Rebar mesh multi-welding point quality detection method and system based on infrared thermal imaging

Abstract

The application provides a steel mesh multi-welding point quality detection method and system based on infrared thermal imaging. For each cross steel welding point, the fusion form of the cross steel welding point is identified based on a thermal radiation image. Based on the fusion forms of all cross steel welding points, a quality detection result of the steel mesh multi-welding point is generated, and the quality detection result includes a geometric judgment result for characterizing whether the fusion form meets a preset form requirement. According to the quality detection result, a corresponding early warning mode is used for adaptive early warning to adjust the welding parameters used in the next resistance welding process according to the early warning information. The technical scheme provided by the application realizes millisecond-level multi-welding point synchronous quality detection and process closed-loop optimization, and greatly improves the production yield and process adaptability of large steel meshes.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of welding point quality detection, and in particular to a steel mesh multi-welding point quality detection method and system based on infrared thermal imaging. BACKGROUND

[0002] In the automatic resistance welding production line of large steel mesh, the welding cycle is extremely short, and the full-width welding point quality detection needs to be completed from the end of welding to the start of mesh pulling. This detection method must complete the detection of dozens of welding points within a preset time, and the detection results need to be fed back to the welding control system in real time to realize the adaptive adjustment of welding parameters.

[0003] The existing scheme adopts a mobile infrared probe array scanning system, which deploys an infrared sensor group that can move horizontally above the welded mesh. The probe is scanned in a partitioned manner by a mechanical arm, and the quality of the nugget is predicted by combining a temperature decay model. The system performs spot checks after the mesh pulling action is completed, and rechecks and verifies the abnormal areas.

[0004] However, the existing scheme has three technical bottlenecks. First, the detection time is much longer than the effective time window due to timing mismatch, and the welding point temperature has dropped to a low temperature stage during acquisition, resulting in the loss of key thermal characteristics. Second, the spatial coverage is limited by the scanning path, and only local welding point sampling detection can be achieved, and the full-width synchronous temperature field cannot be obtained. Third, the dynamic error is amplified due to the movement and vibration of the probe, causing distortion of the temperature gradient curve, and the predicted results of the welding point shear resistance are seriously deviated from the true value. SUMMARY

[0005] The present application provides a steel mesh multi-welding point quality detection method and system based on infrared thermal imaging, to solve the problem of low accuracy of multi-welding point quality detection caused by the lack of timeliness of thermal characteristics, insufficient global coverage, and measurement distortion under dynamic interference in the prior art.

[0006] In a first aspect, the present application provides a steel mesh multi-welding point quality detection method based on infrared thermal imaging, comprising:

[0007] Within the time interval from the end of the resistance welding process to the start of the mesh pulling action, the infrared thermal imager array synchronously acquires thermal radiation images of the surface temperature of all cross-reinforcement welding points obtained by the resistance welding process in this time interval.

[0008] For each cross-reinforcement welding point, the fusion form of the cross-reinforcement welding point is identified based on the thermal radiation image.

[0009] Based on the fusion form of all cross-reinforcement welding points, a quality detection result of the steel mesh multi-welding point is generated, and the quality detection result includes a geometric judgment result for representing whether the fusion form meets the preset form requirement.

[0010] According to the quality detection result, a corresponding early warning mode is used for adaptive early warning, so as to adjust the welding parameters used in the next resistance welding process according to the early warning information.

[0011] Optionally, the fusion form includes a pixel area value and a nugget diameter.

[0012] The fusion form of the cross-reinforcement welding point is identified based on the thermal radiation image, and includes:

[0013] The thermal radiation image is preprocessed to obtain a processed thermal radiation image.

[0014] The boundary region of the cross-reinforcement welding point is identified from the processed thermal radiation image, and a pixel area value of the boundary region is calculated.

[0015] According to the pixel area value of the boundary region, a diameter inversion operation is performed using a pre-established diameter calibration model to obtain the nugget diameter of the cross-reinforcement welding point, and parameters of the diameter calibration model are calibrated according to the reinforcement diameter and the welding parameters.

[0016] Optionally, the boundary region of the cross-reinforcement welding point is identified from the processed thermal radiation image, and includes:

[0017] All continuous pixel regions with a temperature value greater than a preset temperature threshold in the processed thermal radiation image are identified by a temperature threshold segmentation method.

[0018] According to the continuous pixel region, a topological relationship graph representing pixel connectivity is established, boundary node sets in the topological relationship graph are converted into a binary image matrix through coordinate mapping, and an initial boundary mask is generated.

[0019] The initial boundary mask is subjected to a hole filling process to obtain a filled boundary mask.

[0020] The filled boundary mask is subjected to an edge smoothing process to obtain a smoothed boundary mask.

[0021] Based on the smoothed boundary mask, a temperature gradient value of a pixel corresponding to a boundary node is calculated, and pixels with a temperature gradient value greater than or equal to a preset temperature gradient threshold are screened to determine the boundary region of the cross-reinforcement welding point.

[0022] Optionally, based on the smoothed boundary mask, a temperature gradient value of a pixel corresponding to a boundary node is calculated, and pixels with a temperature gradient value greater than or equal to a preset temperature gradient threshold are screened to determine the boundary region of the cross-reinforcement welding point.

[0023] Based on the smoothed boundary mask, the surface temperature of the boundary node in the thermal radiation image is obtained, and the surface temperature of the cross-reinforcement welding point in another thermal radiation image is combined to calculate the temperature gradient value of the pixel corresponding to the boundary node along the boundary normal direction, wherein the boundary normal direction is determined according to the coordinates of the adjacent nodes of the topological relationship diagram.

[0024] A temperature gradient threshold corresponding to the reinforcement diameter is obtained from a preset configuration table.

[0025] Pixels with a temperature gradient value greater than or equal to a preset temperature gradient threshold are screened to determine the boundary region of the cross-reinforcement welding point.

[0026] Optionally, according to the pixel area value of the boundary region, a diameter inversion operation is performed using a pre-established diameter calibration model to obtain the nugget diameter of the cross-reinforcement welding point, and parameters of the diameter calibration model are calibrated according to the reinforcement diameter and the welding parameters, including:

[0027] According to the reinforcement diameter and the welding parameters, the parameters of the diameter calibration model are matched from a preset calibration library.

[0028] According to the pixel area value of the boundary region, a diameter inversion operation is performed using a pre-established diameter calibration model in combination with a welding position compensation factor to obtain the nugget diameter of the cross-reinforcement welding point, and the welding position compensation factor is dynamically calculated by the electrode offset.

[0029] Optionally, the thermal radiation image is preprocessed to obtain a processed thermal radiation image, including:

[0030] The thermal signals of different frequency bands in the thermal radiation image are separated to remove environmental interference and extract nugget edge information to obtain a filtered thermal radiation image.

[0031] Based on the background temperature value of the non-welding area around the welding point in the filtered thermal radiation image, a sensitivity correction value is calculated to eliminate the radiation error caused by the oxidation of the reinforcement surface to obtain a thermal radiation image after radiation error elimination.

[0032] The welding point core region in the thermal radiation image after radiation error elimination is subjected to adaptive histogram equalization to obtain a processed thermal radiation image.

[0033] Optionally, the welding point core region in the thermal radiation image after radiation error elimination is subjected to adaptive histogram equalization to obtain a processed thermal radiation image, including:

[0034] A circular region with a diameter of the reinforcement diameter is circumscribed as the welding point core region with the highest temperature point in the thermal radiation image as the center.

[0035] In the core area of the welding spot, a preset pixel sliding window is used to traverse all pixels, and a probability distribution function of temperature values in the window is calculated;

[0036] A mapping relationship from temperature values to equalized gray values is established according to the probability distribution function;

[0037] The temperature values of all pixels in the window are transformed by using the mapping relationship;

[0038] The transformation results of adjacent windows are fused by using bilinear interpolation, and the transformed temperature values of the core area of the welding spot are subjected to dynamic range compression processing to suppress over-enhanced areas, so as to obtain a processed thermal radiation image.

[0039] In a second aspect, the present application provides a steel mesh multi-welding spot quality detection system based on infrared thermal imaging, comprising:

[0040] A collection module is configured to collect, within a time interval from the end of the current resistance welding process to the start of the self-net drawing action, thermal radiation images of surface temperatures of all cross-reinforcement welding spots by using an infrared thermal imager array, wherein the cross-reinforcement welding spots are welding spots obtained by the current resistance welding process of the steel mesh;

[0041] An identification module is configured to identify, for each cross-reinforcement welding spot, a fusion form of the cross-reinforcement welding spot based on the thermal radiation image;

[0042] A generation module is configured to generate a quality detection result of the steel mesh multi-welding spot based on the fusion form of all cross-reinforcement welding spots, wherein the quality detection result comprises a geometric judgment result for representing whether the fusion form meets a preset form requirement;

[0043] A warning module is configured to perform adaptive warning by using a corresponding warning mode according to the quality detection result, so as to adjust welding parameters used in the next resistance welding process according to warning information.

[0044] In a third aspect, the present application provides a computing device comprising a processing component and a storage component; the storage component stores one or more computer instructions; the one or more computer instructions are used to be called and executed by the processing component to implement a steel mesh multi-welding spot quality detection method based on infrared thermal imaging according to any one of the first aspect.

[0045] In a fourth aspect, the present application provides a computer storage medium storing a computer program, wherein the computer program is executed by a computer to implement a steel mesh multi-welding spot quality detection method based on infrared thermal imaging according to any one of the first aspect.

[0046] In the present application, a steel mesh multi-weld point quality detection method based on infrared thermal imaging is provided, which comprises: within the time interval from the end of the present resistance welding process to the start of the self-net drawing action, synchronously collecting thermal radiation images of the surface temperature of all cross steel reinforcement weld points by an infrared thermal imager array, the cross steel reinforcement weld points being the weld points obtained by the present resistance welding process in the steel mesh; for each cross steel reinforcement weld point, identifying the fusion form of the cross steel reinforcement weld point based on the thermal radiation image; generating a quality detection result of the steel mesh multi-weld point based on the fusion forms of all cross steel reinforcement weld points, the quality detection result comprising a geometric judgment result for characterizing whether the fusion form meets a preset form requirement; and according to the quality detection result, performing adaptive early warning by using a corresponding early warning mode to adjust the welding parameters used in the next resistance welding process according to the early warning information.

[0047] In the present application, by synchronously collecting full-width weld point thermal radiation images within a very short time window from the end of resistance welding to the start of net drawing, the time effectiveness of thermal characteristics is solved, ensuring the capture of transient high temperature field; based on thermal image recognition of each weld point fusion form, millisecond-level parallel quantitative analysis of dozens of weld points is realized; by integrating all weld point quality data to generate a global detection result, the welding control system is linked to perform adaptive early warning and parameter adjustment, forming a closed-loop control from detection to process optimization, and improving the production yield and process stability of large meshes.

[0048] Further, when identifying the weld point fusion form based on the thermal radiation image, first, the image is preprocessed, then the continuous pixel region greater than a preset temperature threshold is extracted by temperature threshold segmentation; the topological relationship graph of the region is constructed and converted into a binary mask, and an optimized boundary mask is generated after hole filling and edge smoothing processing; finally, the weld point accurate boundary region is determined based on temperature gradient threshold screening, the pixel area value is calculated, and the fusion core diameter value is output by using the pre-calibrated diameter inversion model. This scheme improves the boundary anti-interference ability through temperature gradient threshold screening and mask optimization processing, effectively suppressing the recognition deviation of the fusion core region caused by heat diffusion; the pre-calibration model integrates the steel reinforcement diameter and the key variables of the welding parameters, realizing the universality of diameter inversion for multi-specification mesh production lines; the whole-process algorithm has linear calculation efficiency, and the single-weld point analysis time is extremely low, meeting the millisecond-level real-time requirement of synchronous detection of high-density weld point array.

[0049] These aspects or other aspects of the present application will be more apparent in the following description of the embodiments. BRIEF DESCRIPTION OF DRAWINGS

[0050] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings described below are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0051] Figure 1 A flow chart of a steel mesh multi-welding point quality detection method based on infrared thermal imaging provided by an embodiment of the present application is shown in the figure.

[0052] Figure 2 A structural schematic diagram of a steel mesh multi-welding point quality detection system based on infrared thermal imaging provided by an embodiment of the present application is shown in the figure.

[0053] Figure 3 A structural schematic diagram of a computing device provided by an embodiment of the present application is shown in the figure. DETAILED DESCRIPTION

[0054] In order to make those skilled in the art better understand the present application, the technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application.

[0055] In some of the processes described in the specification and claims of the present application and the above-described drawings, a plurality of operations appear in a specific order, but it should be clearly understood that these operations can be executed or performed in parallel or in a different order from that in which they appear in the text. The serial numbers of the operations, such as 11, 12, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these processes can include more or fewer operations, and the operations can be executed in sequence or in parallel. It should be noted that the descriptions of "first", "second", etc. in the text are used to distinguish different messages, devices, modules, etc., and do not represent the order of precedence. Also, "first" and "second" are not of different types.

[0056] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only some embodiments of the present application, not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0057] To solve the problem of low accuracy and low efficiency of multi-welding point quality synchronous detection caused by the lack of heat feature timeliness, the lack of global coverage ability and the distortion of measurement under dynamic interference in the prior art, the embodiment of the present application provides a steel mesh multi-welding point quality detection method based on infrared thermal imaging. The method adopts the following concept: by accurately locking the high-fidelity window period of the heat feature from the end of the resistance welding process to the start of the self-net-drawing action, the spatially distributed infrared thermal imager array is used to synchronously capture the global welding point transient temperature field; the multi-level processing chain based on the thermal radiation image is used to realize the millisecond-level quantitative analysis of the fusion form, and the dynamic calibration model adaptive to the process parameters is used to inverse the nugget diameter; the quality detection result is generated through the full-width geometric feature compliance judgment, and the closed-loop process optimization is triggered based on the defect distribution, thereby systematically solving the problem of low accuracy and low efficiency of multi-welding point quality synchronous detection.

[0058] Figure 1 A flow chart of a steel mesh multi-welding point quality detection method based on infrared thermal imaging provided by the embodiment of the present application is shown in Figure 1 The method comprises the following steps:

[0059] S11, in the time interval from the end of the present resistance welding process to the start of the self-net-drawing action, the thermal radiation images of the surface temperature of all the cross steel welding points are synchronously collected by the infrared thermal imager array. The cross steel welding point is the welding point obtained by the present resistance welding process in the steel mesh.

[0060] The resistance welding process refers to a metal processing method that uses current passing through the contact surface of the steel to generate resistance heat for fusion connection, including three elements of electrode pressure application, welding current conduction and welding time control. The self-net-drawing action refers to the mechanical operation of pulling the steel mesh out of the station by the pulling mechanism after welding, and the start mark detection window is closed. The time interval refers to the process gap period from the electrode separation from the welding point to the triggering of the self-net-drawing mechanism, which is used to accommodate the detection operation. The infrared thermal imager array refers to a spatially distributed imaging system composed of multiple high-frame-rate thermal imagers, covering the full width of the mesh to capture the temperature field. Synchronous acquisition can refer to the simultaneous triggering of exposure of all thermal imagers based on a unified timing signal, ensuring the spatio-temporal alignment of the temperature field data of dozens of welding points. The thermal radiation image is a two-dimensional matrix recording the infrared radiation intensity of the welding point surface, and the pixel value can be mapped to the temperature distribution.

[0061] The cross steel welding point refers to the newly generated welding connection point at the intersection of the steel in a single resistance welding cycle. The steel mesh is a grid-shaped building component composed of longitudinal and transverse steels arranged in a fixed interval. For example, the steel mesh is 32 meters long and 12 meters wide.

[0062] In the embodiment of the present application, first, during the time interval from the end of the resistance welding process to the start of the self-net drawing action, the synchronous acquisition operation is performed by covering the full-width infrared thermal imager array, and the surface temperature thermal radiation image of all the cross steel bar welding points is captured at one time; wherein the cross steel bar welding point can refer to the welding point formed on the steel mesh by the resistance welding process this time, which is a row of steel mesh and can be dozens of welding points, to ensure the timeliness of the data and the accurate correspondence of the welding position.

[0063] S12, for each cross steel bar welding point, based on the thermal radiation image, the fusion form of the cross steel bar welding point is identified.

[0064] Wherein, the fusion form can be a composite index representing the geometric and metallurgical characteristics of the welding point fusion zone, including the fusion zone diameter and the characteristics of the heat affected zone.

[0065] In the embodiment of the present application, for each cross steel bar welding point, first, the image processing technology is used to extract the thermal characteristics of the welding point, and the spatial distribution of the fusion form is determined by the temperature gradient analysis and boundary recognition algorithm; Specifically: calculate the high temperature area pixel area and inverse the fusion zone diameter, finally output the quantitative fusion form parameters containing geometric size and thermodynamic characteristics.

[0066] S13, based on the fusion form of all cross steel bar welding points, the quality detection result of the steel mesh multi-welding point is generated, and the quality detection result includes the geometric judgment result for representing whether the fusion form meets the preset form requirement.

[0067] Wherein, the quality detection result is the structured data output of the welding point compliance state, including the fusion zone diameter threshold determination and the shear resistance prediction conclusion. The geometric judgment result refers to the binary judgment flag based on the proportion relationship between the effective diameter of the fusion zone and the diameter of the steel bar.

[0068] In the embodiment of the present application, based on the fusion form data of all cross steel bar welding points, first, the geometric judgment result of each welding point is compared with the preset form requirement; then the quality detection result covering the whole steel mesh is generated by the spatial aggregation algorithm, which contains the binary judgment of the compliance of each welding point fusion form and the overall quality distribution thermodynamic map.

[0069] S14, according to the quality detection result, adaptive early warning is carried out by using the corresponding early warning mode, so as to adjust the welding parameters used in the next resistance welding process according to the early warning information.

[0070] The pre-warning mode can be an alarm mechanism according to the defect severity classification, including an audible and visual warning, a parameter adjustment instruction and a defect position marker. The adaptive pre-warning is an intelligent feedback mechanism that dynamically selects the pre-warning level according to the real-time detection result and triggers the process optimization. The pre-warning information can be a composite data packet containing the defect position coordinates, the abnormal type and the recommended process correction parameters. The welding parameters refer to the core variables for controlling the resistance welding quality, including the electrode pressure, the welding current and the welding time.

[0071] In the embodiments of the present application, according to the quality detection result, first, the pre-warning mode is matched according to the defect level and the adaptive pre-warning signal is triggered; then the pre-warning information is fed back to the welding control system in real time, and the welding current, the electrode pressure and the welding time of the next resistance welding process are dynamically optimized to form a process closed-loop control chain.

[0072] Alternatively, the following is another steel mesh multi-welding point quality detection method based on infrared thermal imaging, which includes the following processes:

[0073] Step 1: Within the time window of 0.3-0.8 seconds after the resistance welding electrode is separated from the steel mesh and before the net pulling action is started, the thermal radiation images of all welding points are synchronously collected by the infrared thermal imager array arranged above the welding net plane. The field of view of the infrared thermal imager array can cover a welding net width of not less than 2 meters, and the imaging frame rate can be not less than 200 Hz to capture the transient temperature field distribution during the cooling process of the welding points.

[0074] The steel mesh can be a large-diameter large mesh with a maximum length of 32 meters and a maximum width of 12 meters, and the steel bar spacing can be 0.125 meters, so the number of multi-welding points can be 12÷0.125=96. In the process of resistance spot welding, three stages of pre-pressing, power-on and forging pressing should be included. In the power-on stage, the following welding parameters are used: electrode pressure, welding current and welding time. Multiple welding points are welded at the same time, and after welding, there is a holding time in the forging pressing stage. The large mesh can select appropriate electrode pressure, welding current and welding time according to the diameter of the steel bar. During the welding process, the local high temperature is above 1000 degrees, and the steel bars are stacked, so it is not easy to find out whether there is a missing welding or a too large or too small welding seam after welding. In order to realize the synchronous detection of the quality of the multi-welding points based on the thermal imaging data.

[0075] Step 2: The detection system communicates with the welding control system in real time through the industrial bus, receives the welding timing signal to automatically trigger image acquisition, and stores the thermal imaging data and the welding position coordinates in space-time matching.

[0076] Step 3, image analysis based on a pre-established welding spot temperature gradient calibration model, which establishes a corresponding temperature gradient influence map range and a molten core morphology standard feature library according to a parameter combination of a steel bar diameter of 12 mm, a welding current of 25 kA, and a current time of 2.5 s.

[0077] Specifically, the "temperature gradient" is explicitly split into a molten core diameter and a shear resistance, which correspond to the spatial distribution and time evolution characteristics of the thermal image, respectively. The determination process of the molten core diameter is as follows: according to the parameter combination of the steel bar diameter, the welding current, and the current time, a linear regression equation of the area of the high-temperature zone of the welding spot with a temperature greater than or equal to 1000°C and the molten core diameter is established; for shear resistance prediction, the cooling rate of the welding spot temperature gradient curve is associated with the metal bonding strength, the temperature drop slope threshold is calibrated, and when the slope is lower than the critical value, it is determined that the shear resistance is insufficient. Step 3 can be understood as a model establishment stage, and a mapping model between temperature and quality is established through experimental calibration, which provides a criterion for real-time detection in step 4.

[0078] Step 4, real-time calculation of the three-dimensional temperature field distribution of each welding spot by the edge computing unit, extraction of the kurtosis coefficient and the drop rate characteristic value of the temperature gradient curve, and determination of whether the molten core diameter is less than the steel bar diameter in combination with the critical threshold in the calibration model.

[0079] Specifically, the molten core diameter calculation process is as follows: the area of the pixel region with a temperature greater than or equal to 1000°C in the welding spot thermal image is extracted, the molten core diameter is calculated based on the calibration equation, and it is determined whether it is less than the steel bar diameter; the shear resistance evaluation process is as follows: the temperature gradient curve is extracted along the radial direction of the welding spot, the slope in the interval of 600°C to 300°C in the cooling stage is calculated, and if the slope is lower than the calibration threshold, such as 0.15°C / ms, it is determined that the shear resistance is defective. Step 4 is a model application stage, in which the molten core diameter calculation and shear resistance evaluation algorithms are executed in real time based on the model parameters established in step 3 during line detection, and quality determination is completed.

[0080] Step 5, when an abnormal welding spot is detected, a quality warning signal is sent through the line PLC system before the next welding cycle starts, and the defect position coordinates and temperature characteristic parameters are automatically recorded, and the defect welding spot is physically marked by a code spraying device.

[0081] For example, when either the molten core diameter or the shear resistance is abnormal, a hierarchical warning (light, medium, and heavy defects) is triggered by the line PLC system before the next welding cycle, the mechanical arm is controlled to mark the defect welding spot with laser, and the abnormal data is associated with the welding process parameter library to generate a process optimization suggestion report.

[0082] In summary, the embodiment of the present application firstly synchronously collects the thermal radiation image of the whole steel mesh through the infrared thermal imager array across the twelve-meter width when the electrode is lifted at the end of the resistance welding process, capturing the transient temperature field of all cross steel reinforcement welding points. Secondly, for each welding point thermal image, the temperature gradient threshold segmentation and mask optimization algorithm is used to extract the fusion zone boundary area, calculate the pixel area and inverse the fusion zone diameter value. Then the geometric judgment results of all welding points are aggregated to generate a quality detection report containing fusion zone diameter compliance and shear force prediction. Finally, according to the defect distribution, a hierarchical warning is triggered, the current parameters of the next welding cycle are adjusted synchronously, and the problem welding point position is marked, completing the closed-loop control from detection to process optimization.

[0083] By performing S11-S14, the embodiment of the present application solves the problem of capturing the rapid decay of high-temperature welding point thermal characteristics by precisely locking the detection time window between welding and mesh pulling; realizes millisecond-level quantitative analysis of dozens of welding points by using parallel image processing; generates process optimization instructions based on full-width quality data, effectively improving the welding qualification rate of large steel mesh and the intelligent level of the production line.

[0084] In a possible embodiment, the fusion form includes a pixel area value and a fusion zone diameter. S12, for each cross steel reinforcement welding point, based on the thermal radiation image, the fusion form of the cross steel reinforcement welding point is identified, including:

[0085] Step 121, pre-processing the thermal radiation image to obtain a processed thermal radiation image.

[0086] Wherein, the pre-processing is a sequence of noise suppression, radiation error correction and contrast enhancement operations on the original thermal radiation image, providing an optimized data basis for subsequent analysis.

[0087] In the embodiment of the present application, as described in subsequent steps c1-c3, the thermal signals of different frequency bands in the thermal radiation image are separated by frequency domain filtering and the fusion zone edge information is extracted; the sensitivity correction value is calculated based on the background temperature of the non-welding area around the welding point in the filtered thermal radiation image to eliminate the radiation error caused by the oxidation of the steel reinforcement surface; finally, adaptive histogram equalization is implemented in the core area of the welding point to enhance the temperature distribution contrast, and the processed thermal radiation image is finally output.

[0088] For example, in the filtered thermal radiation image, the center of the cross steel reinforcement welding point is taken as the origin, and a ring-shaped area with a radius of 3-5 times the diameter of the steel reinforcement is defined as a candidate area to ensure that the area is not affected by welding heat; a continuous pixel block with uniform temperature distribution (standard deviation of temperature ≤5℃) is selected from the candidate area, and the area of the pixel block is not less than 100 pixels, as a non-welding reference area; the average temperature of all pixels in the non-welding reference area is calculated, denoted as background measurement temperature ; the thermocouple data collected synchronously with the infrared thermal imager is taken as a benchmark, specifically: before the production of the steel mesh, preset the thermocouple sensor in the non-welding area, and the measurement point coincides with the spatial position of the non-welding reference area in the field of view of the infrared thermal imager; after the end of the resistance welding process, the actual temperature measured by the thermocouple is recorded synchronously, which is recorded as the background actual temperature ; standard emissivity of unoxidized steel bar , the standard emissivity is deviated by the oxidation of the steel bar surface, which is calibrated by experiment in advance , so that the measured temperature of the infrared thermal imager deviates from the actual temperature. The calculation formula of the sensitivity correction value K1 is: K1= , wherein K1 represents the radiation measurement deviation coefficient caused by surface oxidation, which can be used as a sensitivity correction value: when K1=1, there is no radiation error; when K1>1, the measured value of the infrared thermal imager is low; when K1<1, the measured value of the infrared thermal imager is high. Multiply the temperature value of each pixel in the filtered thermal radiation image by the sensitivity correction value K1 to obtain the thermal radiation image after the radiation error is eliminated, so as to eliminate the radiation error caused by the oxidation of the steel bar surface.

[0089] Step 122, identify the boundary area of the cross steel bar welding point from the processed thermal radiation image, and calculate the pixel area value of the boundary area.

[0090] , wherein the boundary area is a closed area of the fusion core edge defined by the temperature gradient threshold, and the spatial range determines the calculation accuracy of the fusion core diameter. The pixel area value refers to the total number of all effective pixels in the boundary area of the welding point in the thermal image, which reflects the projection coverage range of the fusion core in the two-dimensional plane.

[0091] In the embodiments of the present application, the cross steel bar welding point boundary is first identified from the pre-processed thermal radiation image: the temperature threshold segmentation technology is used to extract the high-temperature continuous pixel area, a pixel connectivity topology graph is constructed and an initial boundary mask is generated; then the mask is executed with a hole filling and edge smoothing operation; finally, the accurate boundary area is determined based on the temperature gradient threshold screening, and the pixel area value of the area is calculated.

[0092] Step 123, according to the pixel area value of the boundary area, the diameter inversion operation is performed by using the pre-established diameter calibration model, and the fusion core diameter of the cross steel bar welding point is obtained, and the parameters of the diameter calibration model are calibrated according to the steel bar diameter and the welding parameters.

[0093] The diameter calibration model is a mathematical mapping relationship established based on the reinforcement specification and welding process parameters, and is used to convert the pixel area into a physical size. The diameter inversion operation can be understood as a calculation process of inversely deducing the actual nugget diameter from the boundary pixel area value by using the linear regression equation in the calibration model. The nugget diameter represents the key geometric size of the effective bonding surface of the resistance spot welding fusion zone, and needs to reach a certain proportion of the reinforcement diameter to meet the strength requirement. The welding parameters include welding current, welding time, electrode pressure, etc.

[0094] In the embodiments of the present application, first, the corresponding diameter calibration model parameters are matched from the preset calibration library according to the reinforcement diameter and the welding current and welding time; then the boundary area pixel area value is input into the diameter calibration model combined with the welding position compensation factor; finally, the actual physical size value of the nugget diameter is output through the inversion operation.

[0095] The following is a specific example: first, the collected thermal radiation image is preprocessed: the environmental thermal noise is filtered by band separation, the radiation sensitivity error is corrected by using the background temperature value, and the adaptive histogram equalization is implemented in the core area of the welding spot. Second, the welding spot boundary is extracted from the processed image: the high-temperature continuous region is locked by applying temperature threshold segmentation, after mask generation, hole filling and edge smoothing, the final boundary region is determined according to the temperature gradient threshold and the pixel area value is calculated. Finally, according to the reinforcement diameter and welding current parameters, the corresponding calibration model is called, the pixel area value is input into the inversion equation combined with the electrode offset compensation factor, and the actual nugget diameter value is output for quality judgment.

[0096] By performing steps 121-123, the embodiments of the present application eliminate thermal radiation interference through fine preprocessing, ensuring the reliability of boundary recognition; based on the temperature gradient threshold and the mask optimization technology, the sub-pixel level positioning of the nugget region is realized; combined with the self-adaptive calibration model of the process parameters, the physical size inversion of the nugget diameter is completed in a very short time, providing the core geometric basis for the welding quality judgment.

[0097] In one possible embodiment, step 122, the boundary region of the cross-reinforcement welding spot is identified from the processed thermal radiation image, comprising:

[0098] Step a1, by temperature threshold segmentation, all continuous pixel regions with a temperature value greater than a preset temperature threshold in the processed thermal radiation image are identified.

[0099] The temperature threshold segmentation refers to a binary processing method of dividing the thermal image into a target high-temperature area and a background area according to a preset temperature threshold. The continuous pixel region is a connected image subset composed of spatially adjacent pixels that meet the temperature condition.

[0100] In the embodiment of the present application, first, the temperature threshold segmentation technique is used to process the thermal radiation image, the temperature value of each pixel is compared with the preset temperature threshold, and all pixels exceeding the threshold are screened out; then, based on the spatial adjacency relationship of the pixels, a connected pixel cluster is identified, and a continuous pixel region set representing a high-temperature region is formed.

[0101] Step a2, according to the continuous pixel region, a topological relationship graph representing pixel connectivity is established, the boundary node set in the topological relationship graph is converted into a binary image matrix through coordinate mapping, and an initial boundary mask is generated.

[0102] Among them, the topological relationship graph refers to a graph data structure describing the adjacency relationship of pixels, the node represents the pixel position, and the edge represents the spatial connectivity. The coordinate mapping refers to the space conversion operation of converting the node coordinates in the topological graph into the index of the two-dimensional image matrix. The boundary node set refers to the pixel coordinate set located at the edge of the region contour in the topological graph. The binary image matrix refers to a digital matrix containing only two values 0 and 1, where 0 is the background and 1 is the target. The initial boundary mask is an initial binary contour template generated by the boundary node set.

[0103] In the embodiment of the present application, first, a topological relationship graph is established for the continuous pixel region, and the boundary node set of the region is determined by analyzing the adjacency between pixels; second, the boundary nodes in the topological graph are converted into a two-dimensional coordinate point array using coordinate mapping technology; finally, a binary image matrix composed of 0 and 1 is generated, where the 1 value region corresponds to the initial boundary mask.

[0104] Step a3, performing hole filling processing on the initial boundary mask to obtain a filled boundary mask.

[0105] Among them, the hole filling processing refers to a morphological operation of filling pixel values in the closed blank area inside the binary mask. The filled boundary mask refers to a complete region contour template after eliminating internal holes.

[0106] In the embodiment of the present application, first, the closed blank area in the initial boundary mask is scanned; second, the pixel filling operation is performed on the unconnected area inside the mask; finally, the filled boundary mask after eliminating internal holes is output.

[0107] Step a4, performing edge smoothing processing on the filled boundary mask to obtain a smoothed boundary mask.

[0108] Among them, the edge smoothing processing refers to an image optimization process of eliminating boundary sawtooth mutations through a filtering algorithm. The smoothed boundary mask refers to an optimized contour template with continuous and smooth edge characteristics.

[0109] In the embodiments of this application, the jagged edges of the filled boundary mask are first detected; then, a Gaussian convolution kernel is used to perform smoothing filtering along the boundary; finally, a natural boundary contour with continuous edge transition is generated, that is, the smoothed boundary mask.

[0110] Step a5: Based on the smoothed boundary mask, calculate the temperature gradient value of the corresponding pixel of the boundary node, and filter the pixels whose temperature gradient value is greater than or equal to the preset temperature gradient threshold to determine the boundary area of ​​the cross steel bar weld.

[0111] The temperature gradient value represents the rate of temperature change per unit spatial distance and is calculated based on the temperature difference between adjacent pixels. The temperature gradient threshold is a critical value used to determine the thermal conductivity characteristics of the melt nucleus boundary and is set based on the material's thermophysical properties.

[0112] In this embodiment, firstly, boundary node pixels are located based on the boundary mask after smoothing; secondly, the temperature change rate of adjacent pixels in the node normal direction is calculated as the temperature gradient value; finally, pixels with temperature gradient values ​​exceeding a preset temperature gradient threshold are selected to determine the final boundary region of the solder joint.

[0113] Here's a specific example: First, high-temperature continuous pixel regions in the thermal image are extracted by segmentation using a temperature threshold. Next, a topological graph of the region's pixels is established, and the boundary node set is mapped to a binary matrix to generate an initial boundary mask. Then, a hole-filling operation is performed on this mask to eliminate internal gaps, followed by edge smoothing to obtain a continuous contour. Finally, based on the smoothed mask, the temperature gradient values ​​of the boundary nodes are calculated, and nodes exceeding the threshold are selected to determine the precise boundary region of the solder joint.

[0114] By executing steps a1 to a5, this embodiment of the application effectively suppresses the boundary blurring phenomenon caused by thermal diffusion through a multi-level mask optimization and temperature gradient screening mechanism, and achieves sub-pixel-level precise positioning of the melt core edge, providing a high-precision spatial reference for subsequent geometric parameter calculation.

[0115] In one possible embodiment, step a5, based on the smoothed boundary mask, calculates the temperature gradient value of the pixels corresponding to the boundary nodes, and filters pixels whose temperature gradient values ​​are greater than or equal to a preset temperature gradient threshold to determine the boundary region of the intersecting steel bar weld points, includes:

[0116] Step a51: Based on the smoothed boundary mask, obtain the surface temperature of the boundary nodes in the thermal radiation image, and combine it with the surface temperature of the cross steel bar weld in another thermal radiation image. Calculate the temperature gradient value of the corresponding pixel of the boundary node pixel by pixel along the boundary normal direction. The boundary normal direction is determined according to the coordinates of the adjacent nodes in the topology graph.

[0117] Wherein, the surface temperature refers to the surface radiation temperature value of the solder joint captured by the infrared thermal imager, reflecting the heat energy distribution state of the welding area. The boundary normal direction refers to the spatial vector perpendicular to the boundary contour line, which is obtained by rotating ninety degrees after calculating the tangent line through the coordinates of the adjacent nodes in the topological graph. The adjacent nodes refer to the adjacent pixel points directly connected to the target node in the topological relationship graph, which are used to determine the local geometric features.

[0118] In the embodiment of the present application, first, the boundary node positions are located based on the smoothed boundary mask, and the surface temperature values of these nodes in the current thermal radiation image are obtained; second, the surface temperature data of the same solder joint collected at another time are combined, and the temperature change is calculated pixel by pixel along the boundary normal direction; wherein the boundary normal direction is determined through the geometric relationship of the coordinates of the adjacent nodes in the topological relationship graph, and finally the temperature gradient value of each boundary node is output.

[0119] Step a52, obtaining the temperature gradient threshold value corresponding to the diameter of the steel bar from the preset configuration table.

[0120] Wherein, the preset configuration table refers to a structured database storing the mapping relationship between the diameter of the steel bar and the temperature gradient threshold value, which is pre-calibrated and established based on the thermal conduction characteristics of the material.

[0121] In the embodiment of the present application, first, the gradient threshold configuration table is queried according to the diameter specification of the cross steel bar solder joint; second, the temperature gradient threshold value strictly matched with the diameter of the steel bar is extracted from the configuration table, providing a critical reference for subsequent boundary determination.

[0122] Step a53, screening the pixels with temperature gradient values greater than or equal to the preset temperature gradient threshold value to determine the boundary region of the cross steel bar solder joint.

[0123] In the embodiment of the present application, first, the temperature gradient value calculated by the boundary node is compared with the obtained temperature gradient threshold value; second, all effective pixels with temperature gradient values greater than or equal to the threshold value are screened; finally, the pixel coordinate set is determined as the final accurate boundary region of the cross steel bar solder joint.

[0124] The following is a specific example: first, the surface temperature of the boundary node in the current frame is obtained based on the smoothed boundary mask, and the gradient value is calculated pixel by pixel along the boundary normal direction combined with the temperature data at the same position in the previous frame. Second, the corresponding gradient threshold value is obtained by querying the preset configuration table according to the diameter of the steel bar. Finally, the node pixels with gradient values exceeding the threshold value are screened to determine the accurate boundary region of the solder joint for subsequent quality analysis.

[0125] By performing steps a51~a53, the embodiment of the application effectively distinguifies the physical boundary between the nugget zone and the heat affected zone by dynamically calculating the temperature gradient; in combination with the threshold matching mechanism adaptive to the steel bar specification, the physical accuracy of the boundary positioning is improved, laying a spatial reference for the calculation of the nugget diameter.

[0126] In a possible embodiment, step 123, according to the pixel area value of the boundary region, performs diameter inversion operation by using a pre-established diameter calibration model to obtain the nugget diameter of the cross-reinforcement welding point, the parameters of the diameter calibration model being calibrated according to the steel bar diameter and the welding parameters, including:

[0127] Step b1, according to the steel bar diameter and the welding parameters, matches the parameters of the diameter calibration model from a pre-set calibration library.

[0128] The pre-set calibration library refers to a database storing calibration parameters under different combinations of steel bar specifications and welding processes, containing the mapping relationship between the nugget diameter and the pixel area and the error correction coefficient.

[0129] In the embodiment of the application, first, the numerical value of the steel bar diameter associated with the cross-reinforcement welding point, the intensity value of the welding current and the value of the welding current time are extracted; second, the three types of parameters are combined as an index key; then, the pre-set calibration library is searched for a completely matched diameter calibration model parameter record; finally, the regression equation coefficient term and the constant term parameter set in the record are obtained.

[0130] Step b2, according to the pixel area value of the boundary region, in combination with the welding position compensation factor, performs diameter inversion operation by using a pre-established diameter calibration model to obtain the nugget diameter of the cross-reinforcement welding point, the welding position compensation factor being dynamically calculated by the electrode offset.

[0131] The welding position compensation factor is used to correct the proportion coefficient of the nugget deformation error caused by the electrode position deviation, which is calculated by the real-time electrode offset divided by the reference distance reference value . The electrode offset represents the spatial distance deviation between the actual position of the welding electrode and the theoretical center point, which is dynamically measured by a displacement sensor. An example expression of the pre-established diameter calibration model is: wherein, is the physical size of the nugget diameter, is the pixel area value of the boundary region, is the welding position compensation factor, is the proportion coefficient, is the correction intercept.

[0132] In the embodiment of the present application, first, the planar distance between the electrode center point and the theoretical welding position is measured in real time by the electrode displacement sensor as the electrode offset; second, the electrode offset value is divided by the preset reference distance reference value to obtain a welding position compensation factor; then the pixel area value of the boundary region is multiplied by the compensation factor to obtain the effective pixel area after geometric correction; finally, the effective pixel area is input into the pre-established diameter calibration model to perform inversion calculation, and the actual physical size value of the nugget diameter is output.

[0133] The following is a specific example: first, according to the steel bar diameter and welding current parameters, the corresponding model parameters are matched in the calibration library. Second, the electrode offset is calculated to calculate the position compensation factor, and the boundary pixel area value is multiplied by the factor to obtain the corrected area. Finally, the nugget diameter value is output by the calibration model inversion operation.

[0134] By performing steps b1~b2, the embodiment of the present application solves the model generalization problem of multi-specification mesh production through the self-adaptive calibration parameter matching mechanism of process parameters; and introduces dynamic position compensation to correct the spatial distortion of thermal images, thereby improving the inversion accuracy of the nugget diameter.

[0135] In one possible embodiment, step 121, pre-processing the thermal radiation image to obtain a processed thermal radiation image, includes:

[0136] Step c1, separating the thermal signals of different frequency bands in the thermal radiation image to remove environmental interference and extract the nugget edge information, to obtain a filtered thermal radiation image.

[0137] Wherein, the thermal signal refers to the infrared radiation energy data in the thermal radiation image representing the temperature distribution of the object surface. The environmental interference refers to the non-actual temperature signal caused by the non-target heat source. The nugget edge information refers to the thermal distribution mutation feature reflecting the transition boundary between the welding point melting zone and the base material. The filtering refers to the data processing operation of suppressing noise and enhancing target signal through frequency domain or spatial domain transformation.

[0138] In the embodiment of the present application, first, the thermal radiation image is subjected to frequency domain wavelet transform to separate the thermal signal components of different frequency bands; second, a band-stop filter is designed to suppress the feature frequency band of environmental interference; then, the high-frequency component containing the nugget edge information is extracted for signal enhancement; finally, the filtered thermal radiation image is obtained by inverse transform reconstruction, which removes interference and enhances edge features.

[0139] Step c2, based on the background temperature value of the non-welding area around the welding point in the filtered thermal radiation image, a sensitivity correction value is calculated to eliminate the radiation error caused by the oxidation of the steel bar surface, to obtain a thermal radiation image after radiation error elimination.

[0140] Wherein, the non-welding area refers to the steel surface reference area around the welding point which is not affected by heat. The background temperature value refers to the pixel temperature statistical value of the non-welding area, which is used for radiation error correction reference. The sensitivity correction value refers to the temperature calibration coefficient calculated based on the difference of material surface radiation characteristics. The radiation error refers to the temperature measurement deviation caused by the deviation of infrared emissivity from the theoretical value due to material surface oxidation.

[0141] In the embodiments of the present application, first, the non-welding area around the welding point is identified in the filtered image; second, the average value of the temperature of all pixels in the area is calculated as the background temperature value; then, the background temperature value is compared with the standard blackbody radiation temperature, and the ratio of the two is calculated as the sensitivity correction value; finally, the original temperature value of each pixel in the whole image is divided by the correction value, eliminating the radiation error caused by surface oxidation, and generating the thermal radiation image after radiation error elimination.

[0142] Step c3, for the welding point core area in the thermal radiation image after radiation error elimination, adaptive histogram equalization is performed to obtain the processed thermal radiation image.

[0143] Wherein, the histogram equalization refers to a nonlinear transformation method for enhancing image contrast by redistributing pixel gray values. The welding point core area refers to the projection range of the nugget and the adjacent heat affected zone on the thermal radiation image formed during the resistance spot welding process. In the thermal radiation image after radiation error elimination, the highest temperature pixel point coordinates are located by global scanning as the center of the nugget. This point maintains high temperature characteristics in a very short time after welding due to the slow heat dissipation rate of the nugget area; then, a circular analysis area is defined with the center point as the center and half of the current steel diameter as the radius, which covers the complete physical range of the nugget and the heat affected zone, forming the welding point core area.

[0144] In the embodiments of the present application, first, in the image after radiation error elimination, a circular welding point core area is defined with the highest temperature pixel point as the center and the steel diameter value as the diameter; second, a pixel sliding window of a preset size is used to traverse the core area, and the probability distribution function of the temperature value in each window is calculated; then, a piecewise linear mapping relationship from temperature value to equalized gray value is established; next, all pixel values in the window are converted according to the mapping table; finally, the overlapping window area is fused by bilinear interpolation, and the converted result is executed by logarithmic dynamic range compression, and the processed thermal radiation image is output.

[0145] The following is a specific example: first, the original thermal image is band-separated, the nugget edge features are extracted, and the environmental noise is filtered out. Second, the background temperature mean value is calculated in the non-welding area around the welding point in the filtered image, and the sensitivity correction value is derived to eliminate the radiation error. Finally, sliding window histogram equalization is implemented in the welding point core area, and the optimized thermal image is output through dynamic range compression.

[0146] By performing steps c1-c3, the embodiment of the present application effectively separates the target thermal signal from the environmental noise through a multi-stage preprocessing mechanism; achieves accurate compensation of radiation errors based on the background temperature of the non-welding area; and provides a high signal-to-noise ratio thermal image basis for the nugget boundary identification in combination with the adaptive contrast enhancement technology.

[0147] In a possible embodiment, step c3, for the welding point core area in the thermal radiation image after the radiation error elimination, adaptive histogram equalization is performed to obtain a processed thermal radiation image, comprising:

[0148] Step c31, taking the highest temperature point in the thermal radiation image as the center, a circular area with a diameter of the steel bar diameter is defined as the welding point core area.

[0149] Wherein, the highest temperature point refers to the pixel position with the maximum radiation intensity in the thermal radiation image, reflecting the welding pool center point. The circular area is a geometric range with a specified point as the center and a fixed length as the diameter, used to define the core analysis area. The welding point core area refers to the key analysis range containing the main body of the nugget and the heat affected zone, and its diameter is related to the steel bar size.

[0150] In the embodiment of the present application, first, all pixels of the thermal radiation image are traversed, and the single pixel point with the maximum temperature value is identified as the highest temperature point; second, a circular analysis area is defined in the image space with the point coordinates as the center and the physical size value of the current steel bar diameter as the diameter; finally, the circular area is defined as the welding point core area.

[0151] Step c32, in the welding point core area, a preset pixel sliding window is used to traverse all pixels to calculate the probability distribution function of the temperature values in the window.

[0152] Wherein, the pixel sliding window refers to a rectangular analysis unit moving on the image, used to perform local statistical operation. The probability distribution function refers to a statistical model describing the frequency of temperature values appearing in the window, which is the calculation basis of histogram equalization.

[0153] In the embodiment of the present application, first, a rectangular pixel sliding window with a fixed length and width is set; second, the window is moved in the order of row first and column second from the top left corner of the welding point core area until it covers all pixels; then at each window position, the distribution histogram of all pixel temperature values in the window is counted; finally, the distribution function describing the probability density of temperature is generated based on the histogram.

[0154] Step c33, a mapping relationship from the temperature value to the equalized gray value is established according to the probability distribution function.

[0155] Wherein, the mapping relationship refers to the conversion rule table between the original temperature value and the equalized gray value, which is established based on the probability distribution characteristics.

[0156] In the embodiment of the present application, first, the cumulative distribution value of the probability distribution function is calculated; second, the cumulative distribution value is normalized to the target gray scale range; and then the piecewise linear mapping function table between the original temperature input value and the equalized gray scale output value is constructed.

[0157] Step c34, transforming the temperature values of all pixels in the window using the mapping relationship.

[0158] In the embodiment of the present application, first, the original temperature values of all pixels in the current coverage position of the sliding window are read; second, the target gray scale value corresponding to each temperature value is queried in the mapping function table; and finally, the original temperature data is replaced by the mapped gray scale value data.

[0159] Step c35, adopting bilinear interpolation to fuse the transformation results of adjacent windows, and performing dynamic range compression processing on the transformed temperature values in the core area of the welding spot to suppress the over-enhanced area, so as to obtain the processed thermal radiation image.

[0160] The bilinear interpolation refers to a smoothing fusion algorithm for calculating new pixel values in the overlapping area by using the weighted values of the adjacent four pixels. The dynamic range compression processing is an operation of suppressing the contrast enhancement amplitude of the high gray scale value area through nonlinear transformation. The over-enhanced area refers to an image distortion area in which the local contrast is abnormally enlarged due to histogram equalization.

[0161] In the embodiment of the present application, first, for the pixels in the overlapping area of adjacent sliding windows, the bilinear weight calculation is performed according to the distances of the pixels to the centers of the four adjacent windows, so as to fuse the transformation results of different windows; second, the logarithmic function transformation is performed on the full-scale gray scale image after the transformation of the core area, so as to compress the dynamic range of the high gray scale value area; and finally, the processed thermal radiation image in which the over-enhanced phenomenon is suppressed is output.

[0162] The following is a specific example: first, the highest temperature point of the thermal image is located, and a circular core area with the diameter of the steel bar is demarcated. Second, the sliding window is used to traverse the core area, the temperature probability distribution of each window is counted, and the gray scale mapping relationship is established. Then, the pixel values in the window are converted according to the mapping table. Then, the bilinear interpolation fusion is performed on the overlapping area. Finally, the dynamic range compression processing is implemented, and the optimized thermal radiation image is output for boundary recognition.

[0163] By performing steps c1 to c3, the embodiment of the present application enhances the recognizability of the micro-temperature difference features in the nugget area through adaptive equalization; eliminates the window boundary mutation by using bilinear interpolation; and suppresses the over-enhanced distortion by combining dynamic range compression, so as to provide high-fidelity temperature distribution data for the nugget boundary recognition.

[0164] Figure 2A structure diagram of a steel mesh multi-welding point quality detection system based on infrared thermal imaging provided by an embodiment of the present application is shown in FIG. 1, and the system includes: Figure 2

[0165] A collection module 21 is configured to synchronously collect thermal radiation images of surface temperatures of all cross steel welding points in a time interval from the end of the current resistance welding process to the start of the self-net drawing action by using an infrared thermal imager array, the cross steel welding points being welding points obtained by the current resistance welding process in the steel mesh.

[0166] An identification module 22 is configured to identify the fusion form of each cross steel welding point based on the thermal radiation images.

[0167] A generation module 23 is configured to generate a quality detection result of the steel mesh multi-welding points based on the fusion forms of all cross steel welding points, the quality detection result including a geometric judgment result for representing whether the fusion form meets a preset form requirement.

[0168] A warning module 24 is configured to perform adaptive warning by using a corresponding warning mode according to the quality detection result, so as to adjust welding parameters used in the next resistance welding process according to the warning information.

[0169] Figure 2 The steel mesh multi-welding point quality detection system based on infrared thermal imaging can perform Figure 1 The steel mesh multi-welding point quality detection method based on infrared thermal imaging described in the embodiment shown in FIG. 2 will not be described in detail again. The specific manner in which each module and unit in the steel mesh multi-welding point quality detection system based on infrared thermal imaging described in the above embodiment performs an operation has been described in detail in the embodiment related to the method, and will not be described in detail here.

[0170] In one possible design, Figure 2 The steel mesh multi-welding point quality detection system based on infrared thermal imaging described in the embodiment shown in FIG. 3 can be implemented as a computing device, as shown in FIG. 4, which can include a storage component 31 and a processing component 32. Figure 3

[0171] The storage component 31 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 32.

[0172] ​​The processing component 32 is configured to: by controlling the flight height and heading angle of the detection unmanned aerial vehicle, the laser radar and the thermal infrared sensor deployed on the detection unmanned aerial vehicle collect three-dimensional point cloud data of the vertical section of the wildfire area and thermal radiation distribution data of the wildfire area at multiple angles respectively; combine the three-dimensional point cloud data and the thermal radiation distribution data to generate a target fire spread path; use an embedded real-time processing chip deployed in the detection unmanned aerial vehicle to process the thermal radiation distribution data in real time, identify temperature abnormal areas and remove interference areas, and generate fire point position and classification results; based on the target fire spread path, the fire point position and the classification results, send a dynamic scheduling instruction to the cluster of fire extinguishing unmanned aerial vehicles through a low-latency communication link to adjust the position, fire extinguishing agent dosage and coverage range of each fire extinguishing unmanned aerial vehicle in the cluster of fire extinguishing unmanned aerial vehicles.

[0173] The processing component 32 can include one or more processors to execute computer instructions to complete all or part of the steps in the above method. Of course, the processing component can also be one or more Application-Specific Integrated Circuits (ASIC), Digital Signal Processors (DSP), Digital Signal Process Devices (DSPD), Programmable Logic Devices (PLD), Field Programmable Gate Arrays (FPGA), controllers, microcontrollers, microprocessors or other electronic elements for executing the above method.

[0174] The storage component 31 is configured to store various types of data to support the operation of the terminal. The storage component can be implemented by any type of volatile or non-volatile storage device or their combination, such as Random Access Memory (RAM), Static Random-Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk or optical disk.

[0175] Of course, the computing device can also include other components, such as input / output interfaces, display components, communication components, etc.

[0176] The input / output interface provides an interface between the processing component and peripheral interface modules, which can be output devices, input devices, etc.

[0177] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices, etc.

[0178] The computing device can be a physical device or an elastic computing host provided by a cloud computing platform, and the computing device can be a cloud server, and the processing component, the storage component, etc. can be a basic server resource rented or purchased from the cloud computing platform.

[0179] The embodiment of the application also provides a computer storage medium storing a computer program, and the computer program can implement the above-mentioned Figure 1 A steel mesh multi-welding point quality detection method based on infrared thermal imaging is provided.

[0180] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device and unit can refer to the corresponding process in the foregoing method embodiments, and will not be described here.

[0181] The device embodiments described above are only schematic, and the units described as separate components can or can not be physically separated, and the components displayed as units can or can not be physical units, that is, they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected to achieve the purpose of the embodiment scheme according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0182] Through the description of the foregoing embodiments, those skilled in the art can clearly understand that each embodiment can be realized by means of software and a necessary general hardware platform, and of course, it can also be realized by hardware. Based on such understanding, the above technical solutions can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, etc., and includes a plurality of instructions to make a computer device (which can be a personal computer, a server, or a network device, etc.) execute the methods described in each embodiment or some parts of the embodiments.

[0183] Finally, it should be noted that the above examples are only used to illustrate the technical solutions of the present application, and are not intended to limit the same; although the present application has been described in detail with reference to the foregoing examples, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing examples, or make equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for detecting the quality of multiple welding points of a steel mesh based on infrared thermal imaging, characterized in that, The method comprises the following steps: During the time interval from the end of the resistance welding process to the start of the self-net-drawing action, the infrared thermal imager array synchronously collects thermal radiation images of the surface temperatures of all cross-reinforcement welds obtained by the resistance welding process in the steel reinforcement mesh; For each cross-reinforcement weld, the fusion form of the cross-reinforcement weld is identified based on the thermal radiation images; Based on the fusion forms of all cross-reinforcement welds, a quality detection result of the steel reinforcement mesh multi-weld is generated, and the quality detection result comprises a geometric judgment result for characterizing whether the fusion form meets a preset form requirement; According to the quality detection result, a corresponding early warning mode is used for adaptive early warning to adjust the welding parameters used in the next resistance welding process according to the early warning information; The fusion form comprises a pixel area value and a fusion core diameter; The fusion form of the cross-reinforcement weld is identified based on the thermal radiation images, which comprises the following steps: The thermal radiation images are preprocessed to obtain processed thermal radiation images; The boundary region of the cross-reinforcement weld is identified from the processed thermal radiation images, and the pixel area value of the boundary region is calculated; According to the pixel area value of the boundary region, diameter inversion operation is performed by using a pre-established diameter calibration model to obtain the fusion core diameter of the cross-reinforcement weld, and the parameters of the diameter calibration model are calibrated according to the reinforcement diameter and the welding parameters; According to the pixel area value of the boundary region, diameter inversion operation is performed by using a pre-established diameter calibration model to obtain the fusion core diameter of the cross-reinforcement weld, and the parameters of the diameter calibration model are calibrated according to the reinforcement diameter and the welding parameters, which comprises the following steps: According to the reinforcement diameter and the welding parameters, the parameters of the diameter calibration model are matched from a preset calibration library; According to the pixel area value of the boundary region, diameter inversion operation is performed by using a pre-established diameter calibration model to obtain the fusion core diameter of the cross-reinforcement weld, and the welding position compensation factor is dynamically calculated by the electrode offset.

2. The method of claim 1, wherein, The boundary region of the cross-reinforcement weld is identified from the processed thermal radiation images, which comprises the following steps: By a temperature threshold segmentation method, all continuous pixel regions with a temperature value greater than a preset temperature threshold in the processed thermal radiation images are identified; According to the continuous pixel regions, a topological relationship graph characterizing pixel connectivity is established, boundary node sets in the topological relationship graph are converted into a binary image matrix by coordinate mapping, and an initial boundary mask is generated; The initial boundary mask is subjected to a hole filling process to obtain a filled boundary mask; The filled boundary mask is subjected to an edge smoothing process to obtain a smoothed boundary mask; Based on the smoothed boundary mask, the temperature gradient values of the pixels corresponding to the boundary nodes are calculated, and the pixels with a temperature gradient value greater than or equal to a preset temperature gradient threshold value are screened to determine the boundary region of the cross-reinforcement weld.

3. The method of claim 2, wherein, Based on the smoothed boundary mask, the temperature gradient values of the pixels corresponding to the boundary nodes are calculated, and the pixels with a temperature gradient value greater than or equal to a preset temperature gradient threshold value are screened to determine the boundary region of the cross-reinforcement weld. Based on the smoothed boundary mask, the surface temperature of the boundary node in the thermal radiation image is obtained, and combined with the surface temperature of the cross steel bar welding point in another thermal radiation image, the temperature gradient value of the pixel corresponding to the boundary node is calculated along the boundary normal direction, wherein the boundary normal direction is determined according to the coordinates of the adjacent nodes of the topological relationship diagram; Obtain the temperature gradient threshold corresponding to the steel bar diameter from the preset configuration table; Screen the pixels with temperature gradient values greater than or equal to the preset temperature gradient threshold to determine the boundary area of the cross steel bar welding point.

4. The method of claim 1, wherein, The pre-processing of the thermal radiation image to obtain a processed thermal radiation image includes: Separate the thermal signals of different frequency bands in the thermal radiation image to remove environmental interference and extract the nugget edge information to obtain a filtered thermal radiation image; Based on the background temperature value of the non-welding area around the welding point in the filtered thermal radiation image, a sensitivity correction value is calculated to eliminate the radiation error caused by the oxidation of the steel bar surface to obtain a thermal radiation image after radiation error elimination; For the welding point core area in the thermal radiation image after radiation error elimination, adaptive histogram equalization is performed to obtain a processed thermal radiation image.

5. The method of claim 4, wherein, The adaptive histogram equalization for the welding point core area in the thermal radiation image after radiation error elimination to obtain a processed thermal radiation image includes: Centering on the highest temperature point in the thermal radiation image, a circular area with a diameter of the steel bar diameter is defined as the welding point core area; In the welding point core area, a preset pixel sliding window is used to traverse all pixels to calculate the probability distribution function of the temperature values in the window; According to the probability distribution function, a mapping relationship from the temperature value to the equalized gray value is established; The temperature values of all pixels in the window are transformed using the mapping relationship; The transformed results of adjacent windows are fused using bilinear interpolation, and the transformed temperature values of the welding point core area are dynamically range compressed to suppress the over-enhanced area to obtain a processed thermal radiation image.

6. A multi-welding point quality detection system for steel mesh based on infrared thermal imaging, characterized in that, It includes: The acquisition module is configured to acquire, through an infrared thermal imager array, thermal radiation images of the surface temperature of all cross steel bar welding points within a time interval from the end of the current resistance welding process to the start of the self-net drawing action; The identification module is configured to identify, for each cross steel bar welding point, a fusion form of the cross steel bar welding point based on the thermal radiation image; The generation module is configured to generate a quality detection result of the multi-welding point of the steel bar mesh based on the fusion forms of all cross steel bar welding points, the quality detection result including a geometric judgment result for indicating whether the fusion form meets a preset form requirement; The early warning module is configured to adaptively perform early warning in a corresponding early warning mode according to the quality detection result to adjust welding parameters for the next resistance welding process according to early warning information; The fusion form includes a pixel area value and a nugget diameter. The identification of the fusion form of the cross steel bar welding point based on the thermal radiation image includes: Pre-processing the thermal radiation image to obtain a processed thermal radiation image; Identifying a boundary area of the cross-reinforcement welding spot from the processed thermal radiation image, and calculating a pixel area value of the boundary area; According to the pixel area value of the boundary area, a diameter inversion operation is performed by using a pre-established diameter calibration model to obtain a nugget diameter of the cross-reinforcement welding spot, and parameters of the diameter calibration model are calibrated according to a reinforcement diameter and welding parameters; The diameter inversion operation is performed by using the pre-established diameter calibration model according to the pixel area value of the boundary area to obtain the nugget diameter of the cross-reinforcement welding spot, and the parameters of the diameter calibration model are calibrated according to the reinforcement diameter and the welding parameters, and the diameter calibration model comprises: Parameters of the diameter calibration model are matched from a pre-set calibration library according to the reinforcement diameter and the welding parameters; The diameter inversion operation is performed by using the pre-established diameter calibration model according to the pixel area value of the boundary area and in combination with a welding position compensation factor to obtain the nugget diameter of the cross-reinforcement welding spot, and the welding position compensation factor is dynamically calculated by an electrode offset amount.

Images