step S63.

10. The rapid characterization method for average grain size of ferromagnetic materials based on magnetic Barkhausen electromagnetic nondestructive testing according to claim 9, characterized in that, In step S7, the confidence level is defined as: In the formula, N 10 To verify the predicted value Y of the ensemble model ci Compared with the true value Y i The number of validation sets with an error ρ less than 10%, where N is the total number of validation sets, i.e.:

Images