Metal material detection method and device based on multi-mode excitation source

Abstract

The invention relates to the technical field of metal material detection, in particular to a metal material detection method and device based on a multi-mode excitation source. When multi-modal data is adopted for metal material detection, firstly, the defect type of the metal material is determined, then, different multi-modal data feature fusion methods are adopted, when the metal material to be detected is of a single-defect type, feature fusion is carried out by adopting multi-modal features of a convolutional neural network model based on an attention mechanism, and the defect type of the metal material to be detected is detected. When the to-be-detected metal material is of a multi-defect type, performing feature fusion on the multi-modal features by adopting a hierarchical weighted summation method; the feature fusion is more purposeful, the accuracy of the feature fusion is improved, and the accuracy of metal material defect detection is further improved.

Description

Technical Field

[0001] The present invention relates to the technical field of metal material detection, and in particular to a metal material detection method and device based on a multi-mode excitation source. Background Art

[0002] Metal materials are widely used in modern industry, and their quality and performance are directly related to the safety and reliability of products. Traditional metal material detection methods mainly include ultrasonic detection, ray detection, eddy current detection, etc. However, these methods have certain limitations. For example, ultrasonic detection is not accurate enough for qualitative and quantitative analysis of defects, ray detection poses radiation hazards to the environment and personnel, and eddy current detection has limited ability to detect deep defects. Therefore, it is of great significance to develop a detection method and device that is efficient, accurate, non-destructive and applicable to various metal materials. Summary of the Invention

[0003] In order to solve the above technical problems, the present invention provides a metal material detection method and device based on a multi-mode excitation source to solve the problems existing in the prior art.

[0004] The present invention provides a metal material detection method based on a multi-mode excitation source, including the following steps: S1: Set a multi-mode excitation source for metal material detection; S2: Collect multi-mode excitation signals through a detection unit; S3: Perform data preprocessing operations on the multi-mode excitation signals to obtain preprocessed signals; S4: Perform feature fusion operations on the preprocessed signals to obtain multi-modal fusion features; The specific content of S4 is as follows: S4.1: Extract deep features from the preprocessed signals to obtain ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features; S4.2: Input the ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features into the metal material defect model classification model, and output the classification result of the metal material to be detected; S4.3: Select a feature fusion method for the ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features according to the classification result, and output multi-modal fusion features; S5: Input the multi-modal fusion features into a defect detection model to output the defects of the metal material to be detected.

[0005] Preferably, the specific content of S4.1 is as follows: Input the preprocessed ultrasonic signal, electromagnetic signal, thermal imaging signal, and vibration signal into a feature extraction model to obtain ultrasonic signal feature F u , electromagnetic signal feature F e, the thermal imaging signal feature F t and the vibration signal feature F v .

[0006] Preferably, in the S4.2, the metal material defect classification model is a convolutional neural network model. The input of the convolutional neural network model is the ultrasonic signal feature, the electromagnetic signal feature, the thermal imaging signal feature, and the vibration signal feature. The classification result of the metal material to be tested is that the metal material to be tested is of two types: single defect type and multiple defect types.

[0007] Preferably, in the S4.3, when the metal material to be tested is of the single defect type, the feature fusion method is to perform feature fusion on the ultrasonic signal feature, the electromagnetic signal feature, the thermal imaging signal feature, and the vibration signal feature through a convolutional neural network model based on the attention mechanism.

[0008] Preferably, in the S4.3, when the metal material to be tested is of the multiple defect type, the feature fusion method is to perform feature fusion on the ultrasonic signal feature, the electromagnetic signal feature, the thermal imaging signal feature, and the vibration signal feature through the hierarchical weighted summation method.

[0009] Preferably, the attention mechanism is specifically: Calculate the attention weight of each modal feature; The formula is:

[0010] ; In the formula, a i is the attention score of each modal feature, W a and b a are learnable parameters, F i is the feature of the i-th modal, is the attention weight of each modal feature; Perform feature fusion on the ultrasonic signal feature, the electromagnetic signal feature, the thermal imaging signal feature, and the vibration signal feature according to the attention weight of each modal feature to obtain the multi-modal fusion feature; ; In the formula, F fused is the multi-modal fusion feature after fusion.

[0011] Preferably, the hierarchical weighted summation method is specifically: Align the ultrasonic signal feature, the electromagnetic signal feature, the thermal imaging signal feature, and the vibration signal feature to the same size; perform primary fusion on the aligned features to obtain the primary fusion feature; perform secondary fusion on the primary fusion feature to obtain the final multi-modal fusion feature.

[0012] Preferably, the primary fusion is specifically: performing a preliminary weighted sum on the features of each modality; Assume that the primary fusion weight of each modality is w u , w e , w t , w v , then the formula for the primary fusion feature F primary is: ; Among them, the primary fusion weight of each modality feature is determined according to the statistical characteristics of each modality feature; specifically: ; ; ; .

[0013] In the formula, , , , are the variances of the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature respectively; Performing a further weighted sum on the primary fusion feature to obtain the secondary fusion feature F mid , assuming the secondary fusion weight is w mid : ; Among them, the formula for the secondary fusion weight is: ; In the formula, α and β are learnable parameters.

[0014] Preferably, the method for determining α and β is: first, assign initial values to α and β using random numbers, perform multiple training iterations on the convolutional neural network model using the training set. In each iteration, perform forward propagation on the training set, calculate the output and loss value of the convolutional neural network model, and calculate the gradient of the loss value with respect to each parameter through the backpropagation algorithm. Then use the Adam optimizer to update the values of α and β until the loss value meets the requirements, and output the values of α and β at this time as the final values.

[0015] According to another aspect of the present invention, there is provided a metal material detection device based on a multi-mode excitation source. The device uses the above-mentioned metal material detection method based on a multi-mode excitation source. The device includes: Multi-mode excitation source; Detection unit for collecting multi-mode excitation signals; Preprocessing unit for performing data preprocessing operations on the multi-mode excitation signals to obtain preprocessed signals; Feature fusion unit for performing feature fusion operations on the preprocessed signals to obtain multi-modal fusion features; Defect diagnosis unit for inputting the multi-modal fusion features into a defect detection model and outputting the defects of the metal material to be tested.

[0016] The embodiments of the present invention have the following technical effects: When the present invention uses multi-modal data for metal material detection, it first determines the defect types of the metal materials, and then adopts different multi-modal data feature fusion methods. When the metal material to be tested is of a single defect type, a convolutional neural network model multi-modal feature based on the attention mechanism is used for feature fusion. When the metal material to be tested is of multiple defect types, a hierarchical weighted summation method is used for feature fusion of the multi-modal features; this makes the feature fusion more targeted, improves the accuracy of feature fusion, and further improves the accuracy of metal material defect detection; At the same time, when the metal material to be tested is of multiple defect types, through hierarchical weighted summation fusion and dynamic weight adjustment, the primary fusion weights are determined based on the inherent characteristics of the multi-modal data when the primary weights are determined, and then secondary feature fusion is performed based on deep learning methods, making the feature fusion more accurate when facing complex defects. Brief Description of the Drawings

[0017] In order to more clearly illustrate the specific embodiments of the present invention or the technical solutions in the prior art, the following will briefly introduce the drawings required for use in the description of the specific embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0018] Figure 1 It is a flowchart of a metal material detection method based on a multi-mode excitation source provided by an embodiment of the present invention; Figure 2 It is a flowchart of performing feature fusion operations on the preprocessed signals to obtain multi-modal fusion features provided by an embodiment of the present invention. Detailed Embodiments

[0019] To make the objectives, technical solutions and advantages of the present invention clearer, the technical solutions of the present invention will be described clearly and completely below. Apparently, the described embodiments are only a part rather than all of the embodiments of the present invention. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the scope protected by the present invention.

[0020] Appendix Figure 1 shows a flowchart of a metal material detection method based on a multi-mode excitation source. As shown in the appendix Figure 1 shown, a metal material detection method based on a multi-mode excitation source includes the following steps: S1: Set up a multi-mode excitation source for metal material detection; In the metal material detection method based on a multi-mode excitation source, the excitation source is one of the core parts of the detection system. Its function is to excite the response signals inside the metal material through different physical methods, so as to provide basic data for subsequent detection.

[0021] Among them, the multi-mode excitation source includes one or several of an ultrasonic source, a mechanical vibration source, a heat source, and an electromagnetic field source; Among them, the ultrasonic source can emit high-frequency mechanical waves, which can propagate and reflect in the metal material. When the ultrasonic waves encounter defects (such as cracks, pores, etc.) inside the material, reflected waves and transmitted waves will be generated. By detecting the propagation characteristics of these waves, the internal structure information of the material can be obtained; the electromagnetic field source generates an alternating electromagnetic field to cause eddy currents inside the metal material. The distribution and intensity of the eddy currents are closely related to the conductivity, magnetic permeability, and internal defects of the material. By detecting the changes in the eddy currents, the electromagnetic characteristics of the material can be analyzed; the heat source locally heats the metal material to cause a thermal response. Defects (such as cracks, holes, etc.) inside the material will affect heat conduction, resulting in abnormal local temperature distribution. By detecting these temperature changes through thermal imaging technology, the defects can be located; the mechanical vibration source applies an external vibration to cause a vibration response of the metal material. Defects inside the material will affect the propagation and attenuation characteristics of the vibration. By detecting parameters such as the frequency and amplitude of the vibration signal, the internal structure of the material can be analyzed.

[0022] Further, the parameters of the excitation source include that the frequency range of the ultrasonic wave is 1 MHz - 20 MHz, and a suitable frequency is selected according to the detection target. For example, when detecting thicker metal materials, a lower frequency (such as 1 MHz - 5 MHz) is selected; when detecting thinner metal materials or micro-defects, a higher frequency (such as 10 MHz - 20 MHz) is selected. The frequency range of the electromagnetic field source is 10 Hz - 1 MHz, and a suitable frequency is selected according to the conductivity and detection depth of the material. For example, when detecting surface and near-surface defects, a higher frequency (such as 100 kHz - 1 MHz) is selected; when detecting internal defects, a lower frequency (such as 10 Hz - 10 kHz) is selected. The heating temperature range of the heat source is 50°C - 300°C, and a suitable temperature is selected according to the heat conduction characteristics of the material and the detection requirements. For example, when detecting the change of thermal resistance, a higher heating temperature (such as 200°C - 300°C) is selected; when detecting the heat conduction characteristics, a lower heating temperature (such as 50°C - 100°C) is selected. The frequency range of the mechanical vibration source is 10 Hz - 200 Hz, and a suitable frequency is selected according to the elastic modulus of the material and the detection target. For example, when detecting internal cracks, a lower frequency (such as 10 Hz - 50 Hz) is selected; when detecting high-frequency vibration characteristics, a higher frequency (such as 100 Hz - 200 Hz) is selected.

[0023] S2: Collect multi-mode excitation signals through the detection unit; The purpose of signal collection is to collect the response signals of the metal material under the action of different excitation sources, and these signals contain important information such as the internal structure, defect characteristics, and composition distribution of the metal material.

[0024] Among them, the collection of multi-mode excitation signals includes ultrasonic signal collection, vibration signal collection, thermal imaging signal collection, and electromagnetic signal collection.

[0025] Further, a piezoelectric ultrasonic sensor is used to collect ultrasonic signals, which can convert ultrasonic signals into electrical signals; an acceleration sensor or a displacement sensor is used to collect vibration signals, which can convert vibration signals into electrical signals; an infrared thermal imager is used to collect thermal imaging signals, which can convert the temperature distribution into a thermal imaging image; an electromagnetic induction sensor, such as a coil sensor, is used to collect electromagnetic signals, which can detect the change of eddy current signals. Therefore, the detection unit includes a piezoelectric ultrasonic sensor, an acceleration sensor or a displacement sensor, an infrared thermal imager, and a magnetic induction sensor.

[0026] Among them, when collecting multi-modal excitation signals, in some complex detection scenarios (such as weak response signals), according to the real-time feedback during the detection process, the signal acquisition parameters are dynamically adjusted; for example, if the signal-to-noise ratio of the signal collected by the ultrasonic detection unit is low, the intensity of the ultrasonic source can be appropriately increased or the emission frequency can be adjusted; if the temperature distribution image collected by the thermal imaging detection unit is not clear, the heating temperature or heating time of the heat source can be adjusted.

[0027] S3: Perform data preprocessing operations on the multi-modal excitation signals to obtain preprocessed signals; Preprocessing the collected signals is a key step in signal processing. Its purpose is to remove noise and interference signals, extract effective signal features, and provide high-quality data for subsequent analysis and processing; during the signal acquisition process, it is inevitably affected by external noise and internal interference, such as electromagnetic interference, environmental vibration, sensor noise, etc. Preprocessing can effectively remove this noise and improve the signal-to-noise ratio of the signal.

[0028] Among them, the data preprocessing operations include filtering and digital conversion; Among them, the filtering process is specifically: filtering the ultrasonic signal with a band-pass filter, setting the passband frequency to 1 MHz - 5 MHz; filtering the electromagnetic signal with a high-pass filter, performing mean filtering on the thermal imaging signal, and performing median filtering on the vibration signal; The digital conversion is to convert the ultrasonic signal, electromagnetic signal, and vibration signal into digital signals through an analog-to-digital converter (ADC).

[0029] S4: Perform feature fusion operations on the preprocessed signals to obtain multi-modal fusion features; In the detection of metal materials, signals of multiple physical fields are usually collected, such as ultrasonic signals, electromagnetic signals, thermal imaging signals, and vibration signals, etc. These signals each contain different information about the internal structure and defects of the material. In order to comprehensively analyze the characteristics of the material, it is necessary to fuse these multi-modal data. Traditional feature fusion methods (such as principal component analysis, neural networks, etc.) have certain limitations when dealing with multi-modal data, such as insufficient feature extraction and unsatisfactory fusion effects. Therefore, this embodiment proposes a multi-modal feature fusion and analysis method based on deep learning to improve the accuracy of feature fusion and the depth of analysis.

[0030] Specifically, as shown in the appendix Figure 2 shown, the S4 is specifically: S4.1: Perform deep feature extraction on the preprocessed signals to obtain ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features; Among them, the specific content of S4.1 is as follows: input the preprocessed ultrasonic signal, electromagnetic signal, thermal imaging signal, and vibration signal into the feature extraction model to obtain the ultrasonic signal feature F u , electromagnetic signal feature F e , thermal imaging signal feature F t , and vibration signal feature F v .

[0031] In this embodiment, the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature are represented in the form of feature vectors; The feature extraction model is a convolutional neural network model (CNN).

[0032] S4.2: Input the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature into the metal material defect model classification model, and output the classification result of the metal material to be tested; Among them, the metal material defect classification model is a convolutional neural network model. The input of the convolutional neural network model is the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature. The classification result of the metal material to be tested is that the metal material to be tested is of single defect type and multi-defect type; Among them, the single defect type means that there is one defect in the detection part of the metal material to be tested; the multi-defect type means that there are at least two defects in the detection part of the metal material to be tested, including cracks, holes, corrosion, inclusions, uneven composition distribution, etc.; S4.3: Select a feature fusion method for the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature according to the classification result, and output the multi-modal fusion feature; When the metal material to be tested is of single defect type, the feature fusion method is to perform feature fusion on the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature through a convolutional neural network model based on the attention mechanism; Specifically, the attention mechanism is as follows: Calculate the attention weight of each modal feature; The formula is:

[0033] ; In the formula, a i is the attention score of each modal feature, W a and b a are learnable parameters, F i is the feature of the i-th modal, is the attention weight for each modal feature; Perform feature fusion on the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature according to the attention weight of each modal feature to obtain a multi-modal fusion feature; ; In the formula, F fused is the fused multi-modal feature; In this step, the multi-modal data is learned through deep feature extraction and attention mechanism to obtain the fused multi-modal feature, which improves the accuracy of multi-modal feature fusion.

[0034] When the metal material to be measured is of multiple defect types, the feature fusion method is to perform feature fusion on the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature by the hierarchical weighted summation method; Among them, the hierarchical weighted summation method is specifically: Align the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature to the same size; Specifically, perform alignment operations on the sizes of the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature through upsampling; The specific formula is:

[0035] In the formula, 、 、 、 are the aligned features; Perform primary fusion on the aligned features to obtain a primary fusion feature; Among them, the primary fusion is specifically: perform preliminary weighted summation on the features of each modality; Assume that the primary fusion weight of each modality is w u , w e , w t , w v , then the calculation formula of the primary fusion feature F primary is: ; Among them, the primary fusion weight of each modal feature is determined according to the statistical characteristics of each modal feature; specifically: ; ; ; ; In the formula, , , , are the variances of the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature, respectively; Perform further weighted summation on the primary fusion features to obtain the secondary fusion feature F mid , assuming the secondary fusion weight is w mid : ; Among them, the calculation formula for the secondary fusion weight is: ; In the formula, α and β are learnable parameters.

[0036] The determination method is as follows: First, assign initial values to α and β using random numbers, and perform multiple training iterations (epochs) on the convolutional neural network model using the training set. In each iteration, perform forward propagation on the training set, calculate the output and loss value of the convolutional neural network model, and calculate the gradient of the loss value with respect to each parameter through the backpropagation algorithm. Then use the Adam optimizer to update the values of α and β until the loss value meets the requirements, and output the values of α and β at this time as the final values. In this solution, through hierarchical weighted summation fusion and dynamic weight adjustment, the primary fusion weight is determined based on the inherent characteristics of multi-modal data when determining the primary weight, and then secondary feature fusion is performed based on deep learning methods, making the feature fusion more accurate when facing complex defects.

[0037] S5: Input the multi-modal fusion feature into the defect detection model, and output the defect of the metal material to be tested; Among them, the defect monitoring model is a neural network model, the input of the neural network model is the multi-modal feature fusion feature, and the output of the neural network model is the defect type.

[0038] Embodiment 2. The present invention also provides a metal material detection device based on a multi-mode excitation source. The device adopts a metal material detection method based on a multi-mode excitation source in Embodiment 1. The device includes: Multi-mode excitation source; Detection unit, used to collect multi-mode excitation signals; Pretreatment unit, used to perform data pretreatment operations on the multi-mode excitation signals to obtain the pretreated signals; A feature fusion unit for performing feature fusion operations on the preprocessed signals to obtain multi-modal fusion features; A defect diagnosis unit for inputting the multi-modal fusion features into a defect detection model and outputting the defects of the metal material to be tested.

[0039] Example 3. The present invention also provides an electronic device, including one or more processors and a memory.

[0040] The processor may be a central processing unit (CPU) or other forms of processing units with data processing capabilities and / or instruction execution capabilities, and may control other components in the electronic device to perform desired functions.

[0041] The memory may include one or more computer program products, and the computer program products may include various forms of computer-readable storage media, such as volatile memory and / or non-volatile memory. The volatile memory may include, for example, random access memory (RAM) and / or cache memory, etc. The non-volatile memory may include, for example, read-only memory (ROM), hard disk, flash memory, etc. One or more computer program instructions may be stored on the computer-readable storage media, and the processor may run the program instructions to implement a metal material detection method based on a multi-mode excitation source according to any embodiment of the present application above and / or other desired functions. Various contents such as initial external parameters and thresholds may also be stored in the computer-readable storage media.

[0042] In one example, the electronic device may further include: an input device and an output device, and these components are interconnected through a bus system and / or other forms of connection mechanisms (not shown). The input device may include, for example, a keyboard, a mouse, etc. The output device may output various information to the outside, including warning prompt information, braking force, etc. The output device may include, for example, a display, a speaker, a printer, and a communication network and its connected remote output devices, etc.

[0043] Of course, for simplicity, components such as buses, input / output interfaces, etc. are omitted. In addition, according to specific application scenarios, the electronic device may further include any other appropriate components.

[0044] In addition to the above methods and devices, an embodiment of the present application may also be a computer program product, which includes computer program instructions, and when the computer program instructions are run by the processor, the processor is caused to implement the functions of a metal material detection method based on a multi-mode excitation source provided by any embodiment of the present application.

[0045] The computer program product may be written in any combination of one or more programming languages for executing the program code of the operations of the embodiments of the present application. The programming languages include object-oriented programming languages such as Java, C++, etc., and also include conventional procedural programming languages such as the "C" language or similar programming languages. The program code may be executed entirely on the user's computing device, partially on the user's device, executed as a stand-alone software package, partially on the user's computing device and partially on a remote computing device, or entirely on a remote computing device or server.

[0046] In addition, an embodiment of the present application may also be a computer-readable storage medium having computer program instructions stored thereon, and when the computer program instructions are run by a processor, the processor implements a metal material detection method based on a multi-mode excitation source provided by any embodiment of the present application.

[0047] The computer-readable storage medium may employ any combination of one or more readable media. The readable media may be a readable signal medium or a readable storage medium. The readable storage medium may, for example, include but is not limited to electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or components, or any combination of the above. More specific examples (non-exhaustive list) of the readable storage medium include: an electrical connection having one or more wires, a portable disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0048] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them. Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some or all of the technical features. And these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the technical solutions of the embodiments of the present invention.

Claims

1. A metal material detection method based on a multi-mode excitation source, characterized in that, Including the following steps: S1: Set up a multi-mode excitation source for metal material detection; S2: Collect multi-mode excitation signals through a detection unit; S3: Perform data preprocessing operations on the multi-mode excitation signals to obtain preprocessed signals; S4: Perform feature fusion operations on the preprocessed signals to obtain multi-modal fusion features; Specifically, S4 is as follows: S4.1: Perform deep feature extraction on the preprocessed signals to obtain ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features; S4.2: Input the ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features into the metal material defect model classification model, and output the classification result of the metal material to be tested; S4.3: Select a feature fusion method for the ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features according to the classification result, and output multi-modal fusion features; S5: Input the multi-modal fusion features into the defect detection model to output the defects of the metal material to be tested.

2. The method for detecting metal materials based on a multi-mode excitation source according to claim 1, characterized in that: Specifically, S4.1 is as follows: Input the preprocessed ultrasonic signal, electromagnetic signal, thermal imaging signal, and vibration signal into the feature extraction model to obtain the ultrasonic signal feature F u , electromagnetic signal feature F e , thermal imaging signal feature F t , and vibration signal feature F v .

3. The method for detecting metal materials based on a multi-mode excitation source according to claim 2, characterized in that: In S4.2, the metal material defect classification model is a convolutional neural network model. The input of the convolutional neural network model is the ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features. The classification result of the metal material to be tested is that the metal material to be tested is of two types: single defect type and multi-defect type.

4. The method for detecting metal materials based on a multi-mode excitation source according to claim 3, characterized in that: In S4.3, when the metal material to be tested is of a single defect type, the feature fusion method is to perform feature fusion on the ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features through a convolutional neural network model based on an attention mechanism.

5. The method for detecting metal materials based on a multi-mode excitation source according to claim 3, characterized in that: In S4.3, when the metal material to be tested is of a multi-defect type, the feature fusion method is to perform feature fusion on the ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features through a hierarchical weighted summation method.

6. The method for detecting metal materials based on a multi-mode excitation source according to claim 4, characterized in that: The attention mechanism is specifically: Calculate the attention weights of each modal feature; The formula is: ； ； where a i is the attention score of each modal feature, W a and b a are learnable parameters, F i is the feature of the i-th modality, is the attention weight of each modal feature; Perform feature fusion on the ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features according to the attention weights of each modal feature to obtain multi-modal fusion features; ； Where F fused is the fused multi-modal feature.

7. The method for detecting metal materials based on a multi-mode excitation source according to claim 5, characterized in that: The hierarchical weighted summation method is specifically: Align the ultrasonic signal features, electromagnetic signal features, thermal imaging signal features, and vibration signal features to the same size; perform primary fusion on the aligned features to obtain primary fusion features; perform secondary fusion on the primary fusion features to obtain the final multi-modal fusion features.

8. A metal material detection method based on a multi-mode excitation source according to claim 7, wherein: The primary fusion specifically is: perform a preliminary weighted sum on the features of each modality; Assume that the primary fusion weights for each modality are w u , w e , w t , w v , then the formula for calculating the primary fusion feature F primary is as follows: ； Among them, the primary fusion weight of each modality feature is determined according to the statistical characteristics of each modality feature; specifically: ； ； ； ； In the formula, , , , are the variances of the ultrasonic signal feature, electromagnetic signal feature, thermal imaging signal feature, and vibration signal feature, respectively; Further weighted summation is performed on the primary fusion features to obtain the secondary fusion feature F mid , assuming that the secondary fusion weight is w mid : ； Among them, the formula for the secondary fusion weight is: ； In the formula, α and β are learnable parameters.

9. A metal material detection method based on a multi-mode excitation source according to claim 8, wherein: The determination method of α and β is: first assign initial values to α and β using random numbers, perform multiple training iterations on the convolutional neural network model using the training set. In each iteration, perform forward propagation on the training set, calculate the output and loss value of the convolutional neural network model, and calculate the gradient of the loss value with respect to each parameter through the backpropagation algorithm. Then use the Adam optimizer to update the values of α and β until the loss value meets the requirements, and output the values of α and β at this time as the final values.

10. A metal material detection device based on a multi-mode excitation source, characterized in that, The device adopts a metal material detection method based on a multi-mode excitation source according to any one of claims 1-9. The device includes: A multi-mode excitation source; A detection unit for collecting multi-mode excitation signals; A preprocessing unit for performing data preprocessing operations on the multi-mode excitation signals to obtain preprocessed signals; A feature fusion unit for performing feature fusion operations on the preprocessed signals to obtain multi-modal fusion features; A defect diagnosis unit for inputting the multi-modal fusion features into a defect detection model and outputting the defects of the metal material to be tested.

Images