A method and system for detecting surface defects of an antenna phase shifter

Abstract

The application relates to the technical field of infrared defect detection, and discloses an antenna phase shifter surface defect detection method and system, which comprises the following steps: collecting infrared thermal image data of an antenna phase shifter surface through an infrared thermal imager, analyzing thermal resistance change conditions of the phase shifter surface according to the infrared thermal image data, and constructing a thermal resistance gradient graph; extracting a region with thermal resistance gradient greater than a preset gradient threshold value from the thermal resistance gradient graph to obtain a first defect region; in the first defect region, identifying a defect boundary in combination with infrared thermal image data and pre-acquired visible light data to obtain a second defect region; and calculating a defect score in the second defect region through a preset defect early warning model, and early warning a region with a defect score greater than a preset defect threshold value; the application can accurately identify the surface defects of the phase shifter, accurately define the defect boundary of the surface defects, determine the corresponding defect degree, timely issue early warnings, and guarantee the stable operation of the antenna phase shifter.

Description

Technical Field

[0001] This invention relates to the field of infrared defect detection technology, and more specifically to a method and system for detecting surface defects in an antenna phase shifter. Background Technology

[0002] Antennas play a vital role in fields such as communication, radar, and navigation. During operation, phase shifters may develop surface cracks, dents, corrosion, or coating peeling, which can lead to increased signal transmission loss, reduced anti-interference capability, decreased mechanical strength, and even equipment failure. Therefore, in order to avoid economic losses and safety hazards caused by surface defects of phase shifters, it is necessary to identify and warn of surface defects of phase shifters in advance.

[0003] The existing technology has the following problems: manual inspection of surface defects of phase shifters is inefficient and prone to missed detection; ultrasonic testing of surface defects of phase shifters has low sensitivity to small surface defects, resulting in low accuracy of defect identification; the identified defect range is large, making it impossible to accurately define the boundary range of the defects, and the accuracy of surface defect identification of phase shifters cannot meet the actual defect detection requirements; in order to solve at least one of the above problems, the present invention proposes a method and system for detecting surface defects of antenna phase shifters. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a method and system for detecting surface defects in antenna phase shifters, effectively solving the problems described in the background art. The specific technical solution of the present invention is as follows:

[0005] A method for detecting surface defects in an antenna phase shifter includes:

[0006] Infrared thermal image data of the antenna phase shifter surface is acquired by an infrared thermal imager, and the thermal resistance change of the phase shifter surface is analyzed based on the infrared thermal image data to construct a thermal resistance gradient map.

[0007] Based on the thermal resistance gradient map, the region where the thermal resistance gradient is greater than the preset gradient threshold is extracted to obtain the first defect region;

[0008] Within the first defect area, the defect boundary is identified by combining infrared thermal imaging data and pre-acquired visible light data to obtain the second defect area;

[0009] The defect score in the second defect area is calculated by a preset defect warning model, and a warning is issued for areas where the defect score is greater than a preset defect threshold, so as to detect defects on the surface of the antenna phase shifter.

[0010] Specifically, the step of acquiring infrared thermal image data of the antenna phase shifter surface using an infrared thermal imager, analyzing the thermal resistance change of the phase shifter surface based on the infrared thermal image data, and constructing a thermal resistance gradient map includes:

[0011] Infrared thermal image data is collected by applying pulsed thermal excitation to the surface of the antenna phase shifter using an infrared thermal imager.

[0012] Based on the infrared thermal image data, the thermal resistance value and thermal resistance gradient of the phase shifter surface are calculated using a preset thermal resistance analysis model, and a thermal resistance gradient map is constructed.

[0013] Specifically, based on the infrared thermal image data, the thermal resistance value and thermal resistance gradient of the phase shifter surface are calculated using a preset thermal resistance analysis model, and a thermal resistance gradient map is constructed, including:

[0014] Based on infrared thermal image data, the thermal resistance value of each pixel position of the phase shifter is calculated using a preset thermal resistance analysis model.

[0015] The thermal resistance gradient is obtained by calculating the difference in thermal resistance values ​​between adjacent pixels.

[0016] By combining the thermal resistance value and the thermal resistance gradient, a thermal resistance gradient map is constructed.

[0017] Specifically, based on the thermal resistance gradient map, regions with thermal resistance gradients greater than a preset gradient threshold are extracted to obtain the first defect region, including:

[0018] Based on the thermal resistance gradient map, regions with thermal resistance gradients greater than a preset gradient threshold are extracted to obtain candidate defect regions;

[0019] By analyzing the correlation of defects within the candidate defect regions, the candidate defect regions are merged to obtain multiple merged defect regions.

[0020] The first defect region is obtained by combining the merged defect regions.

[0021] Specifically, by analyzing the correlation of defects within the candidate defect regions, the candidate defect regions are merged to obtain multiple merged defect regions, including:

[0022] Identify the defect endpoints within the candidate defect region and construct a set of defect endpoints;

[0023] According to the direction of the increase in thermal resistance gradient in the defect region, connect the defect endpoints in the defect endpoint set where the distance between two points is less than a preset distance threshold to obtain the defect boundary set;

[0024] Calculate the boundary connectivity value of every two defect boundaries in the defect boundary set, and connect the defect boundaries whose boundary connectivity value is greater than a preset connectivity threshold to obtain a set of connected domains;

[0025] Connected regions in the connected region set whose regional connectivity values ​​are greater than a preset connectivity threshold are merged to obtain multiple merged defect regions.

[0026] Specifically, within the first defect area, the defect boundary is identified by combining infrared thermal imaging data and pre-acquired visible light data to obtain the second defect area, including:

[0027] For the first defect region, feature extraction is performed on both infrared thermal imaging data and pre-acquired visible light data to construct a defect boundary response map;

[0028] Based on the defect boundary response map, the boundary of the first defect region is shrunk or expanded to obtain the second defect region.

[0029] Specifically, for the first defect region, feature extraction is performed on both the infrared thermal image data and the pre-acquired visible light data to construct a defect boundary response map, including:

[0030] For the first defect region, the infrared thermal image data and the pre-acquired visible light data are decomposed according to a preset scale to obtain a multi-scale infrared image sequence and a multi-scale visible light image sequence.

[0031] Using a preset feature extraction model, features are extracted from the multi-scale infrared image sequence and the multi-scale visible light image sequence to obtain a multi-scale infrared feature set and a multi-scale visible light feature set.

[0032] The multi-scale infrared feature set and the multi-scale visible light feature set are fused according to the scale level to generate a feature fusion response map for each scale.

[0033] The feature response maps at each scale are fused together to construct a defect boundary response map.

[0034] Specifically, based on the defect boundary response map, the boundary of the first defect region is shrunk or expanded to obtain the second defect region, including:

[0035] Based on the defect boundary response diagram, the curvature and constituent area of ​​the defect boundary are analyzed to construct boundary constraints and spatial constraints;

[0036] The boundary of the first defect region is optimized by the aforementioned boundary constraints to obtain the first optimized boundary;

[0037] The first optimization boundary is shrunk or expanded by the spatial constraints to obtain the second optimization boundary;

[0038] The region formed by the second optimization boundary is taken as the second defect region.

[0039] Specifically, the step of calculating a defect score within the second defect area using a preset defect warning model, and issuing a warning for areas where the defect score exceeds a preset defect threshold, in order to detect surface defects on the antenna phase shifter, includes:

[0040] By extracting the defect features from the second defect region, a defect feature vector is constructed;

[0041] Based on the defect feature vector, the defect type corresponding to the second defect area is identified and the defect score is calculated using a preset defect early warning model;

[0042] When the defect score is greater than the preset defect threshold, an early warning is issued in conjunction with the defect type to detect surface defects of the antenna phase shifter.

[0043] An antenna phase shifter surface defect detection system, used to implement the aforementioned antenna phase shifter surface defect detection method, includes:

[0044] The thermal resistance gradient map construction module collects infrared thermal image data of the antenna phase shifter surface using an infrared thermal imager, analyzes the thermal resistance change of the phase shifter surface based on the infrared thermal image data, and constructs a thermal resistance gradient map.

[0045] The first defect region identification module extracts regions with thermal resistance gradients greater than preset gradient thresholds based on the thermal resistance gradient map, thus obtaining the first defect region.

[0046] The second defect region identification module identifies the defect boundary within the first defect region by combining infrared thermal imaging data and pre-acquired visible light data, thereby obtaining the second defect region.

[0047] The defect warning module calculates the defect score in the second defect area using a preset defect warning model, and issues a warning for areas where the defect score is greater than a preset defect threshold, so as to detect defects on the surface of the antenna phase shifter.

[0048] The beneficial effects of this invention are as follows: A thermal resistance gradient map is constructed based on infrared thermal imaging data. The first defect region with abnormal thermal resistance gradient change is initially identified based on the thermal resistance gradient map. The defect boundary of the first defect region is identified and optimized to obtain a second defect region with a more accurate defect boundary. Feature extraction and defect identification and early warning are performed by combining infrared data and visible light data. This enables accurate identification of surface defects of the antenna phase shifter, accurate definition of the defect boundary of the surface defect, determination of the corresponding defect location, size and severity, timely issuance of early warning, and ensuring the stable operation of the phase shifter. Attached Figure Description

[0049] Figure 1This is a flowchart illustrating the process of a method for detecting surface defects in an antenna phase shifter according to an embodiment of the present invention.

[0050] Figure 2 This is a schematic diagram of the thermal resistance gradient diagram in an embodiment of the present invention;

[0051] Figure 3 This is a schematic diagram of the structure for generating the second optimized boundary in an embodiment of the present invention;

[0052] Figure 4 This is a schematic diagram of the structure of an antenna phase shifter surface defect detection system according to an embodiment of the present invention. Detailed Implementation

[0053] The present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Identical components are denoted by the same reference numerals. It should be noted that the terms "front," "rear," "left," "right," "upper," and "lower" used in the following description refer to directions in the accompanying drawings, and the terms "bottom surface," "top surface," "inner," and "outer" refer to directions toward or away from the geometric center of a specific component, respectively.

[0054] In the embodiments of this application, the terms "exemplary" or "for example" are used to indicate that something is an example, illustration, or description. Any embodiment or design that is described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design. Specifically, the use of the terms "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0055] Hereinafter, the terms "first," "second," and other generic terms are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature. In the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0056] refer to Figure 1 The image shows a specific embodiment of the method for detecting surface defects in an antenna phase shifter according to the present invention, comprising:

[0057] S101. Collect infrared thermal image data of the antenna phase shifter surface using an infrared thermal imager, analyze the thermal resistance change of the phase shifter surface based on the infrared thermal image data, and construct a thermal resistance gradient map.

[0058] S102. Based on the thermal resistance gradient map, extract the region where the thermal resistance gradient is greater than the preset gradient threshold to obtain the first defect region.

[0059] S103. Within the first defect area, the defect boundary is identified by combining infrared thermal image data and pre-acquired visible light data to obtain the second defect area;

[0060] S104. Calculate the defect score in the second defect area using the preset defect warning model, and issue a warning for areas where the defect score is greater than the preset defect threshold, so as to detect defects on the surface of the antenna phase shifter.

[0061] After prolonged operation, antenna phase shifters may develop cracks, corrosion, and coating peeling, leading to increased signal transmission loss, reduced anti-interference capabilities, and even equipment malfunction. Therefore, to prevent surface defects from affecting signal transmission, it is necessary to detect these defects and repair them when they do affect signal transmission. Current methods for detecting surface defects in antenna phase shifters primarily involve visual inspection and ultrasonic testing, which are inefficient and cannot identify minute defects. This invention uses infrared imaging to analyze the resistance distribution on the antenna phase shifter surface. This allows for the identification of resistance changes caused by surface defects, accurately identifying surface defects, analyzing their type and severity, issuing timely warnings, and repairing the defects to ensure stable antenna performance.

[0062] When the antenna phase shifter operates normally without surface defects, the internal circuitry and structure of the antenna are normal, and the heat distribution on the surface is uniform. However, when defects exist on the antenna phase shifter surface, including cracks, solder detachment, or coating peeling, it affects current conduction and heat dissipation, causing a corresponding change in the thermal resistance of the defective area. This embodiment captures infrared radiation from the antenna phase shifter surface using an infrared thermal imager and converts it into temperature data. The temperature data is then used to analyze the changes in thermal resistance on the antenna phase shifter surface, and a thermal resistance gradient map is constructed. By collecting infrared thermal image data and converting it into thermal resistance data, the temperature information of the antenna phase shifter surface can be transformed into thermal resistance information reflecting antenna performance. The constructed thermal resistance gradient map reflects the changes in thermal resistance at different locations on the antenna phase shifter surface, thereby quickly locating areas of abnormal thermal resistance changes and providing data for defect detection on the antenna phase shifter surface.

[0063] Specifically, from the constructed thermal resistance gradient map, regions with thermal resistance gradients greater than a preset gradient threshold are extracted to obtain the first defect region. On the surface of a normal antenna phase shifter, the thermal resistance gradient is uniformly distributed and at a low level. When a defect occurs on the surface, heat transfer in the defect region is hindered, causing the thermal resistance gradient to increase and exceed that of the normal region. At this time, based on the abnormal change in the thermal resistance gradient of the defect region, a gradient threshold is set, and regions with thermal resistance gradients greater than the gradient threshold are extracted as the first defect region. By extracting the defect region through the preset gradient threshold, regions with defects can be quickly and accurately screened from the thermal resistance gradient map, eliminating a large amount of interference from normal regions, narrowing the scope of defect detection, and improving the efficiency of defect detection.

[0064] After identifying the first defect area, a high-definition digital camera is used to acquire images of the antenna phase shifter surface within the first defect area, obtaining visible light image data of the first defect area. Image acquisition is only performed on the first defect area to avoid resource waste and low data acquisition efficiency caused by acquiring images of the entire antenna phase shifter surface. Within the first defect area, infrared thermal imaging data and visible light data are combined to identify the defect boundary of the antenna phase shifter surface defect, determining a second defect area with a more accurate boundary range. Because changes in thermal resistance gradient are affected by the surrounding environment and structure, the boundary may be blurred, making it impossible to accurately determine the actual defect boundary. The acquired visible light data can clearly display the structural features within the antenna surface defect area, showing a clear defect boundary. At this point, combining infrared thermal imaging data and visible light data can more accurately identify the defect boundary, improving the accuracy of defect area positioning and obtaining a more accurate second defect area.

[0065] Specifically, after accurately locating the second defect area, a pre-set defect warning model is used to calculate the defect score within the second defect area. Warnings are issued for areas where the defect score exceeds a pre-set defect threshold. The pre-set defect warning model includes, but is not limited to, a random forest model. This model is trained using a large amount of historical defect data to obtain a pre-trained defect warning model. The defect data of the second defect area is input into the pre-trained model, which analyzes the defect characteristics of the second defect area and calculates the corresponding defect score. A defect threshold is set based on the defect severity analysis of the defect area. When the calculated defect score exceeds the threshold, it indicates that the defect will affect the performance of the antenna phase shifter, and a warning is issued based on the defect type and severity. By using the pre-set defect warning model to score and issue warnings for defects, the timeliness and accuracy of defect analysis and warnings can be improved, allowing for the timely detection of serious defects and reminding staff to take appropriate repair or replacement measures to ensure the normal operation of the antenna phase shifter.

[0066] This invention constructs a thermal resistance gradient map based on infrared thermal imaging data. Based on the thermal resistance gradient map, a first defect region with abnormal thermal resistance gradient changes is initially identified. The defect boundary of the first defect region is identified and optimized to obtain a second defect region with a more accurate defect boundary. By combining infrared and visible light data for feature extraction and defect identification and early warning, the invention achieves accurate identification of surface defects of the antenna phase shifter, accurately defines the defect boundary of the surface defect, thereby determining the corresponding defect location, size and severity, and issuing timely early warnings to ensure the stable operation of the antenna phase shifter.

[0067] Furthermore, infrared thermal image data of the antenna phase shifter surface is acquired using an infrared thermal imager. Based on this data, the thermal resistance variation on the phase shifter surface is analyzed, and a thermal resistance gradient map is constructed, including:

[0068] S201. Infrared thermal image data is collected by applying pulsed thermal excitation to the surface of the antenna phase shifter using an infrared thermal imager.

[0069] S202. Based on the infrared thermal image data, calculate the thermal resistance value and thermal resistance gradient of the phase shifter surface using a preset thermal resistance analysis model, and construct a thermal resistance gradient map.

[0070] In this embodiment, an infrared thermal imager applies pulsed thermal excitation to the surface of the antenna phase shifter. This pulsed thermal excitation causes temperature changes on the antenna phase shifter surface. Infrared thermal image data of the antenna phase shifter surface during the thermal excitation process is continuously collected. Using a preset thermal resistance analysis model, the collected infrared thermal image data is converted into thermal resistance values ​​and thermal resistance gradients, constructing a thermal resistance gradient map. Constructing a thermal resistance gradient map by collecting infrared thermal image data makes the thermal resistance changes corresponding to previously difficult-to-identify surface defects more apparent, improving the accuracy of defect identification.

[0071] Specifically, infrared thermal imager is used to apply pulsed thermal excitation to the surface of the antenna phase shifter and collect infrared thermal image data. When pulsed thermal excitation is applied to the surface of the antenna phase shifter, heat is conducted and diffused inside the antenna phase shifter. The heat conduction in the normal area is uniform and the speed is stable, while in the defect area, the heat conduction will be abnormal due to the difference in thermal resistance. The temperature change process of the antenna phase shifter surface under the action of pulsed thermal excitation is collected in real time by applying pulsed thermal excitation. By applying pulsed thermal excitation, the temperature difference on the surface of the antenna phase shifter can be enhanced, making the thermal resistance change caused by small defects more prominent, thereby improving the accuracy of infrared thermal image data for defect identification.

[0072] After acquiring infrared thermographic data of the antenna phase shifter surface, the thermal resistance value and thermal resistance gradient of the antenna phase shifter surface are calculated using a preset thermal resistance analysis model, and a thermal resistance gradient map is constructed. Based on the heat conduction equation and a large amount of infrared thermographic data, a thermal resistance analysis model is established. The model calculates the thermal resistance value based on infrared thermographic data at different times. The thermal resistance value represents the material's ability to impede heat transfer. The thermal resistance value of defective areas differs from that of normal areas. By calculating the thermal resistance gradient, the degree and direction of thermal resistance changes in different areas can be reflected. A thermal resistance gradient map is constructed based on the calculated thermal resistance gradient, which can transform infrared thermographic data into thermal resistance data that reflects the thermal performance of the antenna phase shifter surface material. The constructed thermal resistance gradient map can display the changes in thermal resistance on the antenna phase shifter surface, thereby quickly identifying defective areas on the antenna phase shifter surface.

[0073] Furthermore, based on infrared thermal imaging data, the thermal resistance value and thermal resistance gradient of the phase shifter surface are calculated using a pre-defined thermal resistance analysis model, and a thermal resistance gradient map is constructed, including:

[0074] S301. Based on the infrared thermal image data, calculate the thermal resistance value of each pixel position of the phase shifter using a preset thermal resistance analysis model.

[0075] S302. The thermal resistance gradient is obtained by calculating the difference in thermal resistance values ​​between adjacent pixels.

[0076] S303. Combining thermal resistance value and thermal resistance gradient, construct a thermal resistance gradient diagram.

[0077] In this embodiment, the infrared thermal image data includes temperature information of each pixel on the surface of the antenna phase shifter. Based on the temperature information, the thermal resistance distribution on the surface of the antenna phase shifter can be calculated. Thermal resistance represents the physical quantity that an object's surface can resist heat transfer. The thermal resistance value at each pixel location of the phase shifter is calculated using a preset thermal resistance analysis model. The thermal resistance analysis model includes, but is not limited to, a multivariate fitting model. A large amount of phase shifter temperature and thermal resistance data is used to fit the multivariate fitting model to obtain the fitted thermal resistance analysis model. The temperature data of each pixel on the surface of the antenna phase shifter is input into the fitted thermal resistance analysis model, and the model outputs the thermal resistance value at each pixel location of the phase shifter. By converting the infrared thermal image data into thermal resistance values ​​through the preset thermal resistance analysis model, the thermal performance of the phase shifter surface can be quantitatively analyzed, thereby quickly identifying defective areas with abnormal thermal resistance changes.

[0078] Specifically, based on the calculated thermal resistance value of each pixel, the difference in thermal resistance values ​​between every two adjacent pixels is calculated to obtain the thermal resistance gradient. The thermal resistance gradient represents the spatial variation of the thermal resistance value, including the degree and direction of the change. On the surface of the phase shifter, the thermal resistance value changes relatively gently in normal areas, resulting in a small thermal resistance gradient. In defective areas, the thermal resistance changes abruptly, resulting in a larger thermal resistance gradient. By calculating the thermal resistance gradient, defective areas with drastic changes in thermal resistance can be quickly identified.

[0079] like Figure 2 As shown, the calculated thermal resistance value and thermal resistance gradient are marked on the surface of the phase shifter, and arrows are used to indicate the direction of increasing thermal resistance value to construct a thermal resistance gradient map. The specific thermal resistance value reflects the magnitude of thermal resistance at each location on the phase shifter surface, and the thermal resistance gradient reflects the change in thermal resistance. Combining the thermal resistance value and thermal resistance gradient can comprehensively describe the thermal resistance distribution characteristics on the phase shifter surface. The constructed thermal resistance gradient map facilitates the observation of the thermal resistance distribution on the phase shifter surface and the areas of drastic thermal resistance changes, thereby quickly identifying the location of defects.

[0080] Furthermore, based on the thermal resistance gradient map, regions with thermal resistance gradients greater than a preset gradient threshold are extracted to obtain the first defect region, including:

[0081] S401. Based on the thermal resistance gradient map, extract the regions where the thermal resistance gradient is greater than the preset gradient threshold to obtain the candidate defect regions.

[0082] S402. By analyzing the correlation of defects within the candidate defect regions, the candidate defect regions are merged to obtain multiple merged defect regions.

[0083] S403. Combine and merge the defective regions to obtain the first defective region.

[0084] In this embodiment, the thermal resistance gradient of the defective region is significantly greater than that of the normal region. Based on a preset gradient threshold, regions with thermal resistance gradients greater than the preset threshold are quickly extracted from the thermal resistance gradient map to obtain candidate defective regions with abnormal thermal resistance gradients. After extracting the candidate defective regions, there will be defect correlation between candidate defective regions that are close in location. The mutually correlated candidate defective regions are merged to obtain multiple merged defective regions. The first defective region is obtained by combining the merged defective regions. By using threshold screening, candidate defective regions can be extracted quickly, avoiding defect analysis on the entire phase shifter surface and improving defect detection efficiency. By merging correlated regions, the accuracy of defect identification and detection can be improved, reducing missed detections and false judgments.

[0085] In the thermal resistance gradient map, the magnitude of the thermal resistance gradient reflects the degree of change in thermal resistance. In areas with defects, the thermal resistance changes abruptly due to impeded heat transfer, resulting in a significant increase in the thermal resistance gradient. Based on the average thermal resistance and standard deviation in the thermal resistance gradient map, a gradient threshold is set to the average thermal resistance plus three times the standard deviation. Areas with thermal resistance gradients greater than the gradient threshold are extracted, yielding multiple candidate defect areas. By using the gradient threshold for screening, areas with defects can be quickly identified from the thermal resistance gradient map, narrowing the detection range and improving the efficiency of defect area identification and processing.

[0086] Specifically, after selecting multiple candidate defect regions, these regions may be segmented into multiple independent small regions due to image noise and defect discontinuities. However, these small regions actually belong to the same defect. By analyzing the positional relationship and thermal resistance gradient characteristics between candidate defect regions, candidate defect regions that are too close to each other and have consistent thermal resistance gradient trends are identified as highly correlated candidate defect regions. These highly correlated candidate defect regions are then merged to obtain multiple merged defect regions. By merging correlated candidate defect regions, the region segmentation problem caused by noise or defect discontinuities can be eliminated, avoiding misjudging the same defect as multiple independent defects and improving the completeness of defect region identification.

[0087] Specifically, by combining multiple merged defect areas and further filtering the defect areas, and by statistically analyzing the historical defect data of the phase shifter, a minimum defect area threshold is set. Based on the set defect area threshold, defect areas smaller than the threshold are excluded, as are small areas that are clearly not defects. The resulting area is the first defect area. By filtering out areas that do not constitute defects through area filtering, the accuracy of the first defect area is improved, thereby enhancing the overall accuracy and reliability of defect detection.

[0088] Furthermore, by analyzing the correlation between defects within the candidate defect regions, the candidate defect regions are merged to obtain multiple merged defect regions, including:

[0089] S501. Identify the defect endpoints within the candidate defect region and construct a set of defect endpoints;

[0090] S502. Connect the defect endpoints in the defect endpoint set whose distance is less than a preset distance threshold according to the direction of the increase of thermal resistance gradient in the defect region to obtain the defect boundary set.

[0091] S503. Calculate the boundary connectivity value of every two defect boundaries in the defect boundary set, and connect the defect boundaries with boundary connectivity values ​​greater than the preset connectivity threshold to obtain a set of connected domains.

[0092] S504. Merge the connected regions in the connected region set whose regional connectivity values ​​are greater than a preset connectivity threshold to obtain multiple merged defect regions.

[0093] In this embodiment, for the selected candidate defect regions, there are multiple defect endpoints on the boundary of the candidate defect regions. The defect endpoints may be the start position, end position, or intersection point of defect branches. The defect endpoints of each candidate defect region are identified to form a set of defect endpoints. By identifying the defect endpoints, the connection position of the defect region can be located, providing a reference point for judging the correlation of regions and improving the accuracy of the region correlation judgment.

[0094] Specifically, based on the direction of the thermal resistance gradient in the thermal resistance gradient map, and following the direction of the increase in thermal resistance gradient in the defect region, defect endpoints in the defect endpoint set whose distance is less than a preset distance threshold are connected to obtain a defect boundary set. Within the defect region, the thermal resistance of the defect core area is greater, and the change in thermal resistance gradient is more drastic. The direction of the increase in thermal resistance gradient points towards the core area of ​​the defect. Different endpoints of the same defect are correlated in the direction of the increase in thermal resistance gradient, and the distance between the endpoints is relatively short. Based on the size of common defects in phase shifters, the average distance between adjacent endpoints on the defects is calculated. A preset distance threshold is set to 1.5 times this average distance. Defect endpoints in the defect endpoint set whose distance is less than the preset distance threshold are connected. This process is repeated for all defect endpoints in the defect endpoint set, resulting in multiple defect boundaries. Combining the direction of the increase in thermal resistance gradient and the distance between defect endpoints, a defect boundary set is constructed, providing a boundary basis for determining regional connectivity.

[0095] Specifically, after constructing the defect boundary set, the boundary connectivity value is calculated by combining every two defect boundaries in the set. There is a certain correlation between defect boundaries. For defect boundaries in the defect boundary set, the boundary connectivity value is obtained by weighting the distance and thermal resistance gradient similarity values ​​of two defect boundaries. Based on a large amount of historical defect data, the connectivity threshold of defect connectivity is calculated. Boundaries with a connectivity value greater than the preset connectivity threshold are connected to form a set of connected domains. By calculating the boundary connectivity value and connecting the corresponding boundaries, it is possible to accurately determine which defect boundaries belong to the same defect. Further integration of different parts of the same defect yields a complete connected domain, improving the coherence and integrity of the defect region.

[0096] Specifically, within the constructed connected domains, there are also correlations between different connected domains. By calculating the regional connectivity values ​​between different connected domains, connected domains with regional connectivity values ​​greater than a preset connectivity threshold are merged to obtain multiple merged defect regions. By weighting the overlapping area and thermal resistance gradient similarity value of two connected domains to obtain regional connectivity values, regions with regional connectivity values ​​greater than a preset connectivity threshold are merged to form multiple merged defect regions. By further merging different parts belonging to the same defect, the final merged defect region is closer to the complete shape of the defect, reducing the occurrence of defect segmentation and improving the completeness and accuracy of defect region identification.

[0097] Furthermore, within the first defect region, the defect boundary is identified by combining infrared thermal imaging data and pre-acquired visible light data to obtain the second defect region, which includes:

[0098] S601. For the first defect region, feature extraction is performed on the infrared thermal image data and the pre-acquired visible light data respectively to construct a defect boundary response map;

[0099] S602. Based on the defect boundary response map, the boundary of the first defect region is shrunk or expanded to obtain the second defect region.

[0100] In this embodiment, for the first defect region, temperature gradient features are extracted from infrared thermal imaging data and edge texture features are extracted from visible light data. By combining the temperature gradient features and edge texture features, a defect boundary response map is constructed, which can identify the precise defect boundary. Based on the distribution of high response value regions in the defect boundary response map, the boundary of the first defect region is shrunk or expanded to obtain the second defect region. By adjusting the boundary of the defect region through the construction of the defect boundary response map, a more accurate defect region can be obtained.

[0101] Specifically, infrared thermographic data can reflect the temperature distribution characteristics of the first defect area, with obvious temperature abrupt changes at the defect boundary; visible light data can reflect the physical structural characteristics of the defect area, with significant changes in color and texture at the defect boundary. By extracting features related to the defect boundary from both infrared thermographic data and visible light data, and combining these features, a defect boundary response map can be constructed to highlight the location of the defect boundary. By combining the temperature characteristics of infrared thermographic data and the structural characteristics of visible light data, the shortcomings of single data can be compensated for, making the defect boundary characteristics more prominent, and the constructed response map can more accurately indicate the location of the defect boundary.

[0102] Specifically, the boundary of the first defect region selected may have some deviation. The high response value region in the constructed defect boundary response map corresponds to a more accurate defect boundary. Based on the defect boundary response map, the boundary of the first defect region is shrunk or expanded to correct and optimize the defect boundary, thus obtaining the second defect region. Adjusting the defect boundary based on the defect boundary response map can combine the dual characteristics of infrared and visible light to correct the defect boundary deviation of the first defect region, making the second defect region more consistent with the boundary of the real defect and improving the accuracy of defect region positioning.

[0103] Furthermore, for the first defect region, feature extraction is performed on both the infrared thermal image data and the pre-acquired visible light data to construct a defect boundary response map, including:

[0104] S701. For the first defect area, the infrared thermal image data and the pre-acquired visible light data are decomposed according to the preset scale to obtain a multi-scale infrared image sequence and a multi-scale visible light image sequence.

[0105] S702. Using a preset feature extraction model, feature extraction is performed on the multi-scale infrared image sequence and the multi-scale visible light image sequence to obtain the multi-scale infrared feature set and the multi-scale visible light feature set.

[0106] S703. Perform feature fusion of the multi-scale infrared feature set and the multi-scale visible light feature set according to the scale level to generate a feature fusion response map for each scale.

[0107] S704. Fuse the feature response maps of each scale to construct the defect boundary response map.

[0108] In this embodiment, for the first defect region, since the defect boundary exhibits different characteristics at different scales, a small scale can reflect the details of the boundary, including the edges of tiny cracks, while a large scale can reflect the overall outline of the boundary, including the extent of large-area defects. By decomposing the infrared thermal image data and visible light data according to preset scales, a multi-scale image sequence including information at different scales can be obtained. Based on the size of the first defect region, corresponding scales are set: small scale, medium scale, and large scale. The infrared thermal image data and visible light data are then decomposed according to the corresponding scales to obtain multi-scale infrared image sequences and multi-scale visible light image sequences. Multi-scale decomposition preserves the detailed information and overall outline information of the defect boundary, avoiding the inaccuracy of boundary results caused by analyzing the defect boundary at a single scale, and improving the accuracy of defect boundary analysis.

[0109] Specifically, after decomposing the images into multi-scale infrared image sequences and multi-scale visible light image sequences, feature extraction is performed using a pre-defined feature extraction model, which includes, but is not limited to, the VGG model. When extracting features from the multi-scale infrared image sequences, the VGG model is trained using a large number of multi-scale infrared image sequences to obtain a pre-trained first VGG model. The multi-scale infrared image sequences are then input into the pre-trained first VGG model, and the model outputs a multi-scale infrared feature set. Similarly, when extracting features from the multi-scale visible light image sequences, the VGG model is trained using a large number of multi-scale visible light image sequences to obtain a pre-trained second VGG model. The multi-scale visible light image sequences are then input into the pre-trained second VGG model, and the model outputs a multi-scale visible light feature set. By extracting features from multi-scale images, feature information at different scales can be obtained, providing rich and comprehensive feature information for defect analysis.

[0110] Specifically, after extracting the multi-scale infrared feature set and the multi-scale visible light feature set, the feature sets are fused according to the corresponding scale level. The features at small scale, medium scale and large scale are weighted and fused respectively to obtain the feature fusion response map of each scale. The fusion according to the scale level can complement the infrared and visible light features at the same scale. The generated feature fusion response map of each scale highlights the defect boundary features at the corresponding scale, and a clear defect boundary analysis result is obtained.

[0111] Specifically, the feature fusion response maps at each scale are fused. The feature fusion response maps at different scales reflect different boundary information: the small-scale map reflects detailed boundaries, the medium-scale map reflects transitional boundaries, and the large-scale map reflects the overall boundary. The feature fusion response maps at each scale are first interpolated to the same size, and then the pixel values ​​are weighted and fused according to the corresponding pixel positions to obtain the fused defect boundary response map. Through multi-scale fusion, defect boundary features at different scales can be combined, so that the defect boundary response map can clearly display small edge details and accurately present the overall outline, avoiding the problem of incomplete boundary features under a single scale, and providing a more accurate basis for defect boundary identification and optimization.

[0112] Furthermore, based on the defect boundary response map, the boundary of the first defect region is shrunk or expanded to obtain the second defect region, which includes:

[0113] S801. Based on the defect boundary response diagram, analyze the curvature and constituent area of ​​the defect boundary, and construct boundary constraints and spatial constraints.

[0114] S802. Optimize the boundary of the first defect region through boundary constraints to obtain the first optimized boundary;

[0115] S803. The first optimization boundary is shrunk or expanded by spatial constraints to obtain the second optimization boundary;

[0116] S804. The region formed by the second optimization boundary is taken as the second defect region.

[0117] In this embodiment, the defect boundary is analyzed based on the constructed defect boundary response map. The curvature change of the defect boundary is continuous, with no sudden large curvature abrupt changes at the crack boundary, and the area formed by the defect boundary matches the actual size of the defect. Boundary constraints and spatial constraints are constructed by analyzing the curvature and area of ​​the defect boundary. The curvature values ​​of the boundary points are calculated based on the defect boundary in the defect boundary response map. The calculated curvature values ​​are statistically analyzed, and 1.5 times the average of all curvature values ​​is used as the curvature threshold. The boundary constraint is set so that the curvature of any point on the defect boundary does not exceed the set curvature threshold. The area of ​​the region enclosed by the defect boundary in the defect boundary response map is analyzed, and a spatial constraint is set so that the difference between the area of ​​the first defect region and the area of ​​the defect boundary response map within the defect region enclosed by the corresponding boundary does not exceed a preset error threshold. By analyzing the curvature and area of ​​the defect boundary and constructing corresponding constraints, the defect boundary can be corrected and optimized based on its geometric and spatial characteristics to obtain accurate defect boundaries and defect regions.

[0118] Specifically, the boundary of the first defect region is optimized through boundary constraints to obtain the first optimized boundary. For the boundary in the first defect region, the curvature value of the boundary points is calculated, and points with curvature values ​​greater than the set curvature threshold are extracted as abnormal curvature points. For a single abnormal curvature point, a smooth curve is fitted using two points before and after the point, and the abnormal curvature point is adjusted onto the curve to optimize the boundary. For multiple consecutive abnormal curvature points, the midpoints of multiple abnormal curvature points are deleted, and then the curve is fitted again to adjust the curvature values ​​of the abnormal curvature points so that all abnormal curvature points meet the boundary constraints. By optimizing the boundary through boundary constraints, the first optimized boundary is obtained, which can eliminate unreasonable curvature abrupt changes in the defect boundary, making the defect boundary smoother and more accurate.

[0119] like Figure 3As shown, for the first optimized boundary after smoothing, the area of ​​the defect region enclosed by it may not conform to the spatial constraints. The first optimized boundary is shrunk or expanded according to the spatial constraints to make the area of ​​the defect region enclosed by the optimized boundary conform to the spatial constraints, resulting in a more accurate second optimized boundary. For the defect region enclosed by the first optimized boundary, the area of ​​each defect region is calculated and compared with the area of ​​the corresponding defect region in the defect boundary response map. The area of ​​the defect region enclosed by the first optimized boundary is subtracted from the area of ​​the corresponding defect region in the defect boundary response map to calculate the difference. When the difference is positive and greater than a preset error threshold, the first optimized boundary is shrunk by pixels. When the difference is negative and its absolute value is greater than the preset error threshold, the first optimized boundary is expanded by pixels. By shrunk or expanded, the first optimized boundary is shrunk until the difference is within the preset error threshold, completing the boundary optimization operation and obtaining the second optimized boundary. Optimizing the boundary through spatial constraints can avoid the boundary being too large and containing too many normal areas, or too small and missing some defects, resulting in a more accurate defect region. Using the area enclosed by the second optimization boundary as the second defect area can accurately reflect the actual range of the defect, providing an accurate defect area for defect type analysis and early warning.

[0120] Furthermore, a defect score is calculated within the second defect area using a preset defect warning model. An early warning is issued for areas with defect scores exceeding a preset defect threshold, thereby detecting surface defects on the antenna phase shifter, including:

[0121] S901. By extracting the defect features in the second defect region, a defect feature vector is constructed;

[0122] S902. Based on the defect feature vector, identify the defect type corresponding to the second defect area through the preset defect early warning model and calculate the defect score;

[0123] S903. When the defect score is greater than the preset defect threshold, an early warning is issued in combination with the defect type to detect defects on the surface of the antenna phase shifter.

[0124] In this embodiment, the second defect region includes defect information, including defect type information such as cracks, desoldering, and corrosion. Different types of defects have different characteristics. By extracting the defect features in the second defect region, a corresponding defect feature vector is constructed. The geometric features, thermal features, texture features, location features, and boundary features of the second defect region are extracted by calculation. Geometric features include, but are not limited to, area, perimeter, and circularity. Thermal features include, but are not limited to, average thermal resistance and thermal resistance standard deviation. Texture features include, but are not limited to, edge density and roughness. Location features include, but are not limited to, distance from the antenna center and the region in which it is located. Boundary features include, but are not limited to, boundary complexity and maximum curvature. The extracted defect features are arranged in order to form a defect feature vector.

[0125] Specifically, based on the constructed defect feature vector, the defect type corresponding to the second defect area is identified and the defect score is calculated through a pre-set defect early warning model. The defect early warning model includes, but is not limited to, the random forest model. The random forest model is trained using a large number of historical defect feature vectors and corresponding defect types to obtain a pre-trained random forest model. The constructed defect feature vector is input into the pre-trained random forest model, and the model calculates the corresponding defect type and defect score. Through feature extraction and model recognition, defect type identification and scoring can be completed quickly and accurately, resulting in accurate defect analysis results.

[0126] When the calculated defect score exceeds the preset defect threshold, a defect warning is issued based on the defect type to detect surface defects on the antenna phase shifter. The average defect score from the historical defect data of the phase shifter is used as the defect threshold. When the defect score exceeds the threshold, it indicates a high degree of defect, requiring a warning. The defect type is then synchronized to the warning information to facilitate appropriate maintenance by staff. By using the preset threshold and combining it with type-based warnings, serious defects affecting the phase shifter's performance can be identified promptly, enabling staff to quickly locate and address problems, reducing phase shifter failures caused by defects.

[0127] like Figure 4 As shown, an antenna phase shifter surface defect detection system is used to implement a method for detecting surface defects in an antenna phase shifter, comprising:

[0128] The thermal resistance gradient map construction module collects infrared thermal image data of the antenna phase shifter surface using an infrared thermal imager, analyzes the thermal resistance change of the phase shifter surface based on the infrared thermal image data, and constructs a thermal resistance gradient map.

[0129] The first defect region identification module extracts regions with thermal resistance gradients greater than preset gradient thresholds based on the thermal resistance gradient map, thus obtaining the first defect region.

[0130] The second defect region identification module identifies the defect boundary within the first defect region by combining infrared thermal imaging data and pre-acquired visible light data, thereby obtaining the second defect region.

[0131] The defect warning module calculates the defect score in the second defect area through a preset defect warning model, and issues a warning for areas where the defect score is greater than the preset defect threshold, so as to detect defects on the surface of the antenna phase shifter.

[0132] In this embodiment, the thermal resistance gradient map construction module uses an infrared thermal imager to acquire infrared thermal image data of the antenna phase shifter surface, analyzes the changes in thermal resistance on the antenna phase shifter surface using the infrared thermal image data, and constructs a thermal resistance gradient map based on the changes in thermal resistance. By constructing the thermal resistance gradient map, the changes in thermal resistance at different locations on the phase shifter surface can be clearly displayed, providing data support for identifying defects on the phase shifter surface.

[0133] The first defect region identification module filters out regions with thermal resistance gradients greater than a preset gradient threshold in the constructed thermal resistance gradient map to obtain the first defect region. By identifying regions with large thermal resistance gradients, the defect detection range can be quickly narrowed down, regions with normal thermal resistance gradients can be removed, and defect detection and identification can be performed on regions with defects, thereby improving the efficiency of defect detection.

[0134] The second defect area identification module, within the first defect area, combines infrared thermal imaging data and pre-acquired visible light data to accurately identify and optimize the defect boundary, thereby selecting a more accurate defect range and obtaining the second defect area. By combining the thermal characteristics of infrared data and the structural characteristics of visible light data, compared to using single infrared data for defect detection, the accuracy of defect area identification can be improved, making the defect area boundary more accurate and providing an accurate defect area range for defect type identification and early warning.

[0135] The defect early warning module uses a preset defect early warning model to calculate the defect score of the second defect area. When the score exceeds the preset defect threshold, an early warning is issued, realizing the detection of surface defects of the phase shifter. By calculating the defect score, the severity of surface defects is quantitatively analyzed, which can promptly detect serious surface defects that affect the normal operation of the phase shifter and issue an early warning. This allows staff to quickly take targeted measures to avoid the phase shifter's performance degradation or failure due to defects, and ensure the stable operation of the phase shifter.

[0136] The above description is merely a preferred embodiment of the present invention. The scope of protection of the present invention is not limited to the above embodiments. All technical solutions falling within the scope of the present invention's concept are within the scope of protection of the present invention. It should be noted that for those skilled in the art, any improvements and modifications made without departing from the principle of the present invention should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting surface defects in an antenna phase shifter, characterized in that, include: Infrared thermal image data of the antenna phase shifter surface is acquired by an infrared thermal imager, and the thermal resistance change of the phase shifter surface is analyzed based on the infrared thermal image data to construct a thermal resistance gradient map. Based on the thermal resistance gradient map, the region where the thermal resistance gradient is greater than the preset gradient threshold is extracted to obtain the first defect region; For the first defect region, the infrared thermal image data and the pre-acquired visible light data are decomposed according to a preset scale to obtain a multi-scale infrared image sequence and a multi-scale visible light image sequence. Using a preset feature extraction model, features are extracted from the multi-scale infrared image sequence and the multi-scale visible light image sequence to obtain a multi-scale infrared feature set and a multi-scale visible light feature set. The multi-scale infrared feature set and the multi-scale visible light feature set are fused according to the scale level to generate a feature fusion response map for each scale. The feature response maps of each scale are fused together to construct a defect boundary response map; Based on the defect boundary response diagram, the curvature and constituent area of ​​the defect boundary are analyzed to construct boundary constraints and spatial constraints; The boundary of the first defect region is optimized by the boundary constraints, the curvature value of the boundary points is calculated, and the points with curvature values ​​greater than the set curvature threshold are regarded as abnormal curvature points. For a single abnormal curvature point, a smooth curve is fitted using two points before and after the point, and the abnormal curvature point is adjusted to be on the curve to optimize the boundary. For multiple consecutive abnormal curvature points, after deleting the midpoints of multiple abnormal curvature points, the curvature values ​​of the abnormal curvature points are adjusted by fitting the curve to make all abnormal curvature points conform to the boundary constraints, thus obtaining the first optimization boundary. The first optimization boundary is shrunk or expanded by the spatial constraints to obtain the second optimization boundary; The region formed by the second optimization boundary is taken as the second defect region; The defect score in the second defect area is calculated by a preset defect warning model, and a warning is issued for areas where the defect score is greater than a preset defect threshold, so as to detect defects on the surface of the antenna phase shifter.

2. The method for detecting surface defects of an antenna phase shifter according to claim 1, characterized in that, The process of acquiring infrared thermal image data of the antenna phase shifter surface using an infrared thermal imager, analyzing the thermal resistance changes on the phase shifter surface based on the infrared thermal image data, and constructing a thermal resistance gradient map includes: Infrared thermal image data is collected by applying pulsed thermal excitation to the surface of the antenna phase shifter using an infrared thermal imager. Based on the infrared thermal image data, the thermal resistance value and thermal resistance gradient of the phase shifter surface are calculated using a preset thermal resistance analysis model, and a thermal resistance gradient map is constructed.

3. The method for detecting surface defects of an antenna phase shifter according to claim 2, characterized in that, Based on the infrared thermal image data, the thermal resistance value and thermal resistance gradient of the phase shifter surface are calculated using a preset thermal resistance analysis model, and a thermal resistance gradient map is constructed, including: Based on infrared thermal image data, the thermal resistance value of each pixel position of the phase shifter is calculated using a preset thermal resistance analysis model. The thermal resistance gradient is obtained by calculating the difference in thermal resistance values ​​between adjacent pixels. By combining the thermal resistance value and the thermal resistance gradient, a thermal resistance gradient map is constructed.

4. The method for detecting surface defects of an antenna phase shifter according to claim 1, characterized in that, Based on the thermal resistance gradient map, regions with thermal resistance gradients greater than a preset gradient threshold are extracted to obtain the first defect region, including: Based on the thermal resistance gradient map, regions with thermal resistance gradients greater than a preset gradient threshold are extracted to obtain candidate defect regions; By analyzing the correlation of defects within the candidate defect regions, the candidate defect regions are merged to obtain multiple merged defect regions. The first defect region is obtained by combining the merged defect regions.

5. The method for detecting surface defects of an antenna phase shifter according to claim 4, characterized in that, By analyzing the correlation of defects within the candidate defect regions, the candidate defect regions are merged to obtain multiple merged defect regions, including: Identify the defect endpoints within the candidate defect region and construct a set of defect endpoints; According to the direction of the increase in thermal resistance gradient in the defect region, connect the defect endpoints in the defect endpoint set where the distance between two points is less than a preset distance threshold to obtain the defect boundary set; Calculate the boundary connectivity value of every two defect boundaries in the defect boundary set, and connect the defect boundaries whose boundary connectivity value is greater than a preset connectivity threshold to obtain a set of connected domains; Connected regions in the connected region set whose regional connectivity values ​​are greater than a preset connectivity threshold are merged to obtain multiple merged defect regions.

6. The method for detecting surface defects of an antenna phase shifter according to claim 1, characterized in that, The step of calculating a defect score within the second defect area using a preset defect warning model, and issuing a warning for areas where the defect score exceeds a preset defect threshold, in order to detect surface defects on the antenna phase shifter, includes: By extracting the defect features from the second defect region, a defect feature vector is constructed; Based on the defect feature vector, the defect type corresponding to the second defect area is identified and the defect score is calculated using a preset defect early warning model; When the defect score is greater than the preset defect threshold, an early warning is issued in conjunction with the defect type to detect surface defects of the antenna phase shifter.

7. A surface defect detection system for an antenna phase shifter, characterized in that, A method for detecting surface defects of an antenna phase shifter as described in any one of claims 1 to 6, comprising: The thermal resistance gradient map construction module collects infrared thermal image data of the antenna phase shifter surface using an infrared thermal imager, analyzes the thermal resistance change of the phase shifter surface based on the infrared thermal image data, and constructs a thermal resistance gradient map. The first defect region identification module extracts regions with thermal resistance gradients greater than preset gradient thresholds based on the thermal resistance gradient map, thus obtaining the first defect region. The second defect region identification module identifies the defect boundary within the first defect region by combining infrared thermal imaging data and pre-acquired visible light data, thereby obtaining the second defect region. The defect warning module calculates the defect score in the second defect area using a preset defect warning model, and issues a warning for areas where the defect score is greater than a preset defect threshold, so as to detect defects on the surface of the antenna phase shifter.

Images