An automated door and window quality inspection method and system based on machine vision

Abstract

The application provides an automatic door and window quality inspection method and system based on machine vision. In the process of pressure increase, a thermal image sequence and pressure data are synchronously collected, a heat conduction baseline is established in the low pressure stage, and a heat conduction mutation caused by deformation of a sealing structure is captured in the high pressure stage. Based on the geometric characteristics of the door and window, the thermal image data are dynamically registered to generate a displacement vector field, and an abnormal heat flow vortex and a discontinuous boundary are identified through a space-time feature extraction model. In the pressure decay stage, a dynamic data compensation mechanism is started to generate a tracking atlas, and finally, the phase correlation of the abnormal area is analyzed in combination with a three-dimensional seepage model, the air leakage path parameters are output, compared with a sealing failure threshold, and a partitioned sealing performance grading evaluation result is generated. The technical scheme provided by the application realizes precise positioning, three-dimensional seepage path reconstruction and quantitative leakage evaluation of non-contact door and window sealing defects through thermodynamic dynamic modeling and multi-source data fusion, and significantly improves the quality inspection accuracy and efficiency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of machine vision, in particular to an automatic door and window quality inspection method and system based on machine vision. BACKGROUND

[0002] With the improvement of building energy-saving standards, door and window air tightness detection has become a core link of building quality acceptance. The traditional method relies on static pressure test and manual visual inspection, which is difficult to accurately quantify the small leakage defects caused by the deformation of the sealing structure during the dynamic pressure process, especially the capture demand of local heat conduction anomalies caused by transient deformation at high pressure stage, low flow rate air leakage path, and an automatic detection technology with non-contact and high space-time resolution is urgently needed to synchronize the pressure load change and thermodynamic response characteristics, realize defect positioning and dynamic modeling of leakage parameters.

[0003] The current mainstream scheme adopts an infrared thermal imaging based static pressure detection technology, which acquires the temperature field distribution of the door and window surface under a fixed pressure, and identifies the temperature difference abnormal area by combining an image segmentation algorithm. Specifically, the system acquires a single thermal image under a preset pressure value, extracts the sealing strip contour by using an edge detection algorithm, and screens suspected heat leakage points by using a thermal gradient threshold, and finally estimates the leakage equivalent by combining a pressure decay curve. Some improved schemes introduce time series thermal image superposition analysis to improve the sensitivity to small temperature fluctuations.

[0004] This scheme relies on static pressure conditions and cannot capture the transient deformation characteristics of the sealing structure under dynamic pressure changes, resulting in the local heat conduction mutation caused by elastic deformation at the high pressure stage being ignored. In addition, single frame or low sampling rate thermal image data is difficult to analyze the heat flow continuity characteristics (such as vortex, boundary fracture), and is easy to cause the loss of transient abnormal signals due to insufficient time resolution. At the same time, the static registration method does not consider the frame displacement caused by pressure changes, which causes the spatial alignment error of the thermal image sequence to accumulate, further affecting the reconstruction accuracy of the leakage path. In the pressure mutation scene, the existing scheme lacks a dynamic data compensation mechanism, and it is difficult to synchronize the tracking of the thermodynamic response and the pressure phase relationship, which finally leads to the calculation deviation of the leakage equivalent and the defect misjudgment. SUMMARY

[0005] The present application provides an automatic door and window quality inspection method and system based on machine vision to solve the error problem caused by the transient deformation of the sealing structure under dynamic pressure changes in the prior art.

[0006] In a first aspect, the present application provides an automatic door and window quality inspection method based on machine vision, comprising:

[0007] Synchronously acquire thermal image sequence data and pressure data in a door and window pressurization process, collect an initial distribution of a temperature field in a low-pressure stage of the pressurization process and use the initial distribution to establish a heat conduction baseline, and capture a heat conduction mutation feature caused by deformation of a sealing structure in a high-pressure stage of the pressurization process;

[0008] Perform heat flow correction processing on the thermal image sequence data, establish a dynamic registration relationship based on geometric features of a door and window frame, the heat conduction baseline, and the heat conduction mutation feature, and generate a displacement vector field with spatial constraints according to the dynamic registration relationship;

[0009] Input the displacement vector field into a preset space-time feature extraction model, extract a heat flow continuity feature of a sealing edge region through a multi-layer feature fusion structure, and identify an abnormal heat flow vortex and a discontinuous boundary;

[0010] Trigger a dynamic data compensation mechanism in a pressure decay process, enhance a thermal image data sampling frequency when a pressure mutation is detected, and generate a dynamic tracking atlas;

[0011] According to a topological structure of the abnormal heat flow vortex, an abnormal region divided by the discontinuous boundary, and a phase correlation of the pressure data, in combination with a response event of the dynamic tracking atlas, a three-dimensional seepage model of a sealing defect is established, and geometric parameters of a wind leakage path and a leakage equivalent are output according to the three-dimensional seepage model;

[0012] Based on the leakage equivalent, the geometric parameters, and a preset sealing failure threshold curve, a hierarchical evaluation result associated with a door and window structure partition is generated.

[0013] Optionally, according to a topological structure of the abnormal heat flow vortex, an abnormal region divided by the discontinuous boundary, and a phase correlation of the pressure data, in combination with a response event of the dynamic tracking atlas, a three-dimensional seepage model of a sealing defect is established, including:

[0014] The topological structure of the abnormal heat flow vortex is decomposed into a plurality of annular closed paths, and based on spatial geometric constraints of the abnormal region divided by the discontinuous boundary, a center point of each closed path and a minimum distance from an edge of the abnormal region are calculated;

[0015] According to the phase correlation of the pressure data, a pressure decay process is divided into a linear decay stage and a nonlinear mutation stage, and a time domain interval in which a pressure change rate exceeds a preset value in the nonlinear mutation stage is extracted;

[0016] The response event of the dynamic tracking atlas is matched with the time domain interval, a set of image frames in which the heat flow direction in the thermal image sequence has a reverse mutation in the time domain interval is marked, a three-dimensional grid of the permeability distribution of the porous medium is constructed based on the spatial superposition relationship of the center point distribution density of the ring-shaped closed path, the minimum interval and the set of image frames;

[0017] The heat flow vector angle difference between adjacent grid cells in the three-dimensional grid is calculated, when the angle difference exceeds a structural deformation threshold, a seepage channel connecting the ring-shaped closed path and the joint of the door and window frame is generated, and finally a three-dimensional seepage model of the sealing defect is established.

[0018] Optionally, based on the leakage equivalent, the geometric parameters and a preset sealing failure threshold curve, a hierarchical evaluation result associated with the door and window structure partition is generated, including:

[0019] The seepage channel extension length in the geometric parameters is mapped to the physical division line of the structural partition of the door and window frame, and the number of intersection points of the seepage channel and the frame joint in each partition is calculated;

[0020] According to the projection position of the leakage equivalent on the sealing failure threshold curve, the air tightness level attenuation coefficient corresponding to each partition is determined;

[0021] When the number of intersection points in the same partition exceeds a structural strength threshold and the attenuation coefficient simultaneously exceeds a critical value, the partition is marked as a high-risk area of sealing failure;

[0022] According to the consistency of the extension direction of the seepage channel of the adjacent partition of the high-risk area, a hierarchical evaluation result containing a priority repair path indication associated with the door and window structure partition is generated.

[0023] Optionally, a dynamic registration relationship is established based on the geometric features of the door and window frame, the heat conduction baseline and the heat conduction mutation feature, and a displacement vector field with spatial constraints is generated according to the dynamic registration relationship, including:

[0024] The corner points and the joint lines in the geometric features of the door and window frame are extracted in the heat flow stable distribution mode of the heat conduction baseline to generate a reference heat flow direction constraint template;

[0025] The deformation area of the frame is identified in the heat conduction mutation feature, and the heat flow direction deviation of the deformation area edge and the matching error of the reference heat flow direction constraint template are calculated;

[0026] When the matching error exceeds a deformation compensation threshold, the vector weight of the reference heat flow direction constraint template is adjusted according to the gradient distribution of the heat conduction mutation feature;

[0027] Based on the adjusted vector weight and the geometric topological relationship of the door and window frame joint, a proportional mapping function of heat flow displacement and physical deformation is constructed;

[0028] The heat flow vector difference of adjacent frames in the thermal image sequence data is converted into a spatial displacement with millimeter-level precision through the proportional mapping function, and a displacement vector field with spatial constraints is generated.

[0029] Optionally, the displacement vector field is input into a preset spatiotemporal feature extraction model, and the heat flow continuity features of the sealing edge region are extracted through a multi-layer feature fusion structure to identify abnormal heat flow vortexes and discontinuous boundaries, including:

[0030] The displacement vector field is input into a preset spatiotemporal feature extraction model, and the vector direction consistency of the strip-shaped region on both sides of the sealing strip edge is analyzed through the multi-layer feature fusion structure in the model;

[0031] The detection unit is divided at equal intervals in the strip-shaped region, and the dynamic smoothness of the direction included angle of the continuous three frames of displacement vectors in each detection unit is calculated, and when the smoothness is lower than the continuity threshold, it is marked as a heat flow fracture point;

[0032] Based on the spatial distribution density of the heat flow fracture points, a discontinuous boundary along the extension direction of the sealing strip is generated, and the geometric accuracy of the boundary contour is corrected according to the fracture point spacing change rate;

[0033] The annular closed path of the displacement vector is detected outside the strip-shaped region, and when two or more non-concentric annular paths are detected and their rotational directions are opposite to the pressure decay direction, they are marked as abnormal heat flow vortexes;

[0034] According to the consistency of the modified contour of the discontinuous boundary and the rotational direction of the abnormal heat flow vortex, the topological structure containing the preferential direction of air leakage diffusion is output.

[0035] Optionally, a dynamic data compensation mechanism is triggered during pressure decay, and when a pressure mutation is detected, the thermal image data sampling frequency is increased and a dynamic tracking map is generated, including:

[0036] The first derivative of the real-time monitored pressure decay rate is monitored, and when the absolute value of the derivative exceeds the pressure mutation threshold, the fast sampling mode of the thermal image acquisition device is started;

[0037] In the fast sampling mode, the dynamic data compensation mechanism is triggered, the thermal image frame rate is increased to a preset sampling rate, and the microsecond-level response delay of the pressure sensor is recorded synchronously;

[0038] According to the difference between the pressure mutation threshold duration and the response delay, the increase rate of the thermal image frame rate is dynamically adjusted;

[0039] After the fast sampling mode ends, the thermal image data higher than the standard sampling rate is time-stamped and interpolated to align, generating a tracking atlas containing sub-frame level thermal flow dynamics.

[0040] Optionally, the initial temperature field distribution is collected during the low pressure stage of the pressurization process and used to establish a thermal conduction baseline, and the thermal conduction mutation characteristics triggered by the deformation of the sealing structure during the high pressure stage of the pressurization process are captured, including:

[0041] During the low pressure stage of the pressurization process, the pressure is maintained at a preset interval lower than the standard test pressure, the initial temperature field distribution is collected, and a thermal conduction baseline containing the steady-state thermal flow diffusion rate at the sealing joint is established;

[0042] When the pressure during the pressurization process switches to the high pressure stage, the thermal conduction mutation characteristics triggered by the deformation of the sealing structure are captured, including the detection of a significant change event in the heat flow rate at the joint and the synchronous reversal of the heat flow direction.

[0043] According to the spatiotemporal correlation between the thermal conduction baseline and the thermal conduction mutation characteristics, a deformation compensation parameter containing the deformation hysteresis effect of the frame material is established; during the high pressure stage maintenance process, when the phase difference between the thermal conduction mutation characteristics and the pressure change exceeds the dynamic tolerance threshold, the iterative update of the deformation compensation parameter is triggered.

[0044] In a second aspect, the application provides an automatic door and window quality inspection system based on machine vision, comprising:

[0045] The acquisition module is used for synchronously acquiring thermal image sequence data and pressure data during the pressurization process of the door and window, and the initial temperature field distribution collected during the low pressure stage of the pressurization process is used to establish a thermal conduction baseline, and the thermal conduction mutation characteristics triggered by the deformation of the sealing structure during the high pressure stage of the pressurization process are captured.

[0046] The generation module is used for performing heat flow correction processing on the thermal image sequence data, establishing a dynamic registration relationship based on the geometric characteristics of the door and window frame, the thermal conduction baseline, and the thermal conduction mutation characteristics, and generating a displacement vector field with spatial constraints.

[0047] The recognition module is used for inputting the displacement vector field into a spatiotemporal feature extraction model, extracting thermal flow continuity features of the sealing edge region through a multi-layer feature fusion structure, and recognizing abnormal thermal flow vortices and discontinuous boundaries.

[0048] The compensation module is used for triggering a dynamic data compensation mechanism during the pressure decay process, and when a pressure mutation is detected, the thermal image data sampling frequency is enhanced and a dynamic tracking atlas is generated.

[0049] an output module configured to establish a three-dimensional seepage model of the sealing defect according to the topological structure of the abnormal heat flow vortex, the phase correlation between the abnormal region divided by the discontinuous boundary and the pressure data, and the response event of the dynamic tracking atlas, and output geometric parameters of the air leakage path and a leakage equivalent according to the three-dimensional seepage model;

[0050] The generation module is further configured to generate a hierarchical evaluation result associated with a partition of the door and window structure based on the leakage equivalent, the geometric parameters, and a preset sealing failure threshold curve.

[0051] In a third aspect, an embodiment of the present application provides a computing device, including 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, so as to realize the automatic door and window quality inspection method based on machine vision as described in the first aspect.

[0052] In a fourth aspect, an embodiment of the present application provides a computer storage medium, which stores a computer program; when the computer program is executed by a computer, a method for automatic door and window quality inspection based on machine vision as described in the first aspect is realized.

[0053] In the embodiment of the present application, the thermal image sequence and the pressure data are synchronously acquired during the door and window pressurization process, the heat conduction baseline is established in the low-pressure stage, and the heat conduction mutation characteristics are captured in the high-pressure stage, so that the pressure deformation and the thermodynamic response can be dynamically associated, and the sealing structure elastic deformation and the leakage defect characteristics can be accurately distinguished; the displacement vector field is generated through the heat flow correction and the dynamic registration, so that the image space deviation caused by the displacement of the door and window frame can be eliminated, and the dynamic mapping relationship between the heat flow characteristics and the geometric structure can be established; the abnormal heat flow vortex and the discontinuous boundary are identified by using the space-time feature extraction model, so that the limitation of single-frame thermal image analysis is broken, and the sensitivity to the continuous abnormality such as the micro heat flow fracture and the vortex is enhanced; the high-frequency sampling is triggered and the tracking atlas is generated through the dynamic data compensation mechanism, so that the transient heat flow characteristics during the pressure mutation can be completely captured; the topological structure of the abnormal region and the pressure phase correlation are analyzed based on the three-dimensional seepage model, so that the three-dimensional reconstruction of the air leakage path and the physical quantization of the leakage equivalent are realized; finally, the partition evaluation of the sealing failure threshold curve is combined, so that the weak leakage area of the door and window structure can be accurately positioned, the interpretable defect hierarchical conclusion can be formed, and the detection efficiency and the engineering applicability of the defect modeling can be significantly improved.

[0054] Furthermore, the construction of the three-dimensional seepage model specifically includes: decomposing the abnormal heat flow vortex into a closed loop and calculating its distance from the defect edge; dividing the time domain interval by combining the nonlinear pressure change stage; matching image frames of reverse heat flow change in the dynamic tracking spectrum; generating a three-dimensional mesh of porous media permeability based on the center point density, spacing, and image frame superposition relationship; and finally determining the seepage channel by the difference in the angle between the heat flow vectors. By decoupling the loop path topology and matching the pressure-heat flow time domain, the limitations of traditional two-dimensional thermal gradient analysis are overcome, and the three-dimensional expansion path of the seepage channel in the porous medium is accurately quantified. At the same time, combined with the deformation threshold dynamic correction model, the reconstruction accuracy and spatial positioning reliability of small tortuous leakage defects are significantly improved.

[0055] These or other aspects of this application will become more apparent in the following description of the embodiments. Attached Figure Description

[0056] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0057] Figure 1 A flowchart of an automated door and window quality inspection method based on machine vision provided in this application is shown;

[0058] Figure 2 The illustration shows a scenario of an automated door and window quality inspection method based on machine vision provided in this application.

[0059] Figure 3 A schematic diagram of the structure of an automated door and window quality inspection system based on machine vision provided in this application is shown.

[0060] Figure 4 A schematic diagram of the structure of a computing device provided in this application is shown. Detailed Implementation

[0061] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings.

[0062] In some of the flowcharts described in the specification and claims of this application and in the above-described figures, there are multiple operations contained in each description which are in a particular order when performed. However, it should be clear that these operations can be performed in an order other than that in which they appear or in parallel, and the serial numbers of the operations, such as 101, 102, etc., are only used to distinguish different operations, and the serial numbers themselves do not represent any execution order. In addition, these flowcharts 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 this paper are used to distinguish different messages, devices, modules, etc., and do not represent the order and do not limit that "first" and "second" are different types.

[0063] In the current door and window air tightness detection technology, the static pressurization scheme based on infrared thermal imaging has a significant bottleneck: it relies on single-frame thermal image analysis under fixed pressure, and cannot capture the transient deformation characteristics of the sealing structure during dynamic pressurization, resulting in the omission of local thermal conduction mutations caused by elastic deformation at high pressure. At the same time, low sampling rate thermal image data is difficult to analyze thermal flow continuity anomalies (such as vortex, boundary fracture), and the static registration method ignores the frame displacement caused by pressure changes, causing spatial alignment errors of the thermal image sequence to accumulate, further affecting the reconstruction accuracy of the leakage path. In addition, the existing scheme lacks a dynamic response mechanism for pressure mutations, and cannot synchronously track the phase relationship between thermodynamic characteristics and pressure decay, ultimately leading to leakage equivalent calculation bias and defect misjudgment.

[0064] To overcome the above-mentioned defects, the present application proposes a door and window sealing defect detection method based on dynamic thermal-force coupling modeling, the core of which is to synchronously collect thermal image sequences and pressure data during the pressurization process, capture thermal conduction mutation characteristics in the high-pressure stage, and realize the decoupling of deformation and leakage conduction modes by establishing a thermal conduction baseline in the low-pressure stage. Further, a displacement vector field is generated based on the dynamic registration of the geometric characteristics of the door and window, eliminating the thermal image offset error caused by frame displacement; through a spatiotemporal feature extraction model, thermal flow vortexes and discontinuous boundaries are identified, and a dynamic data compensation mechanism is combined to enhance the transient feature capture capability. Finally, the topological structure of the abnormal area and the pressure phase are associated using a three-dimensional seepage model to quantify the geometric parameters and equivalent of the leakage path, and the partition evaluation results are generated in combination with the sealing failure threshold curve. This scheme breaks through the limitations of static analysis, accurately reconstructs the seepage channel and eliminates registration errors through dynamic thermal-force data fusion and spatiotemporal modeling, solves the problem of transient signal loss through high-frequency sampling triggered by pressure and phase correlation algorithms, significantly improves the detection rate of small defects and the evaluation accuracy of leakage volume, and provides reliable technical support for the automatic quality inspection of door and window air tightness.

[0065] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be described clearly and completely. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments of the present application, all the other embodiments obtained by those skilled in the art without creative work fall within the scope of the present application.

[0066] Figure 1 A flow chart of an embodiment of the present application provides an automatic door and window quality inspection method based on machine vision, as shown in Figure 1 The method comprises the following steps.

[0067] 101. Synchronously acquiring thermal image sequence data and pressure data in a door and window pressurization process, collecting temperature field initial distribution in a low pressure stage of the pressurization process and establishing a heat conduction baseline, and capturing heat conduction mutation characteristics caused by sealing structure deformation in a high pressure stage of the pressurization process.

[0068] Optionally, step 101 can specifically comprise the following steps.

[0069] 1011. Maintaining pressure in a preset interval lower than a standard test pressure in the low pressure stage of the pressurization process, collecting temperature field initial distribution and establishing a heat conduction baseline containing steady-state heat flow diffusion rate at a sealing joint.

[0070] 1012. When the pressure in the pressurization process switches to the high pressure stage, capturing heat conduction mutation characteristics including a significant change event of heat flow rate at a detection joint and a synchronous reverse phenomenon of heat flow direction caused by sealing structure deformation.

[0071] 1013. According to the space-time correlation of the heat conduction baseline and the heat conduction mutation characteristics, establishing a deformation compensation parameter containing deformation hysteresis effect of frame material; when the phase difference between the heat conduction mutation characteristics and pressure change exceeds a dynamic tolerance threshold, triggering iterative update of the deformation compensation parameter in the high pressure stage maintenance process.

[0072] In the above scheme, the heat conduction baseline is a reference model established by maintaining a preset interval below the standard test pressure during the low pressure stage of the door and window pressurization, specifically referring to the heat flow diffusion rate calculated by collecting the steady-state temperature field data at the sealing joint using an infrared thermal imager and combining the Fourier heat conduction equation, which is used to represent the normal heat conduction law when there is no structural deformation; the heat conduction mutation feature is the abnormal change and direction reversal of the heat flow rate caused by the deformation of the sealing structure during the high pressure stage, for example, the heat flow direction of the sealing strip changes from outward diffusion to inward convergence after being pressed; the deformation compensation parameter is a dynamic correction coefficient generated by fitting the pressure-heat conduction phase delay based on the time and space correlation between the baseline data and the mutation feature, which is used to offset the influence of the deformation hysteresis effect of the frame material on the detection accuracy.

[0073] In the embodiment of the present application, first, the pressure control module is stabilized in a preset interval during the low pressure stage in the pressurization process through step 1011, the initial distribution data of the temperature field of the sealing area is collected at a fixed frequency using an infrared thermal imager, the steady-state heat flow diffusion rate at the sealing joint is calculated based on the Fourier heat conduction equation after random noise is eliminated using the mean filtering algorithm, and a heat conduction baseline model is generated; second, when the pressure rises to the high pressure stage, the pressure sensor data and the thermal image sequence are synchronously collected in step 1012, the heat flow rate and direction of each pixel point in the sealing area are calculated in real time using the heat flow vector field algorithm, and if it is detected that the heat flow rate of a region significantly exceeds the baseline threshold and the direction is reversed, it is marked as a heat conduction mutation feature and its time sequence position is recorded; finally, the baseline data of the heat conduction baseline and the heat conduction mutation feature are analyzed in time and space through step 1013, the phase difference between pressure change and heat conduction response is calculated, when the phase difference exceeds the preset tolerance threshold, the least squares method is used to fit the pressure-heat conduction delay curve, the deformation compensation parameter is iteratively updated to correct the phase deviation of subsequent data processing, and finally a closed-loop dynamic correction mechanism is formed.

[0074] In actual application, taking the air tightness detection of a certain aluminum alloy sliding window as an example, the pressure is stably controlled in the 50% interval of the standard test pressure during the low pressure stage, the temperature field distribution data of the sealing strip area is continuously collected by the infrared thermal imager, and the baseline model of the steady-state heat flow diffusion rate is generated based on the heat conduction equation after filtering and noise reduction; when the pressure rises to the high pressure stage, the pressure sensor data and the thermal image sequence are synchronously monitored, and the heat flow rate of the sealing strip area at the lower right corner of the window frame is detected to increase sharply and the direction changes from outward diffusion to inward convergence, which is marked as a heat conduction mutation feature; then the time sequence difference between the pressure rising moment and the heat flow mutation response is analyzed, it is found that the phase difference exceeds the tolerance threshold, the least squares method is used to fit the delay curve and update the deformation compensation parameter, so as to correct the phase deviation of the heat conduction data and the pressure change in real time during the subsequent high pressure maintenance stage.

[0075] The scheme establishes a heat conduction baseline in the low-pressure stage, effectively distinguishes environmental interference from real heat conduction characteristics, and provides a reliable reference for defect identification; the heat flow mutation characteristics are dynamically captured in the high-pressure stage, the abnormal area caused by the deformation of the sealing structure is accurately positioned, and the misjudgment of elastic deformation and leakage defects is avoided; the iteration updating mechanism of the deformation compensation parameter is combined to real-time correct the hysteresis effect of the pressure-heat conduction response, significantly improve the sensitivity and positioning accuracy of the defect detection, and ensure the data reliability of the subsequent seepage model reconstruction.

[0076] 102. performing heat flow correction processing on the thermal image sequence data, establishing a dynamic registration relationship based on the geometric features of the door and window frame, the heat conduction baseline and the heat conduction mutation characteristics, and generating a displacement vector field with spatial constraints according to the dynamic registration relationship;

[0077] Optionally, step 102 can specifically include the following steps:

[0078] 1021. extracting the heat flow stable distribution pattern of the corner points and the joint lines in the geometric features of the door and window frame in the heat conduction baseline, and generating a reference heat flow direction constraint template;

[0079] 1022. identifying the frame deformation area in the heat conduction mutation characteristics, and calculating the heat flow direction deviation of the deformation area edge and the matching error of the reference heat flow direction constraint template;

[0080] 1023. when the matching error exceeds the deformation compensation threshold, adjusting the vector weight of the reference heat flow direction constraint template according to the gradient distribution of the heat conduction mutation characteristics;

[0081] 1024. constructing a proportional mapping function of heat flow displacement and physical deformation based on the adjusted vector weight and the geometric topological relationship of the door and window frame joints;

[0082] 1025. converting the heat flow vector difference of adjacent frames in the thermal image sequence data into a spatial displacement with millimeter-level precision through the proportional mapping function, and generating a displacement vector field with spatial constraints.

[0083] In the above scheme, the reference heat flow direction constraint template refers to a direction reference template generated based on the stable heat flow distribution pattern of the corner points and the joint lines of the door and window frame in the low-pressure stage heat conduction baseline, which is used to constrain the heat flow direction deviation calculation of the high-pressure deformation area; the matching error refers to the difference value of the heat flow direction of the deformation area in the high-pressure stage and the reference template, which is quantified by the vector angle cosine similarity; the vector weight is a template correction coefficient dynamically adjusted according to the gradient distribution of the heat conduction mutation characteristics, which is used to balance the difference between the reference template and the local deformation characteristics; the proportional mapping function is a conversion function established based on the geometric topological relationship and the adjusted weight, which maps the heat flow vector difference to the physical deformation to generate the displacement field.

[0084] The embodiment of the application first extracts the stable heat flow distribution pattern of the door and window frame corner points and the joint line in the heat conduction baseline through step 1021, for example, the Harris corner point detection algorithm is used to locate the four corners of the window frame, and a reference heat flow direction constraint template is generated by combining the consistency analysis of the heat flow vector direction, which is specifically manifested as the heat flow direction of the four corners of the window frame all diffuses outward; secondly, the frame deformation area is identified in the heat conduction mutation feature in the high pressure stage through step 1022, for example, the heat flow direction of the left upper corner sealing strip area is detected to be suddenly changed to inward convergence, the heat flow direction change of the edge of the deformation area is tracked by using the optical flow method, the matching error is obtained by calculating the cosine similarity degree of the heat flow direction deviation of the edge of the deformation area and the reference template, and when the matching error is lower than the preset threshold, it is determined that there is significant deformation; then the vector weight of the reference template is adjusted according to the heat conduction gradient distribution of the deformation area through step 1023, for example, when the heat flow gradient amplitude of the left upper corner is higher, the template weight of this area is increased to 1.5 times to enhance the correction priority; then the vector weight after adjustment and the geometric topological relationship of the door and window frame joint are used to convert the heat flow vector difference into physical displacement through step 1024, for example, the joint included angle is 90 degrees, the proportional mapping function of the heat flow displacement and the physical deformation is constructed by fitting the historical deformation data by the least square method, for example, the heat flow difference 5W·m⁻² corresponds to the deformation 4mm; finally, the heat flow vector difference of adjacent frames in the thermal image sequence data is input into the proportional mapping function through step 1025, the displacement vector field with millimeter level precision is generated by superimposing the joint continuity constraint, for example, the output shows that the left upper corner frame is deformed inward by 4mm, and the displacement direction is consistent with the extension direction of the joint.

[0085] In actual application, taking the high pressure air tightness detection of the broken bridge aluminum door and window as an example, the four corners of the window frame and the joint line are located by the Harris corner point detection in the low pressure stage, and the reference template is generated by analyzing the outward diffusion heat flow direction; the heat flow direction of the left upper corner sealing strip area is detected to be suddenly changed to inward convergence in the high pressure stage, the cosine similarity degree with the reference template is calculated by tracking the deformation edge by using the optical flow method, and the matching error is significantly lower than the threshold; the template weight is increased based on the regional heat conduction gradient amplitude, and the high gradient area is preferentially corrected; the proportional mapping function is fitted by combining the right angle topological relationship of the window frame joint, the heat flow difference is converted into physical displacement; finally, the displacement vector field is generated by the joint continuity constraint, the left upper corner frame is deformed inward, the displacement direction is consistent with the extension of the joint, and the millimeter level spatial positioning of the deformation area is completed.

[0086] This scheme effectively eliminates the interference of frame geometric features on the heat flow direction by generating a reference heat flow direction constraint template and dynamically adjusting the weights, thereby improving the accuracy of deformation region identification. Based on the matching error threshold determination and gradient-driven weight optimization mechanism, it enhances the sensitivity to local deformation features and the priority of correction. Combined with the proportional mapping function of geometric topological constraints, it achieves high-precision conversion of heat flow data to physical displacement, ensuring the spatial consistency of the displacement vector field and the reliability of deformation modeling. This provides accurate spatial deformation input for subsequent seepage path analysis, and improves the robustness and engineering applicability of defect detection as a whole.

[0087] 103. Input the displacement vector field into a preset spatiotemporal feature extraction model, and extract the heat flow continuity features of the sealed edge region through a multi-layer feature fusion structure to identify abnormal heat flow vortices and discontinuous boundaries;

[0088] Optionally, step 103 may specifically include the following steps:

[0089] 1031. Input the displacement vector field into a preset spatiotemporal feature extraction model, and perform vector direction consistency analysis on the strip-shaped regions on both sides of the sealing strip edge through the multi-layer feature fusion structure in the model;

[0090] 1032. Divide the strip-shaped area into equally spaced detection units, and perform dynamic smoothness calculation on the directional angle of three consecutive frames of displacement vectors in each detection unit. When the smoothness is lower than the continuity threshold, it is marked as a heat flow break point.

[0091] 1033. Based on the spatial distribution density of the heat flow break points, a discontinuous boundary is generated along the extension direction of the sealing strip, and the geometric accuracy of the boundary profile is corrected according to the rate of change of the distance between break points.

[0092] 1034. Detect the closed-loop path of the displacement vector around the periphery of the strip region. When more than two non-concentric closed-loop paths are detected and their rotation direction is opposite to the pressure attenuation direction, they are marked as abnormal heat flow vortices.

[0093] 1035. Based on the consistency between the corrected profile of the discontinuous boundary and the rotation direction of the anomalous heat flow vortex, output a topology containing the preferred direction of air leakage diffusion.

[0094] In the above scheme, the strip-shaped region refers to a strip-shaped detection range on both sides of the edge of the sealant strip, which is used to define the analysis region of the heat flow continuity feature; the dynamic smoothness calculation is a quantitative index of the fluctuation degree of the direction included angle of continuous multiple frames of displacement vectors, which is used to evaluate the risk of mutation or fracture of the heat flow direction; the heat flow fracture point refers to the local discontinuous position marked when the dynamic smoothness is lower than the threshold value; the discontinuous boundary is a defect boundary line extending along the sealant strip generated based on the spatial density of the heat flow fracture point; the annular closed path is a ring-shaped track formed by the change of the continuous vector direction in the displacement vector field, which represents the topological structure of the abnormal heat flow vortex; the rotational consistency refers to the physical correlation between the rotation direction of the vortex and the pressure attenuation direction, which is used to determine the air leakage diffusion trend.

[0095] Firstly, the displacement vector field is input into a preset spatiotemporal feature extraction model through step 1031, for example, a graph convolutional neural network is used to perform consistency analysis on the vector direction of the left strip-shaped region of the sealant strip, and it is detected that the average vector direction of the strip-shaped region is outward diffusion, and the right strip-shaped region is inward convergence; secondly, the strip-shaped region is divided into equidistant detection units through step 1032, for example, the dynamic smoothness calculation is performed on the direction included angle of the continuous three frames of displacement vectors of a certain unit at the lower right corner at an interval of 5 mm, the dynamic smoothness calculation is performed on the direction included angle of the continuous three frames of displacement vectors in each detection unit, and if the angle sequence is 45°, 120° and 90°, and the smoothness calculation value is lower than the threshold value 0.7, the heat flow fracture point is marked; then, the initial discontinuous boundary is generated by counting the spatial distribution density of the heat flow fracture point through step 1033, for example, when the number of fracture points in a unit length is 3 / cm, the jagged edge of the boundary profile is corrected based on the interval change rate, so that the geometric accuracy is improved; subsequently, the annular closed path is detected outside the strip-shaped region through step 1034, for example, two non-concentric annular paths are found, and the rotational direction thereof is opposite to the pressure attenuation direction, which is marked as an abnormal heat flow vortex; finally, the air leakage path is output by combining the corrected discontinuous boundary profile and the rotational consistency of the vortex through step 1035, for example, the vortex diffusion trend counterclockwise matches the boundary fracture direction, and the topological structure of the air leakage path is output.

[0096] In practical applications, taking the detection of the sealing performance of a broken bridge aluminum door and window as an example, the displacement vector field is analyzed by a space-time feature extraction model, it is found that the vector direction consistency of the left strip-shaped area of the sealant tape is relatively high, and there is a local direction mutation on the right side; in the detection unit divided in the lower right corner, the direction angle of the continuous three frames of displacement vectors fluctuates significantly, and the dynamic smoothness calculation value is lower than the threshold value, which is marked as a heat flow fracture point; after generating the initial discontinuous boundary based on the distribution density of the fracture point, the jagged profile is corrected according to the interval change rate; two non-concentric ring-shaped paths are detected on the periphery of the strip-shaped area, the rotation direction of which is opposite to the direction of pressure attenuation, and it is determined to be an abnormal heat flow vortex; finally, the modified boundary and the vortex rotation consistency are combined to output the topological structure of the air leakage path, which is preferentially diffused along the lower right corner boundary, and the preferential direction of leakage diffusion is clear.

[0097] The scheme accurately locates the heat flow abnormal area of the edge of the sealant tape through multi-layer feature fusion and vector direction consistency analysis; determines the heat flow fracture point based on the dynamic smoothness threshold value, effectively identifies the microscopic direction mutation defect; combines the fracture point density and interval to correct the discontinuous boundary profile, improves the geometric precision; detects the ring-shaped closed path and matches the rotation consistency, and reconstructs the diffusion trend of the abnormal heat flow vortex. Finally, the topological structure modeling and air leakage path dynamic prediction of the sealing defect are realized, which significantly enhances the detection robustness and result reliability of the complex leakage scene.

[0098] 104、In the process of pressure attenuation, a dynamic data compensation mechanism is triggered, and when a pressure mutation is detected, the heat image data sampling frequency is increased and a dynamic tracking map is generated;

[0099] Optionally, step 104 can specifically include the following steps:

[0100] 1041、Real-time monitoring of the first derivative of the pressure decay rate, when the absolute value of the derivative is detected to exceed the pressure mutation threshold, the fast sampling mode of the heat image acquisition device is started;

[0101] 1042、In the fast sampling mode, a dynamic data compensation mechanism is triggered, the heat image frame rate is increased to a preset sampling rate, and the microsecond-level response delay of the pressure sensor is recorded synchronously;

[0102] 1043、According to the difference between the pressure mutation threshold duration and the response delay, the increase rate of the heat image frame rate is dynamically adjusted;

[0103] 1044、After the fast sampling mode ends, the heat image data higher than the standard sampling rate is time-stamped and interpolated to align, and a tracking map containing sub-frame-level heat flow dynamics is generated.

[0104] In the above scheme, the first derivative of the pressure decay rate refers to the instantaneous slope of the change of pressure with time, which is used to quantify the intensity of the pressure mutation; the fast sampling mode is a high-frequency sampling state of the thermal imaging device triggered when the pressure mutation occurs, which captures the transient heat flow change by increasing the frame rate; the microsecond-level response delay refers to the signal transmission time deviation between the pressure sensor and the thermal imaging device, which needs to be synchronized and calibrated to ensure data synchronization; the timestamp interpolation alignment is a technology that aligns high-frequency thermal imaging data with a standard time axis through an algorithm, generating a heat flow dynamic tracking map with sub-frame-level accuracy.

[0105] In the embodiment of the present application, first, the first derivative of the pressure decay rate is monitored in real time through step 1041, for example, when the pressure change slope suddenly increases from -5 kPa / s to -15 kPa / s, the absolute value exceeds the threshold value of 10 kPa / s, and the fast sampling mode of the thermal imaging acquisition device is triggered immediately; second, through step 1042, the fast sampling mode is triggered, and the dynamic data compensation mechanism is triggered to increase the thermal imaging frame rate to a preset sampling rate, for example, the thermal imaging frame rate is increased from 30 Hz to 120 Hz, for example, during the pressure drop process of the aluminum alloy door and window, the response delay of the pressure sensor is recorded synchronously and stored as a time calibration parameter, which is 500 μs; then, through step 1043, the frame rate increase rate is dynamically adjusted according to the duration of the pressure mutation and the delay difference, for example, when the pressure mutation lasts for 80 ms and the delay is 500 μs, the frame rate is increased to 150 frames per second to match the dynamic characteristics of the pressure decay; finally, through step 1044, the high-frequency thermal imaging data is timestamped and aligned using a cubic spline interpolation algorithm, for example, 100 frames of data collected at 120 Hz are interpolated and mapped to a standard 30 Hz time axis, generating a tracking map that displays the dynamic process of the heat flow direction at the window frame joint changing from outward expansion to inward cohesion within 3 ms, and completely restoring the thermodynamic response details of the transient leakage event.

[0106] In actual application, taking the pressure decay detection of an aluminum alloy sliding window as an example, the first derivative of the pressure decay rate is monitored in real time to suddenly decrease from -8 kPa / s to -18 kPa / s, which exceeds the preset threshold value of 10 kPa / s, triggering the fast sampling mode of the thermal imaging device; the frame rate is increased from 30 Hz to 120 Hz, and the response delay of the pressure sensor is recorded synchronously, which is 550 μs; according to the difference between the duration of the pressure mutation of 70 ms and the delay, the frame rate is dynamically adjusted to 140 frames per second; after sampling, the high-frequency data is aligned to the standard time axis through cubic spline interpolation, generating a tracking map that displays the heat flow rate of the sealant strip at the lower left corner of the window frame increasing suddenly and reversing direction within 2.5 ms, completely capturing the heat flow dynamic details of the transient leakage event.

[0107] The scheme accurately captures the transient heat flow characteristics at the pressure mutation point through a dynamic data compensation mechanism, eliminates the time sequence deviation caused by equipment delay, combines frame rate adaptive adjustment and timestamp interpolation alignment, realizes sub-frame level heat flow dynamic restoration and data synchronization improvement, ensures the strict space-time correlation of pressure-heat flow response, finally enhances the detection sensitivity and data integrity of transient leakage events, provides high-precision transient process data support for dynamic modeling and leakage path analysis of sealing defects, and significantly improves the defect diagnosis reliability under complex working conditions.

[0108] 105. According to the topological structure of the abnormal heat flow vortex, the abnormal area divided by the discontinuous boundary, and the phase correlation of the pressure data, a three-dimensional seepage model of the sealing defect is established combined with the response event of the dynamic tracking map, and the geometric parameters of the air leakage path and the leakage equivalent are output according to the three-dimensional seepage model;

[0109] Optionally, step 105 can specifically include the following steps:

[0110] 1051. The topological structure of the abnormal heat flow vortex is decomposed into a plurality of annular closed paths, and the center point of each closed path and the minimum distance from the edge of the abnormal area are calculated based on the spatial geometric constraint of the abnormal area divided by the discontinuous boundary;

[0111] 1052. According to the phase correlation of the pressure data, the pressure decay process is divided into a linear decay stage and a nonlinear mutation stage, and the time domain interval in which the pressure change rate exceeds a preset value in the nonlinear mutation stage is extracted;

[0112] 1053. The response event of the dynamic tracking map is matched with the time domain interval, and the image frame set in which the heat flow direction in the thermal image sequence in the time domain interval reverses suddenly is marked; based on the center point distribution density of the annular closed path, the minimum distance, and the spatial superposition relationship of the image frame set, a three-dimensional grid of porous medium permeability distribution is constructed;

[0113] 1054. The heat flow vector angle difference between adjacent grid cells in the three-dimensional grid is calculated, and when the angle difference exceeds a structural deformation threshold, a seepage channel connecting the annular closed path and the door and window frame joint is generated, and finally a three-dimensional seepage model of the sealing defect is established.

[0114] In the above scheme, the annular closed path refers to the annular trajectory formed by the continuous vector direction change in the abnormal heat flow vortex; the phase correlation represents the matching relationship between the pressure decay time domain characteristics and the heat flow response; the three-dimensional grid is a three-dimensional unit structure divided based on spatial topological constraints, used to simulate the permeability distribution; the heat flow vector angle difference is a quantitative indicator of the direction difference between adjacent cells, used to determine the generation of the seepage channel.

[0115] In the embodiment of the present application, firstly, the topological structure of the abnormal heat flow vortex is decomposed into multiple annular closed paths by step 1051, for example, three annular contours are detected by using Hough transform, and the minimum distance from the center point of each closed path to the boundary is calculated based on the abnormal area space geometry constraint of the discontinuous boundary division, for example, the center point A is 1.5 mm away from the boundary, and the center point B is 2.0 mm away from the boundary; secondly, the pressure decay process is divided into linear and nonlinear stages according to the phase correlation of the pressure data by step 1052, for example, by analyzing the second derivative of the pressure-time curve, the time interval of the nonlinear mutation stage in which the pressure change rate exceeds 18 kPa / s is extracted, which is 10 ms~35 ms; then, the response events in the dynamic tracking atlas are matched with the above time interval by step 1053, and the image frame set in which the heat flow direction in the thermal image sequence in the time interval reverses is marked, based on the center point distribution density of the annular closed path, the minimum distance and the spatial superposition relationship of the image frame set, for example, 5 frames of data with reversed heat flow direction are marked within 10 ms~35 ms, and a three-dimensional grid is constructed combined with the center point distribution density of the annular path, such as 3 / cm² and the minimum distance, for example, the permeability distribution is divided into 1 mm³ units; finally, the heat flow vector angle difference of adjacent grid units is calculated by step 1054, for example, when the vector angle of adjacent units is 60° and exceeds the deformation threshold value 45°, an S-shaped seepage channel of the connected annular path and the window frame joint is generated, for example, the path extends from the center point B to the lower right window corner joint, the air leakage path length and the leakage equivalent parameter are output, and finally the three-dimensional seepage model of the sealing defect is established.

[0116] In practical applications, taking the detection of sealing defects of a broken bridge aluminum window as an example, the abnormal heat flow vortex is decomposed into three annular closed paths by Hough transform, and the minimum distances of the center points of the paths to the discontinuous boundary are 1.5 mm, 2.0 mm and 1.8 mm respectively; the nonlinear mutation stage is divided according to the second derivative analysis of the pressure decay data, and the time domain interval in which the pressure change rate exceeds the preset value is extracted; the 5 frames of thermal image data in which the heat flow direction is reversed in the dynamic tracking atlas in the time period are matched, and a three-dimensional grid is constructed in combination with the distribution density and minimum distance of the center points of the annular paths; when the difference degree of the heat flow vector included angles of adjacent units in the grid exceeds the deformation threshold, an S-shaped seepage channel of the connected annular path and the window frame joint is generated, and the geometric parameters and leakage equivalent of the air leakage path are output, so that the seepage diffusion path and the defect severity are determined. Through the topological decomposition of the annular path and the boundary constraint analysis, the spatial expansion trend of the seepage path is accurately located; combined with the pressure phase correlation division and the dynamic matching of the heat flow, the strict space-time correlation of the pressure-heat flow response is realized; based on the three-dimensional grid permeability modeling and the heat flow vector difference degree judgment, the seepage channel of the sealing defect area to the frame joint is reconstructed, and the geometric parameters and leakage equivalent of the air leakage path are provided. Finally, the three-dimensional visual modeling and dynamic diffusion simulation of the leakage defect are realized, and the engineering explainability of the defect diagnosis and the reliability of the maintenance decision are significantly improved.

[0117] 106、Based on the leakage equivalent, the geometric parameters and the preset sealing failure threshold curve, a hierarchical evaluation result associated with the partition of the door and window structure is generated.

[0118] Optionally, step 106 can specifically include the following steps:

[0119] 1061、Map the seepage channel extension length in the geometric parameters to the physical segmentation line of the structural partition of the door and window frame, and calculate the number of intersection points of the seepage channel and the frame joint in each partition;

[0120] 1062、According to the projection position of the leakage equivalent on the sealing failure threshold curve, determine the air tightness level attenuation coefficient corresponding to each partition;

[0121] 1063、When the number of intersection points in the same partition exceeds the structural strength threshold and the attenuation coefficient simultaneously exceeds the critical value, mark the partition as a high-risk area of sealing failure;

[0122] 1064、According to the consistency of the extension direction of the seepage channel in the adjacent partitions of the high-risk area, generate a hierarchical evaluation result containing a priority maintenance path indication and associated with the partition of the door and window structure.

[0123] In the above scheme, the physical division line of the structural partition refers to the area boundary line of the window frame divided according to mechanical properties or functions, such as the four corners of the window frame, the middle joint, and the like; the air tightness level attenuation coefficient is a quantitative index of the sealing performance decline calculated based on the projection position of the leakage equivalent on the sealing failure threshold curve; the structural strength threshold is the upper limit of the number of intersection points of the seepage channel and the joint in the window partition, and the structure is determined to be at risk of failure when the number of intersection points exceeds the threshold; the priority repair path indication is generated according to the consistency of the extension direction of the seepage channel of the adjacent partition of the high-risk area, for example, the area with concentrated diffusion direction is repaired preferentially.

[0124] In the embodiment of the application, firstly, the extension length of the seepage channel in the geometric parameter is mapped to the physical division line of the window structure partition in step 1061, for example, the aluminum window frame is divided into four partitions of upper left, upper right, lower left and lower right, the number of intersection points of the seepage channel and the joint in each partition is calculated based on the geometric topological relationship of the joint, such as the middle joint line, for example, the lower right corner partition detects that the seepage channel and the joint have 2 intersection points; secondly, the air tightness level attenuation coefficient is determined according to the projection position of the leakage equivalent on the sealing failure threshold curve in step 1062, for example, when the leakage equivalent is 0.75 m³ / h·m², the projection on the threshold curve is located in the middle-high risk transition interval, and the attenuation coefficient is calculated to be 0.82 by linear interpolation method; then, whether the number of intersection points in the partition exceeds the structural strength threshold is determined in step 1063, when the number of intersection points in the same partition exceeds the structural strength threshold and the attenuation coefficient exceeds the critical value at the same time, the partition is marked as a high-risk area of sealing failure; for example, the threshold is 2 and the attenuation coefficient exceeds the critical value, for example, 0.8, for example, the intersection point number of the lower right corner partition is 2, which is equal to the threshold, but the attenuation coefficient 0.82 is over limit, which is still marked as a high-risk area; finally, the priority repair path indication is generated according to the consistency of the extension direction of the seepage channel of the adjacent partition of the high-risk area in step 1064, for example, the cosine similarity of the seepage direction of the lower right corner partition and the left lower corner partition is 0.9, and the priority repair path indication is generated from the lower right corner to the left lower corner, and the grading evaluation result is output in combination with the window structure partition.

[0125] In practical application, taking the sealing performance evaluation of an aluminum alloy casement window as an example, the seepage channel extends to the lower right corner partition of the window frame and the number of intersection points exceeds the structural strength threshold, the leakage equivalent is projected to the high-risk interval of the sealing failure threshold curve, the air tightness level attenuation coefficient exceeds the critical value, and the partition is marked as a high-risk area; the extension direction of the seepage channel of the adjacent upper right corner partition is consistent with that of the high-risk area and has a high cosine similarity, a priority repair path indication is generated from the lower right corner to the upper right corner, guiding the engineering team to preferentially handle the continuous seepage diffusion path across the partitions.

[0126] The scheme accurately locates the high-risk sealing failure area through the space mapping of seepage channels and structural partition and the dynamic calculation of air tightness decay coefficient; combines the seepage direction consistency analysis and the maintenance path priority generation, and converts the abstract leakage parameter into the operable engineering decision basis; finally realizes the visual expression of defect classification and the accurate allocation of maintenance resources, significantly improves the efficiency of door and window maintenance and the economy of defect management, and guarantees the long-term reliability of building air tightness performance.

[0127] Figure 2 A scene diagram of an automatic door and window quality inspection method based on machine vision is provided for the embodiments of the application, as shown in Figure 2 for a complete embodiment of steps 101-106, including:

[0128] In the air tightness detection of the broken bridge aluminum casement window, the heat conduction baseline of the window frame sealant strip area is established at the low pressure stage 300 Pa in step 101, the heat flow diffusion rate is 0.12 W / (m·K), the heat flow rate of the upper right corner is detected to increase to 0.38 W / (m·K) and the direction is reversed at the high pressure stage 600 Pa, triggering the iteration update of the deformation compensation parameter; step 102 calculates the deformation area deviation of the upper right corner based on the corner heat flow direction constraint template, generates a displacement vector field to display the inward deformation of the frame 3.5 mm; step 103 divides the sealant strip area detection unit, finds that the dynamic smoothness of the upper right corner unit is 0.6, which is lower than the threshold, and marks it as a heat flow fracture point, and detects that the rotation directions of the two non-concentric ring paths are opposite to the pressure decay direction, and determines that it is an abnormal heat flow vortex; step 104 triggers the heat image frame rate to increase to 120 Hz at the pressure mutation stage-20 kPa / s, records the sensor delay 500 μs, and interpolates to generate a tracking map to display the heat flow direction reversal of the upper right corner within 2 ms; step 105 decomposes the abnormal vortex into two ring-shaped closed paths, the distance between the center point and the fracture boundary is 1.8 mm, constructs a three-dimensional grid and calculates the heat flow angle difference of adjacent units 55°, generates an S-shaped seepage channel, and outputs the leakage equivalent 0.75 m³ / (h·m²); step 106 extends the seepage channel to the upper right corner partition, the intersection number is 4 which is higher than the threshold value 2, the leakage equivalent is projected to the corresponding decay coefficient 0.88 in the high-risk interval, and the maintenance path "upper right corner→lower right corner" is generated combining the cosine similarity of the seepage direction of the adjacent lower right corner 0.91. Finally, the whole process analysis from dynamic data acquisition to leakage path three-dimensional modeling is realized, the high-risk area is accurately located and the engineering maintenance is guided through the heat-force coupling response and the seepage topology reconstruction, which significantly improves the defect diagnosis efficiency and the accuracy of building air tightness maintenance.

[0129] The scheme realizes full-link closed-loop analysis from dynamic data acquisition to defect grading evaluation through multi-step cooperative technology: synchronous acquisition of thermal and force data and baseline modeling eliminate environmental interference, accurately quantify sealing structure deformation characteristics; high-frequency sampling and spatiotemporal feature extraction capture transient heat flow fracture and abnormal vortex details; three-dimensional seepage model reconstruction seepage path topology, output quantitative leakage parameters; combined with partition threshold judgment and direction consistency analysis to generate maintenance priority decision. Finally, the abstract thermodynamic response is converted into an engineering executable defect positioning and maintenance strategy, significantly improving the accuracy of air tightness detection, defect diagnosis efficiency and maintenance resource allocation rationality, providing reliable technical support for building energy saving and long-term sealing performance guarantee.

[0130] Figure 3 A structure diagram of an automatic door and window quality inspection system based on machine vision is provided for the embodiments of the present application, as shown in Figure 3 The system comprises:

[0131] The acquisition module 31 is configured to synchronously acquire thermal image sequence data and pressure data during the pressurization process of the door and window, and the initial distribution of the temperature field collected in the low-pressure stage of the pressurization process is used to establish a heat conduction baseline, and the heat conduction mutation characteristics caused by the deformation of the sealing structure are captured in the high-pressure stage of the pressurization process.

[0132] The generation module 32 is configured to perform heat flow correction processing on the thermal image sequence data, establish a dynamic registration relationship based on the geometric characteristics of the door and window frame, the heat conduction baseline and the heat conduction mutation characteristics, and generate a displacement vector field with spatial constraints.

[0133] The identification module 33 is configured to input the displacement vector field into a spatiotemporal feature extraction model, extract the heat flow continuity characteristics of the sealing edge area through a multi-layer feature fusion structure, and identify abnormal heat flow vortexes and discontinuous boundaries.

[0134] The compensation module 34 is configured to trigger a dynamic data compensation mechanism during the pressure decay process, and when a pressure mutation is detected, the sampling frequency of the thermal image data is increased and a dynamic tracking map is generated.

[0135] The output module 35 is configured to establish a three-dimensional seepage model of the sealing defect according to the topological structure of the abnormal heat flow vortex, the abnormal area divided by the discontinuous boundary and the phase correlation of the pressure data, combined with the response event of the dynamic tracking map, and output the geometric parameters of the air leakage path and the leakage equivalent according to the three-dimensional seepage model.

[0136] The generation module 32 is further configured to generate a grading evaluation result associated with the partition of the door and window structure based on the leakage equivalent, the geometric parameters and a preset sealing failure threshold curve.

[0137] Figure 3The machine vision-based automatic door and window quality inspection system can perform Figure 1 The machine vision-based automatic door and window quality inspection method of the embodiment has the same implementation principle and technical effects as the machine vision-based automatic door and window quality inspection system. The specific operation modes of each module and unit of the machine vision-based automatic door and window quality inspection system in the above embodiment have been described in detail in the embodiment of the method, and will not be described in detail here.

[0138] In one possible design, Figure 3 The machine vision-based automatic door and window quality inspection system of the embodiment can be implemented as a computing device, such as a computer. Figure 3 As shown, the computing device can include a storage component 41 and a processing component 42.

[0139] The storage component 41 stores one or more computer instructions, wherein the one or more computer instructions are called and executed by the processing component 42.

[0140] The processing component 42 is configured to perform the above Figure 1 The machine vision-based automatic door and window quality inspection method of the embodiment.

[0141] The processing component 42 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 (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors or other electronic components, for executing the above method.

[0142] The storage component 41 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 a combination thereof, such as 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.

[0143] Of course, the computing device can also include other components, such as an input / output interface, a display component, a communication component, etc.

[0144] The input / output interface provides an interface between the processing component and the peripheral interface module, which can be an output device, an input device, etc.

[0145] The communication component is configured to facilitate wired or wireless communication between the computing device and other devices.

[0146] 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. The processing component, the storage component, and the like can be basic server resources rented or purchased from the cloud computing platform.

[0147] The embodiments of the present application also provide a computer storage medium storing a computer program, and the computer program can implement the above-described method when executed by a computer. Figure 1 An automatic door and window quality inspection method based on machine vision is provided.

[0148] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the above-described system, device and unit can refer to the corresponding processes in the foregoing method embodiments, which will not be described herein.

[0149] 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 embodiments according to actual needs. Those skilled in the art can understand and implement without creative labor.

[0150] From the above description of the embodiments, those skilled in the art can clearly understand that the embodiments can be realized by means of software and necessary general hardware platforms, and of course, 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 ROM / RAM, magnetic disk, optical disk, etc., and includes a number 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.

[0151] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some 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 machine vision-based automated door and window quality inspection method, characterized in that, The application relates to a method for detecting and evaluating sealing defects of a door and window structure. Synchronous acquisition of thermal image sequence data and pressure data during a door and window pressurization process, acquisition of an initial temperature field distribution during a low-pressure stage of the pressurization process and use of the initial temperature field distribution to establish a heat conduction baseline, and capture of heat conduction mutation characteristics caused by deformation of a sealing structure during a high-pressure stage of the pressurization process; Thermal flow correction processing of the thermal image sequence data, establishment of a dynamic registration relationship based on geometric characteristics of a door and window frame, the heat conduction baseline and the heat conduction mutation characteristics, and generation of a displacement vector field with spatial constraints according to the dynamic registration relationship; Input of the displacement vector field into a preset space-time feature extraction model, extraction of thermal flow continuity features of a sealing edge region through a multi-layer feature fusion structure, and identification of abnormal thermal flow vortexes and discontinuous boundaries; Triggering of a dynamic data compensation mechanism during a pressure decay process, enhancement of thermal image data sampling frequency when a pressure mutation is detected, and generation of a dynamic tracking atlas; Establishment of a three-dimensional seepage model of a sealing defect according to a topological structure of the abnormal thermal flow vortexes, an abnormal area divided by the discontinuous boundaries and phase correlation of the pressure data, combination of a response event of the dynamic tracking atlas, output of geometric parameters of a wind leakage path and a leakage equivalent according to the three-dimensional seepage model; Generation of a hierarchical evaluation result associated with a partition of a door and window structure based on the leakage equivalent, the geometric parameters and a preset sealing failure threshold curve.

2. The method of claim 1, wherein, The application relates to a method for detecting and evaluating sealing defects of a door and window structure. Decomposition of a topological structure of the abnormal thermal flow vortexes into a plurality of annular closed paths, calculation of a center point of each closed path and a minimum distance between the center point and an edge of an abnormal area based on spatial geometric constraints of the abnormal area divided by discontinuous boundaries; Division of a pressure decay process into a linear decay stage and a nonlinear mutation stage according to phase correlation of the pressure data, extraction of a time domain interval in which a pressure change rate exceeds a preset value in the nonlinear mutation stage; Matching of a response event of the dynamic tracking atlas with the time domain interval, marking of a set of image frames in which a thermal flow direction in a thermal image sequence in the time domain interval reverses suddenly, construction of a three-dimensional grid of permeability distribution of a porous medium based on a center point distribution density of the annular closed paths, the minimum distance and a spatial superposition relationship of the set of image frames; Calculation of a difference degree of a thermal flow vector angle between adjacent grid units in the three-dimensional grid, generation of a seepage channel connecting the annular closed paths and a door and window frame joint when the difference degree of the thermal flow vector angle exceeds a structural deformation threshold, and establishment of a three-dimensional seepage model of a sealing defect.

3. The method of claim 1, wherein, Generation of a hierarchical evaluation result associated with a partition of a door and window structure based on the leakage equivalent, the geometric parameters and a preset sealing failure threshold curve, including: Mapping of a seepage channel extension length in the geometric parameters to a physical segmentation line of a structural partition of a door and window frame, and calculation of a number of intersection points of the seepage channel and a frame joint in each partition. determining a corresponding air-tightness level attenuation coefficient of each sub-area according to the projection position of the leakage equivalent on the sealing failure threshold curve; when the number of intersection points in the same sub-area exceeds the structural strength threshold and the attenuation coefficient simultaneously exceeds the critical value, marking the sub-area as a high-risk area of sealing failure; generating a hierarchical evaluation result containing a priority repair path indication associated with the door and window structure sub-area according to the consistency of the seepage channel extension direction of adjacent sub-areas of the high-risk area.

4. The method of claim 1, wherein, establishing a dynamic registration relationship based on the geometric characteristics of the door and window frame, the heat conduction baseline and the heat conduction mutation feature, and generating a displacement vector field with spatial constraints according to the dynamic registration relationship, including: extracting the heat flow stable distribution pattern of the corner points and the joint lines in the geometric characteristics of the door and window frame in the heat conduction baseline to generate a reference heat flow direction constraint template; identifying the frame deformation area in the heat conduction mutation feature, calculating the heat flow direction deviation of the deformation area edge and the matching error of the reference heat flow direction constraint template; when the matching error exceeds the deformation compensation threshold, adjusting the vector weight of the reference heat flow direction constraint template according to the gradient distribution of the heat conduction mutation feature; based on the adjusted vector weight and the geometric topological relationship of the door and window frame joint, constructing a proportional mapping function of heat flow displacement and physical deformation; convert the heat flow vector difference of adjacent frames in the thermal image sequence data into spatial displacement with millimeter-level precision through the proportional mapping function to generate a displacement vector field with spatial constraints.

5. The method of claim 1, wherein, input the displacement vector field into a preset spatiotemporal feature extraction model, extract the heat flow continuity feature of the sealing edge area through the multi-layer feature fusion structure in the model, identify the abnormal heat flow vortex and discontinuous boundary, including: input the displacement vector field into a preset spatiotemporal feature extraction model, and analyze the vector direction consistency of the belt-shaped area on both sides of the sealing strip edge through the multi-layer feature fusion structure in the model; divide the detection unit at equal intervals in the belt-shaped area, and calculate the dynamic smoothness of the direction included angle of the continuous three frames of displacement vectors in each detection unit, and mark it as a heat flow fracture point when the smoothness is lower than the continuity threshold; based on the spatial distribution density of the heat flow fracture point, generate a discontinuous boundary along the extension direction of the sealing strip, and correct the geometric accuracy of the boundary contour according to the fracture point spacing change rate; detect the annular closed path of the displacement vector outside the belt-shaped area, and when two or more non-concentric annular paths are detected and their rotation direction is opposite to the pressure decay direction, mark it as an abnormal heat flow vortex; according to the consistency of the modified contour of the discontinuous boundary and the rotation direction of the abnormal heat flow vortex, output the topological structure containing the air leakage diffusion priority direction.

6. The method of claim 1, wherein, trigger the dynamic data compensation mechanism during the pressure decay process, and when a pressure mutation is detected, increase the thermal image data sampling frequency and generate a dynamic tracking map, including: real-time monitoring of the first derivative change of the pressure decay rate, when the absolute value of the derivative exceeds the pressure mutation threshold, starting the fast sampling mode of the thermal image acquisition device; In the fast sampling mode, a dynamic data compensation mechanism is triggered to increase the thermal image frame rate to a preset sampling rate, and the microsecond-level response delay of the pressure sensor is recorded synchronously; According to the difference between the pressure mutation threshold duration and the response delay, the increase rate of the thermal image frame rate is dynamically adjusted; After the fast sampling mode ends, the thermal image data higher than the standard sampling rate is time-stamped and interpolated to align, and a tracking atlas containing sub-frame-level thermal flow dynamics is generated.

7. The method of claim 1, wherein, In the low-pressure stage of the pressurization process, the initial distribution of the temperature field is collected and used to establish a thermal conduction baseline, and in the high-pressure stage of the pressurization process, the thermal conduction mutation characteristics caused by the deformation of the sealing structure are captured, including: In the low-pressure stage of the pressurization process, the pressure is maintained in a preset interval lower than the standard test pressure, the initial distribution of the temperature field is collected, and a thermal conduction baseline containing the steady-state thermal flow diffusion rate at the sealing joint is established; When the pressure in the pressurization process switches to the high-pressure stage, the thermal conduction mutation characteristics caused by the deformation of the sealing structure are captured, including the detection of the significant change event of the heat flow rate at the joint and the synchronous reversal of the heat flow direction; According to the spatiotemporal correlation between the thermal conduction baseline and the thermal conduction mutation characteristics, a deformation compensation parameter containing the deformation hysteresis effect of the frame material is established; when the phase difference between the thermal conduction mutation characteristics and the pressure change exceeds the dynamic tolerance threshold, the iterative update of the deformation compensation parameter is triggered during the maintenance in the high-pressure stage.

8. A machine vision-based automated door and window quality inspection system, characterized in that, including: a collection module for synchronously acquiring thermal image sequence data and pressure data during the door and window pressurization process, collecting the initial distribution of the temperature field in the low-pressure stage of the pressurization process to establish a thermal conduction baseline, and capturing the thermal conduction mutation characteristics caused by the deformation of the sealing structure in the high-pressure stage of the pressurization process; a generation module for performing heat flow correction processing on the thermal image sequence data, establishing a dynamic registration relationship based on the geometric characteristics of the door and window frame, the thermal conduction baseline, and the thermal conduction mutation characteristics, and generating a displacement vector field with spatial constraints; an identification module for inputting the displacement vector field into a spatiotemporal feature extraction model, extracting the thermal flow continuity features of the sealing edge region through a multi-layer feature fusion structure, and identifying abnormal thermal flow vortices and discontinuous boundaries; a compensation module for triggering a dynamic data compensation mechanism during the pressure decay process, and increasing the thermal image data sampling frequency and generating a dynamic tracking atlas when a pressure mutation is detected; an output module for establishing a three-dimensional seepage model of the sealing defect based on the topological structure of the abnormal thermal flow vortex, the abnormal area divided by the discontinuous boundary, and the phase correlation of the pressure data, combining the response event of the dynamic tracking atlas, outputting the geometric parameters of the air leakage path and the leakage equivalent according to the three-dimensional seepage model; The generation module is further configured to generate a hierarchical evaluation result associated with the partition of the door and window structure based on the leakage equivalent, the geometric parameters, and a preset sealing failure threshold curve.

9. A computing device, comprising: The method comprises 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, and realize the machine vision-based automatic door and window quality inspection method according to any one of claims 1-7.

10. A computer storage medium, characterized in that A computer program is stored, and when the computer program is executed by a computer, the machine vision-based automatic door and window quality inspection method according to any one of claims 1-7 is realized.

Images