Absorbing coating defect detection model training method, defect diagnosis method and system

Abstract

The invention discloses a defect detection model training method for an absorbing coating, a defect detection method and a system. The absorbing coating is subjected to microwave heating and a thermal image sequence is obtained. Since the thermal images of a normal area and a defective area are different, feature information of the normal area and the defective area can be extracted. Then, a defect mark is performed on a time-temperature curve of each pixel point in the thermal image sequence according to the feature information. A defect detection model is trained using a training sample consisting of the time-temperature curve and the defect mark. Finally, the trained defect detection model is used to realize intelligent detection of the absorbing coating. The maximum detection depth is 0.3 mm, the minimum diameter-to-depth ratio is 1.5, and pixel-level detection of the absorbing coating is realized. The resonance cavity + microwave heating technology is used to ensure that the microwave maintains a uniform field strength in the resonance cavity. When a test piece or the absorbing coating is heated, large-area uniform heating is realized, thereby improving the heating efficiency and heating effect.

Description

Technical Field

[0001] The present invention relates to a microwave absorbing coating defect detection technology, and in particular to a microwave absorbing coating defect detection model training method based on a resonant cavity and microwave thermal imaging, a defect detection method and a system. Background Art

[0002] Radar-absorbing coatings are widely used in fighter jets and other weapons and equipment because of their advantages such as simple and convenient construction and no restrictions on equipment shape. At high altitudes and high speeds, the temperature of the fuselage is higher than expected, causing the radar-absorbing coating on the surface of the horizontal tail to fall off, greatly affecting the stealth function of the fighter jet. Therefore, defect detection of radar-absorbing coatings becomes particularly important.

[0003] Existing nondestructive testing methods, such as eddy current method, capacitance method, ultrasonic method, classical microwave method and terahertz method, can be applied to the detection and evaluation of absorbing coatings in principle. However, these methods all have some limitations:

[0004] The eddy current method can detect the thickness of non-conductive absorbing coatings, but due to the limitations of basic principles, it cannot detect damage inside the non-conductive absorbing coatings and damage between the metal substrate and the coating interface, and imaging is time-consuming and the detection efficiency is low; the capacitance method can detect the thickness and internal damage of non-conductive absorbing coatings, but it is difficult to detect damage in the metal substrate; the biggest disadvantage of the ultrasonic method is that it requires a coupling agent and there is a detection blind spot, which makes it difficult to apply it to in-situ in-service non-destructive testing of coatings. In addition, the special structure of the absorbent particles in the absorbing coating leads to complex ultrasonic echo and reflection phenomena, making it difficult to distinguish damage information; the classical microwave method is mainly used to test the absorbing properties of absorbing coatings, but the microwave network analyzer has a high procurement cost and a large size, making it difficult to use for in-situ non-destructive testing; the biggest disadvantage of the terahertz method is the long imaging time, complex system and high cost.

[0005] Microwave thermal imaging technology is a large-area, non-contact measurement technology that can meet the requirements for the detection and evaluation of absorbing coatings. In recent years, domestic and foreign scholars have used microwave waveguide antennas as microwave heating sources for defect detection in composite materials, but microwave antennas have problems such as uneven heating and a small heating area. Summary of the invention

[0006] The purpose of the present invention is to provide a method for training a defect detection model for an absorbing coating, a defect detection method and a system, so as to solve the problems that the existing detection technology cannot detect internal defects of the absorbing coating, defects between the metal substrate and the coating interface, and defects within the metal substrate, as well as low detection efficiency and high detection cost.

[0007] The present invention solves the above technical problems through the following technical solutions: a method for training a radar absorbing coating defect detection model, comprising the following steps:

[0008] Making test pieces with different defects, wherein the locations of the defects include within the absorbing coating, between the substrate and the absorbing coating, and / or within the substrate;

[0009] Placing the defective test piece in the resonant cavity and controlling the microwave heating device to perform microwave heating on the test piece;

[0010] Acquiring a thermal image sequence of the test piece after microwave heating;

[0011] According to the thermal image sequence, a time-temperature variation curve of each pixel point is obtained;

[0012] Performing feature dimensionality reduction processing on the thermal image sequence to extract feature information, wherein the feature information includes defect area features and normal area features, and the defect area features include defect features at different defect locations;

[0013] Defect marking is performed on the time-temperature variation curve of each pixel point according to the characteristic information to obtain a time-temperature variation curve with defect markings, and the time-temperature variation curves of all pixel points and the corresponding defect markings constitute a training sample;

[0014] A defect detection model is constructed, and the time-temperature variation curve of the pixel point is used as input and the corresponding defect mark is used as output. The defect detection model is trained to obtain a trained defect detection model.

[0015] Furthermore, the test piece is subjected to a temperature reduction treatment while acquiring the thermal image sequence.

[0016] Furthermore, the PCA (Principal Component Analysis, PCA) method is used to perform feature dimension reduction processing on the thermal image sequence to extract feature information.

[0017] Furthermore, the defect detection model is a convolutional neural network model, a BP neural network model or a GRU network model.

[0018] Furthermore, the training method further includes: processing the time-temperature variation curve of the current defect through the trained defect detection model to generate and output a defect detection result of the current defect.

[0019] The present invention also provides a method for detecting defects in an absorbing coating, comprising:

[0020] Placing the absorbing coating to be detected in the resonant cavity, and controlling the microwave heating device to perform microwave heating on it;

[0021] Cooling down the microwave-heated microwave-absorbing coating, and acquiring a thermal image sequence of the microwave-absorbing coating during the cooling process;

[0022] According to the thermal image sequence, a time-temperature variation curve of each pixel point is obtained;

[0023] The defect detection model trained as described above is used to process the time-temperature variation curve of the pixel point to generate and output the defect detection result of the absorbing coating.

[0024] The present invention also provides a radar absorbing coating defect detection model training system, comprising:

[0025] A resonant cavity used to place defective specimens during microwave heating;

[0026] A microwave heating device is provided on the resonant cavity;

[0027] A storage platform for placing defective test pieces after microwave heating;

[0028] An infrared thermal imager disposed above the storage platform;

[0029] A cooling device disposed above or inside the storage platform;

[0030] and a control device electrically connected to the microwave heating device, the infrared thermal imager and the cooling device;

[0031] The control device is used to: control the microwave heating device to perform microwave heating on the defective test piece in the resonant cavity; control the cooling device to cool down the defective test piece on the placement platform, and control the infrared thermal imager to collect a thermal image sequence of the defective test piece during the cooling process; obtain the thermal image sequence, and obtain a time-temperature change curve of each pixel point according to the thermal image sequence; perform feature dimensionality reduction processing on the thermal image sequence to extract feature information, wherein the feature information includes defect information and normal information, and the defect information includes defect information of different defect positions; perform defect marking on the time-temperature change curve of each pixel point according to the feature information to obtain a time-temperature change curve with defect marking, and the time-temperature change curves of all pixel points and the corresponding defect markings constitute a training sample; construct a defect detection model, take the time-temperature change curve of the pixel point as input and the corresponding defect marking as output, train the defect detection model, and obtain a trained defect detection model.

[0032] Furthermore, the cooling device is a fan arranged above the storage platform.

[0033] Furthermore, the infrared thermal imager is an online infrared thermal imager of model MAG62.

[0034] The present invention also provides a radar absorbing coating defect detection system, comprising:

[0035] A resonant cavity for placing the absorbing coating to be detected during microwave heating;

[0036] A microwave heating device is provided on the resonant cavity;

[0037] A placement platform for placing the absorbing coating to be tested after microwave heating;

[0038] An infrared thermal imager disposed above the storage platform;

[0039] A cooling device disposed above or inside the storage platform;

[0040] and a control device electrically connected to the microwave heating device, the infrared thermal imager and the cooling device;

[0041] The control device is used to: control the microwave heating device to perform microwave heating on the absorbing coating to be detected in the resonant cavity; control the cooling device to cool down the absorbing coating to be detected on the storage platform, and control the infrared thermal imager to collect a thermal image sequence of the absorbing coating to be detected during the cooling process; obtain the thermal image sequence, and obtain a time-temperature variation curve of each pixel point according to the thermal image sequence; and use the defect detection model trained as described above to process the time-temperature variation curve of the pixel point to generate and output a defect detection result of the absorbing coating.

[0042] Beneficial Effects

[0043] Compared with the prior art, the advantages of the present invention are:

[0044] The present invention provides a microwave absorbing coating defect detection model training method, defect detection method and system, wherein microwave heating is performed on the microwave absorbing coating and a thermal image sequence is obtained. Since the thermal images of the normal area and the defective area are different, feature information of the normal area and the defective area can be extracted, and then the time-temperature curve of each pixel in the thermal image sequence is defect-marked according to the feature information. The defect detection model is trained using training samples consisting of the time-temperature curve and the defect mark, and finally the trained defect detection model is used to realize intelligent detection of the microwave absorbing coating, with a maximum detection depth of 0.3 mm and a minimum diameter-to-depth ratio. The ratio of the microwave field intensity to the microwave field intensity is 1.5, and the pixel-level detection of the absorbing coating is realized; the resonance cavity + microwave heating technology is adopted to ensure that the microwave maintains a uniform field intensity in the resonance cavity, and when heating the test piece or the absorbing coating, a large area of ​​uniform heating is realized, the heating efficiency and heating effect are improved, and a large area of ​​continuous detection capability can be formed, and the detection efficiency is improved, which will provide an effective means for the quality control and in-service detection of the absorbing coating composite material; the method and system of the present invention can realize the detection of the internal defects of the absorbing coating, the defects between the metal substrate and the coating interface, and the defects in the metal substrate, and the whole detection process is contactless, which can realize the in-situ in-service nondestructive detection of the absorbing coating. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the drawings described below are only one embodiment of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0046] Figure 1 is a flow chart of a method for training a radar absorbing coating defect detection model in an embodiment of the present invention;

[0047] Figure 2 Schematic diagram of the detachment defect of the absorbing coating in an embodiment of the present invention;

[0048] Figure 3 Schematic diagram of the separation defect between the absorbing coating and the substrate produced on the test piece in the embodiment of the present invention, wherein (a) is a top view of the substrate, and (b) is a side view of the absorbing coating and the substrate;

[0049] Figure 4 Schematic diagram of internal defects of the absorbing coating produced on the test piece in the embodiment of the present invention, wherein (a) is a top view of the absorbing coating, and (b) is a side view of the absorbing coating and the substrate;

[0050] Figure 5 Schematic diagram of internal defects of a substrate produced on a test piece in an embodiment of the present invention, wherein FIG (a) is a top view of the substrate, and FIG (b) is a side view of the absorbing coating and the substrate;

[0051] Figure 6 It is a schematic diagram of the structure of an absorbing coating defect detection model training system or an absorbing coating defect detection system in an embodiment of the present invention.

[0052] Among them, 1-test piece or absorbing coating to be tested, 11-absorbing coating, 12-substrate, 13-detachment defect, 2-resonant cavity, 3-microwave heating device, 4-infrared thermal imager, 5-control device, 6-fan. DETAILED DESCRIPTION

[0053] The following is a clear and complete description of the technical solutions in the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0054] The technical solution of the present application is described in detail with specific embodiments below. The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described in detail in some embodiments.

[0055] like Figure 1 As shown, a radar absorbing coating defect detection model training method provided in this embodiment includes the following steps:

[0056] Step 1: Prepare test pieces with different defects, where the defects are located in the absorbing coating, between the substrate and the absorbing coating, and / or in the substrate.

[0057] like Figure 2 The adhesion and debonding damage between the metal substrate and the absorbing coating shown in the figure damages the coating performance and even makes it ineffective. The surface damage can be directly observed, but the adhesion and debonding damage between the metal substrate and the absorbing coating, the internal damage of the metal substrate and the internal damage of the absorbing coating cannot be directly observed. The defect detection model trained by the defect detection model training method for absorbing coating provided in this embodiment can realize non-destructive detection of these internal damages.

[0058] Take the bonding and debonding damage between the substrate and the absorbing coating on the specimen as an example. The specimen consists of a CFRP substrate and an absorbing coating. Figure 3 The CFRP substrate is 5 mm thick. Four rows of flat-bottom holes are processed on the CFRP substrate to simulate debonding defects, with 5 flat-bottom holes in each row (bonding defects or debonding defects). Figure 3As shown in (a), from left to right, the diameters of the holes are 10 mm, 8 mm, 5 mm, 3 mm and 1.5 mm respectively; from top to bottom, the first row is pasted with a 2 mm thick absorbing coating, the second row is pasted with a 1 mm thick absorbing coating, the third row is pasted with a 0.5 mm thick absorbing coating, and the fourth row is pasted with a 0.3 mm thick absorbing coating. Figure 3 (b) shows that the middle dotted line represents the debonding defect. When heating, microwaves start heating from the side of the absorbing coating, so the thickness of the absorbing coating can be regarded as the burial depth of the debonding defect, that is, the burial depths of the four rows of defects are 2mm, 1mm, 0.5mm and 0.3mm respectively. Table 1 lists the parameters of all defects of different specimens, including diameter D, burial depth L, and diameter-to-depth ratio τ (calculated by τ = D / L).

[0059] Table 1 Defect parameters or information of specimens with different numbers

[0060] Label Diameter D(mm) Buried depth L(mm) Diameter-to-depth ratio τ #1 10 2.0 5 2 8 2.0 4 3 5 2.0 2.5 4 3 2.0 1.5 5 1.5 2.0 0.75 6 10 1.0 10 7 8 1.0 8 8 5 1.0 5 9 3 1.0 3 10 1.5 1.0 1.5 11 10 0.5 20 12 8 0.5 16 13 5 0.5 10 14 3 0.5 6 15 1.5 0.5 3 16 10 0.3 33.33 17 8 0.3 26.67 18 5 0.3 16.67 19 3 0.3 10 20 1.5 0.3 5

[0061] Take the creation of internal defects in the absorbing coating on the specimen as an example. Figure 4 (a) is a top view of the internal defects of the absorbing coating. The absorbing coating is 2 mm thick. Two rows of flat-bottom holes are processed in the absorbing coating to simulate internal defects. There are 5 defects in each row and the defect depth is 1 mm. From left to right, the diameters of the flat-bottom holes are 10 mm, 8 mm, 5 mm, 3 mm and 1.5 mm, respectively. Figure 4 (b) is a schematic diagram of the absorbing coating being pasted onto the CFRP substrate. The thickness of the CFRP substrate is 5 mm, and the dotted line in the middle represents the internal defects of the absorbing coating.

[0062] Take the creation of internal defects of the substrate on the specimen as an example. Figure 5 (a) is a top view of the internal defects of the CFRP substrate. The CFRP substrate is 5 mm thick. Two rows of flat-bottom holes are machined in the CFRP substrate to simulate internal damage defects. There are 5 defects in each row and the defect depth is 2 mm. From left to right, the diameters of the holes are 10 mm, 8 mm, 5 mm, 3 mm and 1.5 mm, respectively. Figure 5 (b) is a schematic diagram of the absorbing coating being pasted onto the CFRP substrate. The thickness of the absorbing coating is 1 mm, and the dotted line in the middle represents the internal damage of the CFRP.

[0063] Step 2: Place the defective specimen in the resonant cavity and control the microwave heating device to perform microwave heating on it.

[0064] The resonant cavity is a metal cavity used to keep the high-frequency electromagnetic field oscillating continuously. Since the electromagnetic field is completely concentrated in the resonant cavity, there is no radiation loss, so it has a high quality factor. The microwave heating device generates microwaves, and the resonant cavity can ensure that the microwaves maintain a relatively uniform field strength inside it. When heating the specimen, it can achieve uniform heating over a large area, which improves the heating speed and heating effect.

[0065] Microwave heating methods can be divided into volumetric heating and surface heating, so the heating process can be divided into volumetric heating (i.e. dielectric loss) and surface heating (i.e. eddy current heating). For conductive materials, such as carbon fiber composite substrates, microwave heating is eddy current heating. Since the CFRP (carbon fiber reinforced plastic) substrate is conductive, microwaves cannot penetrate the conductive material. Therefore, the principle of microwave heating is to radiate energy to the surface of the conductive material through microwaves to generate an alternating electric field, and then stimulate the induced surface current from the alternating electric field to generate an alternating magnetic field, and then generate a vortex electric field through the alternating magnetic field. The vortex electric field promotes the movement of electrons, thereby generating Joule heat.

[0066] The microwave absorbing coating defect detection model training system based on the resonant cavity and microwave heating includes a resonant cavity 2, a microwave heating device 3, a storage platform, an infrared thermal imager 4, a cooling device and a control device 5. Figure 6 As shown. The microwave heating device 3 is arranged on the resonant cavity 2, a storage platform is arranged on one side of the resonant cavity 2, an infrared thermal imager 4 and a cooling device are arranged above the storage platform, and a control device 5 is electrically connected to the microwave heating device 3, the infrared thermal imager 4 and the cooling device respectively. The control device 5 can remotely control the microwave heating device to heat the specimen in the resonant cavity. The specimen 1 is composed of an absorbing coating 11 and a substrate 12.

[0067] Step 3: Cool down the specimen after microwave heating, and obtain a thermal image sequence of the specimen during the cooling process.

[0068] The microwave-heated specimen is taken out and placed on the storage platform. In order to obtain thermal images at different temperatures and improve the efficiency of thermal image acquisition, a cooling device is used to cool the specimen, and a control device controls the infrared thermal imager to collect a thermal image sequence of the specimen.

[0069] In this embodiment, the cooling device is a fan.

[0070] Step 4: Based on the thermal image sequence, obtain the time-temperature change curve of each pixel.

[0071] The microwave-absorbing coating has a high absorption rate for microwaves, and converts them into heat after absorption. A microwave heating device is used to generate microwaves to excite the coating. The heat generated by the excitation will be conducted in the coating. Defects will directly affect the microwave absorption process, heat generation or conduction process. Therefore, the temperature of the defect area and the normal area is different, that is, the thermal image is different. For example, the temperature of the debonding defect area is lower than the surrounding non-debonding defect area. The defect can be initially located by the thermal image. The temperature value within a period of time is extracted from the thermal image sequence. The temperature value sequence of each pixel point in the thermal image within this period of time (that is, the time-temperature change curve) is taken as a sample. There are as many samples as there are pixels.

[0072] Step 5: Perform feature dimensionality reduction processing on the thermal image sequence to extract feature information, where the feature information includes defect area features and normal area features. The defect area features include defect features at different defect locations, such as within the absorbing coating, between the substrate and the absorbing coating, and / or within the substrate.

[0073] This embodiment uses the PCA method to perform feature dimensionality reduction processing on the thermal image sequence to extract feature information. The PCA method for extracting feature information is a prior art, and the specific extraction steps are not repeated here.

[0074] Step 6: Defects are marked on the time-temperature variation curve of each pixel according to the feature information to obtain a time-temperature variation curve with defect marks. The time-temperature variation curves of all pixels and the corresponding defect marks constitute a training sample.

[0075] Assume that the defect feature in the absorbing coating is marked as 1, the defect feature between the substrate and the absorbing coating is marked as 2, the defect feature in the substrate is marked as 3, and the feature of the normal area is marked as 0. The time-temperature change curve of each pixel corresponds to a different defect feature, so the time-temperature change curve of each pixel corresponds to a different label. For example, the training sample vector corresponding to a certain pixel is recorded as in represents the temperature of pixel j at time t2, F j represents the defect mark of pixel j, F j =0,1,2,3.

[0076] Step 7: Construct a defect detection model, take the time-temperature change curve of the pixel point as input, take the corresponding defect mark as output, train the defect detection model, and obtain the trained defect detection model.

[0077] There are many defect detection models, including but not limited to convolutional neural network models, BP neural network models or GRU network models (gated recurrent unit network models).

[0078] This embodiment uses a one-dimensional convolutional neural network model (One-Dimensional Convolutional Neural Network, 1D-CNN) as a defect detection model. The structure of the one-dimensional convolutional neural network model mainly includes three convolution layers, three pooling layers, a Flatten layer (flattening the data) and two fully connected layers, as shown in Table 2. Since the feature extraction of time series is related to the size of the convolution kernel, if more features need to be detected, the convolution kernel must be set large enough, but it is easy to cause a surge in the amount of calculation in the convolution process and is not conducive to the deepening of the network; setting a smaller convolution kernel, the detected features are more delicate, but the globality of the detected features is not good, and the operation speed is low. Therefore, this embodiment comprehensively considers setting the convolution kernel size of the convolution layer to 9, the step size is 2, the activation function is selected as ReLU, the pooling layer uses the maximum value pooling with a pooling kernel of 4, and the last two fully connected layers are activated using the Sigmoid function.

[0079] Table 2.1D-CNN microwave thermal imaging defect detection model

[0080]

[0081]

[0082] Step 8: Process the time-temperature variation curve of the current defect through the trained defect detection model to generate and output the defect detection result of the current defect.

[0083] The absorbing coating to be detected is placed in the resonant cavity, and the microwave heating device is controlled to perform microwave heating on the absorbing coating. The absorbing coating is taken out after heating, and a thermal image sequence is collected using an infrared thermal imager while cooling the absorbing coating. The time-temperature variation curve of each pixel point is obtained according to the thermal image sequence, and the time-temperature variation curve of each pixel point is input into the trained defect detection model to obtain the defect mark of each pixel point, thereby obtaining the defect detection result of the absorbing coating.

[0084] This embodiment also provides a method for detecting defects in an absorbing coating, comprising the following steps:

[0085] Step 1: Place the absorbing coating to be detected in the resonant cavity, and control the microwave heating device to perform microwave heating on it.

[0086] Step 2: Cooling down the microwave-heated microwave-absorbing coating, and obtaining a thermal image sequence of the microwave-absorbing coating during the cooling process;

[0087] Step 3: According to the thermal image sequence, obtain the time-temperature change curve of each pixel;

[0088] Step 4: Use the defect detection model trained as described above to process the time-temperature variation curve of the pixel point in step 3 to generate and output the defect detection result of the absorbing coating.

[0089] This embodiment also provides a microwave absorbing coating defect detection model training system, including a resonant cavity 2, a microwave heating device 3, a storage platform, an infrared thermal imager 4, a cooling device and a control device 5. Figure 6 As shown. The microwave heating device 3 is arranged on the resonant cavity 2, a storage platform is arranged on one side of the resonant cavity 2, an infrared thermal imager 4 and a cooling device are arranged above the storage platform, and a control device 5 is electrically connected to the microwave heating device 3, the infrared thermal imager 4 and the cooling device respectively. The control device 5 can remotely control the microwave heating device to heat the specimen in the resonant cavity. The specimen 1 is composed of an absorbing coating 11 and a substrate 12.

[0090] The control device 5 is used to control the microwave heating device 3 to perform microwave heating on the defective specimen 1 in the resonant cavity 2; control the cooling device to cool down the defective specimen 1 on the storage platform, and control the infrared thermal imager 4 to collect a thermal image sequence of the defective specimen 1 during the cooling process; obtain the thermal image sequence, and obtain the time-temperature change curve of each pixel point according to the thermal image sequence; perform feature dimensionality reduction processing on the thermal image sequence to extract feature information, the feature information includes defect information and normal information, and the defect information includes defect information at different defect positions; perform defect marking on the time-temperature change curve of each pixel point according to the feature information to obtain a time-temperature change curve with defect marking, and the time-temperature change curves of all pixel points and the corresponding defect markings constitute a training sample; construct a defect detection model, take the time-temperature change curve of the pixel point as input, take the corresponding defect mark as output, train the defect detection model, and obtain a trained defect detection model.

[0091] In this embodiment, the resonant cavity 2 and the microwave heating device 3 use a controllable microwave excitation system of model WBMW-JL01, with an output power of 100-1450W. The system has a resonant cavity and a magnetron, and microwaves are generated by the magnetron. The controllable microwave excitation system excites the coating, and the heat generated by the excitation will be conducted in the coating system. Defects will directly affect the microwave absorption process, the generation or conduction of heat. The infrared thermal imager 4 uses the giant brother SN MAG62, which can realize real-time acquisition of the surface temperature of the test piece, and the field of view is 640×480 pixels. The control device 5 controls the infrared thermal imager 4 to set parameters and observe real-time thermal images, generate clear thermal images of 640×480 pixels, record the transient temperature field of the coating surface over time, and convert it into MATLAB data format for storage.

[0092] This embodiment also provides a microwave absorbing coating defect detection system, including a resonant cavity 2, a microwave heating device 3, a storage platform, an infrared thermal imager 4, a cooling device and a control device 5. Figure 6 As shown. The microwave heating device 3 is arranged on the resonant cavity 2, a storage platform is arranged on one side of the resonant cavity 2, an infrared thermal imager 4 and a cooling device are arranged above the storage platform, and a control device 5 is electrically connected to the microwave heating device 3, the infrared thermal imager 4 and the cooling device respectively. The control device 5 can remotely control the microwave heating device to heat the specimen in the resonant cavity. The specimen 1 is composed of an absorbing coating 11 and a substrate 12.

[0093] The control device 5 is used to control the microwave heating device 3 to perform microwave heating on the absorbing coating 1 to be detected in the resonant cavity 2; control the cooling device to cool down the absorbing coating 1 to be detected on the placement platform, and control the infrared thermal imager 4 to collect a thermal image sequence of the absorbing coating 1 to be detected during the cooling process; obtain the thermal image sequence, and obtain a time-temperature variation curve of each pixel point according to the thermal image sequence; use the defect detection model trained as described above to process the time-temperature variation curve of the pixel point, so as to generate and output the defect detection result of the absorbing coating 1.

[0094] What is disclosed above is only a specific implementation mode of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can easily think of changes or modifications within the technical scope disclosed in the present invention, which should be covered within the protection scope of the present invention.

Claims

1. A method for training a radar absorbing coating defect detection model, characterized in that: The following steps are involved: Making test pieces with different defects, wherein the locations of the defects include within the absorbing coating, between the substrate and the absorbing coating, and / or within the substrate; Placing the defective test piece in the resonant cavity and controlling the microwave heating device to perform microwave heating on the test piece; Acquiring a thermal image sequence of the test piece after microwave heating; According to the thermal image sequence, a time-temperature variation curve of each pixel is obtained; Performing feature dimensionality reduction processing on the thermal image sequence to extract feature information, wherein the feature information includes defect area features and normal area features, and the defect area features include defect features at different defect locations; Defect marking is performed on the time-temperature variation curve of each pixel point according to the characteristic information to obtain a time-temperature variation curve with defect markings, and the time-temperature variation curves of all pixel points and the corresponding defect markings constitute a training sample; A defect detection model is constructed, and the time-temperature variation curve of the pixel point is used as input and the corresponding defect mark is used as output. The defect detection model is trained to obtain a trained defect detection model.

2. The method for training a radar absorbing coating defect detection model according to claim 1, wherein: The test piece is subjected to a temperature reduction treatment while acquiring the thermal image sequence.

3. The method for training a radar absorbing coating defect detection model according to claim 1, wherein: The PCA method is used to perform feature dimension reduction processing on the thermal image sequence to extract feature information.

4. The method for training a radar absorbing coating defect detection model according to claim 1, wherein: The defect detection model is a convolutional neural network model, a BP neural network model or a GRU network model.

5. The method for training a defect detection model for an absorbing coating according to any one of claims 1 to 4, characterized in that: Also includes: The time-temperature variation curve of the current defect is processed by the trained defect detection model to generate and output a defect detection result of the current defect.

6. A method for detecting defects in an absorbing coating, characterized in that: include: Placing the absorbing coating to be detected in the resonant cavity, and controlling the microwave heating device to perform microwave heating on it; Cooling down the microwave-heated microwave-absorbing coating, and acquiring a thermal image sequence of the microwave-absorbing coating during the cooling process; According to the thermal image sequence, a time-temperature variation curve of each pixel is obtained; The defect detection model trained as claimed in any one of claims 1 to 5 is used to process the time-temperature variation curve of the pixel point to generate and output the defect detection result of the absorbing coating.

7. A radar absorbing coating defect detection model training system, characterized in that: include: A resonant cavity used to place defective specimens during microwave heating; A microwave heating device is provided on the resonant cavity; A storage platform for placing defective test pieces after microwave heating; An infrared thermal imager disposed above the storage platform; A cooling device disposed above or inside the storage platform; and a control device electrically connected to the microwave heating device, the infrared thermal imager and the cooling device; The control device is used to: control the microwave heating device to perform microwave heating on the defective test piece in the resonant cavity; control the cooling device to cool down the defective test piece on the storage platform, and control the infrared thermal imager to collect a thermal image sequence of the defective test piece during the cooling process; obtain the thermal image sequence, and obtain a time-temperature change curve of each pixel point according to the thermal image sequence; perform feature dimensionality reduction processing on the thermal image sequence to extract feature information, wherein the feature information includes defect information and normal information, and the defect information includes defect information at different defect locations; perform defect marking on the time-temperature change curve of each pixel point according to the feature information to obtain a time-temperature change curve with defect marking, and the time-temperature change curves of all pixel points and the corresponding defect markings constitute a training sample; A defect detection model is constructed, and the time-temperature variation curve of the pixel point is used as input and the corresponding defect mark is used as output. The defect detection model is trained to obtain a trained defect detection model.

8. The radar absorbing coating defect detection model training system according to claim 7, characterized in that: The cooling device is a fan arranged above the storage platform.

9. The radar absorbing coating defect detection model training system according to claim 7 or 8, characterized in that: The infrared thermal imager is an online infrared thermal imager of model MAG62.

10. A radar absorbing coating defect detection system, characterized in that: include: A resonant cavity for placing the absorbing coating to be detected during microwave heating; A microwave heating device is provided on the resonant cavity; A placement platform for placing the absorbing coating to be tested after microwave heating; An infrared thermal imager disposed above the storage platform; A cooling device disposed above or inside the storage platform; and a control device electrically connected to the microwave heating device, the infrared thermal imager and the cooling device; The control device is used to: control the microwave heating device to perform microwave heating on the absorbing coating to be detected in the resonant cavity; control the cooling device to cool down the absorbing coating to be detected on the storage platform, and control the infrared thermal imager to collect a thermal image sequence of the absorbing coating to be detected during the cooling process; obtain the thermal image sequence, and obtain a time-temperature variation curve of each pixel point according to the thermal image sequence; The defect detection model trained as claimed in any one of claims 1 to 5 is used to process the time-temperature variation curve of the pixel point to generate and output the defect detection result of the absorbing coating.

Images