A strain gauge mounting quality detection method and system, device, storage medium and program product

Abstract

The application discloses a strain gauge mounting quality detection method and system, equipment, a storage medium and a program product, and relates to the technical field of force / torque sensor manufacturing and detection. The method comprises the following steps: obtaining product information of a sensor to be detected; obtaining a visible light image and determining measured geometric information of each strain gauge; determining an executable excitation object according to bridge topology information, and applying an electric excitation signal; obtaining transient temperature information of each strain gauge region; performing spatial registration on the visible light image and the transient temperature information; extracting thermal response feature information of each strain gauge region; determining geometric error information of each strain gauge; performing differential comparison on the geometric error information and the thermal response feature information in the same preset symmetrical strain gauge group, and determining a mounting quality detection result. The application can comprehensively represent the geometric appearance quality, internal bonding interface quality and electric connection state of the strain gauge mounting, and realizes high repeatability, quantification and traceability of online detection.

Description

Technical Field

[0001] This application relates to the field of force / torque sensor manufacturing and testing technology, and in particular to a strain gauge mounting quality testing method and system, equipment, storage medium and program product. Background Technology

[0002] The quality of strain gauge mounting in strain gauge sensors directly affects the sensor's measurement accuracy and long-term stability. Relevant detection methods mainly rely on manual visual inspection or single-modal testing: visible light detection can identify positional deviations, angular deviations, and surface anomalies of the strain gauge, but it cannot detect voids in the internal adhesive layer covered by the strain gauge or interface debonding; infrared thermal imaging can reflect heat transfer anomalies, but its repeatability is insufficient due to factors such as ambient temperature, surface emissivity, and excitation power. Related technologies typically employ only one of these methods or a simple combination thereof, which cannot simultaneously and reliably characterize the geometric appearance quality, internal adhesive interface quality, and electrical connection status of the strain gauge mounting under mass production conditions.

[0003] Therefore, how to provide an online inspection solution that can comprehensively characterize multi-dimensional quality information of strain gauge mounting and has high repeatability, quantifiability, and traceability has become an urgent technical problem to be solved. Summary of the Invention

[0004] This application provides a method for inspecting the mounting quality of strain gauges to solve the above-mentioned technical problems.

[0005] A method for inspecting the mounting quality of strain gauges, applied to an inspection system, the method comprising: Obtain product information of the sensor to be tested, including the design location information of each strain gauge and the bridge topology information; Acquire visible light images of the sensor and determine the measured geometric information of each strain gauge based on the visible light images; Based on the bridge topology information, an executable excitation object is determined, and an electrical excitation signal is applied to the excitation object to induce a Joule thermal response in each strain gauge region. During the application of the electrical excitation signal and after the application is stopped, the transient temperature information of each strain gauge region is acquired; Spatial registration is performed between the visible light image and the transient temperature information; Based on the spatial registration results, thermal response characteristic information of each strain gauge region is extracted from the transient temperature information; Based on the designed location information and the measured geometric information, the geometric error information of each strain gauge is determined; The geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetrical strain gauge group are compared differentially, and the mounting quality test result of each strain gauge is determined based on the result of the differential comparison.

[0006] In some embodiments, before obtaining the product information of the sensor to be tested, the method further includes: confirming that the strain gauge adhesive of the sensor has reached a specified curing state and that the electrical connection has been completed, and placing the sensor under a specified load state.

[0007] In some embodiments, before obtaining the product information of the sensor to be tested, the method further includes: obtaining the coating process version information of the sensor; if the sensor has been coated, then calling the corresponding thermal reference parameters and model version according to the coating process version information; if the sensor has not been coated, then calling the thermal reference parameters and model version corresponding to the absence of a coating; wherein the thermal reference parameters are used to correct the transient temperature information, and the model version is used to perform defect identification on each strain gauge.

[0008] In some embodiments, before applying the electrical excitation signal to the excitation object, the method further includes: determining the waveform parameters of the electrical excitation signal based on the resistance of each strain gauge of the sensor, the adhesive material, the elastomer material, and the maximum allowable temperature rise, wherein the waveform parameters include one or more of peak power, cumulative energy, and maximum temperature rise.

[0009] In some embodiments, after the heating phase of the electrical excitation signal applied to the excitation object has ended, the method further includes: acquiring transient temperature information of each strain gauge region during the cooling phase; and simultaneously acquiring voltage and current signals of the excitation object.

[0010] In some implementations, determining the executable excitation object based on the bridge topology information includes selecting one or more from a single strain gauge, a single bridge arm, a half-bridge, and a full bridge as the excitation object based on the bridge topology information and accessible terminal information.

[0011] In some embodiments, before extracting the thermal response characteristic information of each strain gauge region from the transient temperature information based on the spatial registration result, the method further includes performing one or more of the following processes on the transient temperature information: baseline subtraction, ambient temperature correction, and emissivity compensation.

[0012] In some embodiments, after performing one or more of the following processing steps on the transient temperature information: baseline subtraction, ambient temperature correction, and emissivity compensation, the method further includes: energy normalization of the processed transient temperature information based on the excitation energy of the electrical excitation signal.

[0013] In some embodiments, spatial registration of the visible light image with the transient temperature information includes: mapping the positions of each strain gauge in the visible light image to the infrared image coordinates corresponding to the transient temperature information based on camera calibration parameters and workpiece positioning information.

[0014] In some embodiments, the preset symmetrical strain gauge group is pre-established based on the symmetrical structure information of the sensor, which includes one or more of geometric symmetry relationships and bridge circuit functional symmetry relationships.

[0015] In some embodiments, determining the mounting quality inspection results of each strain gauge based on the differential comparison results further includes: inputting the geometric error information, the thermal response characteristic information, and the symmetry difference index obtained from the differential comparison into the defect identification model to obtain the defect category, defect severity, and model uncertainty of each strain gauge.

[0016] In some embodiments, after obtaining the defect category, defect severity, and model uncertainty of each strain gauge, the following determination steps are further included: if the model uncertainty exceeds a preset threshold, the sensor detection result is marked as re-inspection; if the defect category belongs to the reworkable category and the defect severity is lower than a preset value, the sensor detection result is marked as rework; if the defect category belongs to the non-reworkable category or the defect severity exceeds a preset value, the sensor detection result is marked as isolation or scrap.

[0017] In some implementations, before performing the determination step, the method further includes: performing a detection validity gating check, which includes one or more of the following: workstation status check, environmental stability check, excitation waveform validity check, image quality check, and excitation temperature rise safety check; if the detection validity gating check fails, the current detection result is marked as invalid, and a re-inspection process is triggered.

[0018] In some embodiments, the method further includes: storing detection process data, the detection process data including the visible light image, the transient temperature information, the waveform data of the electrical excitation signal, the thermal response characteristic information, the mounting quality detection result, and the corresponding product serial number; and statistically analyzing the trend changes of process indicators based on the detection process data of multiple sensors in the same batch, and issuing a process warning signal when the process indicators exceed the warning range.

[0019] A strain gauge mounting quality inspection system, comprising: The information acquisition module is used to acquire product information of the sensor to be tested, including the design location information of each strain gauge and the bridge topology information; The visible light imaging module is used to acquire the visible light image of the sensor and determine the measured geometric information of each strain gauge based on the visible light image; An electrical excitation module is used to determine an executable excitation object based on the bridge topology information and to apply an electrical excitation signal to the excitation object; An infrared thermal imaging module is used to acquire transient temperature information of each strain gauge region during and after the application of the electrical excitation signal. A spatial registration module is used to spatially register the visible light image with the transient temperature information; The feature extraction module is used to extract the thermal response feature information of each strain gauge region from the transient temperature information based on the spatial registration result. The geometric error determination module is used to determine the geometric error information of each strain gauge based on the design location information and the measured geometric information. The defect identification module is used to perform differential comparison of the geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetrical strain gauge group, and determine the mounting quality inspection result of each strain gauge based on the result of the differential comparison. The output module is used to output the mounting quality inspection results.

[0020] In some embodiments, the electrical excitation module is further configured to determine the waveform parameters of the electrical excitation signal based on the resistance of each strain gauge of the sensor, the adhesive material, the elastomer material, and the maximum allowable temperature rise. The waveform parameters include one or more of peak power, cumulative energy, and maximum temperature rise.

[0021] In some embodiments, the defect identification module includes: a symmetry difference evaluation unit, used to compare the geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetric strain gauge group to determine the symmetry difference index; and a multimodal identification unit, used to input the geometric error information, the thermal response characteristic information and the symmetry difference index into the defect identification model to obtain the defect category, defect severity and model uncertainty of each strain gauge.

[0022] In some embodiments, the system further includes a quality judgment module, configured to perform the following judgments according to preset process rules based on the defect category, defect severity, and model uncertainty output by the defect identification module: if the model uncertainty exceeds a preset threshold, the sensor's detection result is marked as re-inspection; if the defect category is reworkable and the defect severity is lower than a preset value, the sensor's detection result is marked as rework; if the defect category is non-reworkable or the defect severity exceeds a preset value, the sensor's detection result is marked as isolation or scrap; if the above conditions are not triggered, the sensor's detection result is marked as qualified and released.

[0023] In some implementations, the quality judgment module is further configured to perform a detection validity gating check before performing the quality judgment. The detection validity gating check includes one or more of the following: workstation status check, environmental stability check, excitation waveform validity check, image quality check, and excitation temperature rise safety check. If the detection validity gating check fails, the current detection result is marked as invalid and a re-inspection process is triggered.

[0024] In some embodiments, the system further includes: a data traceability module for storing detection process data, the detection process data including the visible light image, the transient temperature information, the waveform data of the electrical excitation signal, the thermal response characteristic information, the mounting quality detection result, and the corresponding product serial number; and for statistically analyzing the trend changes of process indicators based on the detection process data of multiple sensors in the same batch, and issuing a process warning signal when the process indicators exceed the warning range.

[0025] A computer-readable storage medium for storing a computer program that causes a computer to perform the method described in any of the preceding claims.

[0026] A computer program product includes computer program instructions that cause a computer to perform the method described in any of the preceding claims.

[0027] The technical solution provided in this application obtains the measured geometric information of the strain gauge through visible light detection, generates Joule heat in the strain gauge itself through electrical excitation and obtains transient thermal response information, achieves multimodal information fusion through spatial registration, and performs differential comparison using symmetrical strain gauges. It can comprehensively characterize the geometric appearance quality, internal bonding interface quality and electrical connection status of the strain gauge mounting, and realizes online detection with high repeatability, quantifiability and traceability. Attached Figure Description

[0028] Figure 1 A schematic flowchart illustrating a strain gauge mounting quality inspection method provided in this application embodiment; Figure 2 This is a schematic diagram of the overall inspection process in the strain gauge mounting quality inspection method provided in the embodiments of this application; Figure 3 This is a schematic diagram of a strain gauge mounting quality inspection system provided in an embodiment of this application; Figure 4 This is a block diagram of the detection system in the strain gauge mounting quality inspection system provided in the embodiments of this application; Figure 5 This is a schematic diagram illustrating the timing of electrical excitation and transient thermal imaging synchronization provided in an embodiment of this application. Figure 6 This is a schematic diagram of the multimodal fusion and symmetry difference evaluation process provided in the embodiments of this application; Figure 7 This is a schematic diagram of the closed-loop process for online quality assessment and process handling provided in the embodiments of this application; Figure 8 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Detailed Implementation

[0029] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0030] In this embodiment, the strain gauge mounting quality inspection system is used to perform online quality inspection on sensors that have completed strain gauge mounting and electrical connection. This inspection system can be installed in the sensor manufacturing production line as an intermediate inspection station between the mounting process and the subsequent covering process. The main body executing the inspection system can be a dedicated inspection device integrating a visible light imaging unit, an infrared thermal imaging unit, an electrical excitation unit, and a data processing unit, or it can be an inspection system composed of multiple discrete devices.

[0031] The object of strain gauge mounting quality inspection is a resistance strain gauge sensor, including but not limited to a six-dimensional force / torque sensor, a load cell, a torque sensor, or other strain gauge sensing devices that use a resistance-sensitive element and can generate a measurable thermal response through controlled electrical excitation.

[0032] The object measured by a six-dimensional force / torque sensor can be represented as:

[0033] in, , , These represent the force components along the three orthogonal directions: x, y, and z. , , These represent the torque components about the three orthogonal directions: x, y, and z.

[0034] The key parameters involved in this application and their physical meanings are as follows: : No. The design center coordinates of the strain gauge are the theoretical position coordinates of the strain gauge in the sensor design drawings.

[0035] The first one obtained by visible light detection Measured center coordinates of the strain gauge.

[0036] : No. The positional error of a strain gauge represents the deviation distance between the measured center and the design center.

[0037] : No. The angular error of a strain gauge represents the angular deviation between the measured direction and the design direction.

[0038] : No. Each incentive recipient at time The voltage waveform.

[0039] : No. Each incentive recipient at time The current waveform.

[0040] : No. Each incentive recipient at time Instantaneous electrical power .

[0041] : Cumulative electrical excitation energy during a single heating phase.

[0042] : No. Each strain gauge region is located at... ,time The infrared temperature field.

[0043] : No. Temperature rise field of each strain gauge region relative to the baseline, .

[0044] Thermal response normalized to excitation energy.

[0045] The maximum allowable temperature rise threshold during the testing process is determined based on the strain gauge, adhesive, solder joint, elastomer material, and long-term stability verification results.

[0046] The equivalent thermal time constant during the heating phase reflects the rate at which the temperature rises during the heating process.

[0047] The equivalent thermal time constant during the cooling phase reflects the rate at which the temperature drops during the power-off cooling process.

[0048] : No. Thermal / electrical eigenvectors of a strain gauge.

[0049] Symmetrical strain gauge , The difference between the abnormal indicators, The larger the value, the more obvious the difference in thermal response between the two symmetrical strain gauges.

[0050] : Symmetric region feature orientation / scale mapping matrix, used to perform unified orientation and scale transformation of the features of two symmetrical strain gauges.

[0051] The feature weight matrix in symmetric difference calculation is used to weight the importance of different features.

[0052] : Defect category or category probability output.

[0053] : Defect severity output, which is a quantitative indicator.

[0054] Model uncertainty or low confidence index.

[0055] Manufacturing quality assessment and handling results.

[0056] Strain gauge six-dimensional force sensors typically consist of strain gauges bonded to multiple strain-sensitive regions of an elastomer, arranged in multiple Wheatstone bridges according to structural symmetry and bridge circuit functional relationships. The center position of the strain gauges, the bonding angle, the integrity of the sensitive grid, the adhesive layer thickness and interface bonding state, as well as the quality of the leads and solder joints, all affect the bridge circuit's zero point, sensitivity, consistency, hysteresis, creep, temperature stability, and inter-axial coupling stability.

[0057] In related technologies, the quality inspection after strain gauge mounting is mainly carried out using the following methods: On the one hand, manual visual inspection relies on magnifying glasses, microscopes, or ordinary industrial cameras. Visible light images can effectively observe surface-visible problems such as strain gauge position, angle, grid wire (i.e., the metal wire of the sensitive grid) / sensitive grid appearance, solder joint appearance, adhesive overflow, and surface contamination. However, visible light cannot directly penetrate the strain gauge substrate to observe interface defects located between the strain gauge substrate and the elastomer, such as local voids, bubbles, debonding (i.e., the state where the strain gauge substrate and the elastomer surface are partially separated), and abnormal adhesive layer thickness. The efficiency of manual visual inspection is affected by the skill level of the personnel, the judgment criteria are difficult to standardize, and it is difficult to form quantifiable mass production quality control data.

[0058] On the other hand, some related technologies have explored the use of infrared thermal imaging for defect detection. Strain gauges, as resistance-sensitive elements, generate Joule heating (i.e., the heat generated when current flows through a resistor) under controlled voltage or current excitation. Under normal bonding conditions, heat is conducted to the elastomer through the sensitive grid, substrate, and adhesive layer. When local debonding, voids, abnormal adhesive layer thickness, or interface contamination occur, the equivalent thermal resistance, heat capacity, and heat diffusion path will change. By collecting the temperature distribution in the strain gauge area, the state of the bonding interface can be indirectly inferred.

[0059] However, related infrared detection schemes have many limitations. If only the absolute temperature of a single infrared image is compared, the results are easily affected by factors such as ambient temperature, material surface emissivity (the ratio of surface radiative emissivity to that of a blackbody at the same temperature; different materials have different emissivity, for example, the emissivity of strain gauge surfaces, solder joint surfaces, and elastomer surfaces differs), reflection, airflow, excitation power, strain gauge resistance, elastomer geometry, and thermal conductivity, resulting in insufficient repeatability. If the electrical topology of a standard Wheatstone bridge is ignored and each strain gauge is directly assumed to be independently energized, this may not be feasible in actual production and could even alter the bridge circuit connection or introduce additional contact errors. Excessive electrical excitation energy can cause strain gauge self-heating (i.e., the strain gauge itself generates heat, leading to temperature rise), adhesive softening, thermal strain (i.e., thermal expansion deformation of the strain gauge caused by temperature changes), or zero-point changes, causing the detection process to negatively impact the tested product; conversely, insufficient energy may result in an inadequate signal-to-noise ratio. Furthermore, directly overlaying visible light and infrared images into a two-dimensional image cannot guarantee accurate correspondence of defect locations when strain gauges are located on curved surfaces, beam sides, or multi-height structures. Directly using convolutional neural networks to output "pass / fail" results in a black-box decision-making process, failing to distinguish between geometric defects, adhesive interface defects, and electrical connection defects, which is detrimental to rework decisions and manufacturing traceability. In addition, the relevant inspection process often lacks differential comparison of the functional symmetry areas of symmetrical strain gauges / bridge circuits, fails to fully utilize the structural symmetry of the six-dimensional force sensor itself to offset environmental and measurement common-mode errors, and lacks a closed loop from online inspection, re-inspection, rework to release / isolation, as well as management of models, parameters, images, electrical waveforms, and process versions, making it difficult to form stable batch quality control capabilities.

[0060] In view of the problems existing in the above-mentioned related technologies, the embodiments of this application propose the following technical solutions.

[0061] Example 1 Figure 1 This is a schematic flowchart illustrating a strain gauge mounting quality inspection method provided in an embodiment of this application. The method is applied to an inspection system. Figure 1 As shown, the method includes the following steps: Step 101: Obtain product information of the sensor to be tested.

[0062] Specifically, the detection system acquires product information about the sensor under test, including the design location information of each strain gauge and the bridge circuit topology information. The design location information includes the theoretical center coordinates, theoretical direction angle, sensitive grid area boundary, substrate boundary, and pad area location of each strain gauge in the design drawings. The bridge circuit topology information includes the sensor's bridge connection method (e.g., full bridge, half bridge, quarter bridge), the position of each strain gauge in the bridge circuit, accessible electrical terminals, and the functional definitions of each terminal.

[0063] In one example, the sensor to be detected is a six-dimensional force sensor with multiple strain gauges attached to its elastic body. Each strain gauge is configured to form multiple Wheatstone bridges according to a specific bridge circuit configuration. The system reads the design location information and bridge circuit topology information of each strain gauge from computer-aided design documents or process databases.

[0064] In some implementations, before obtaining product information of the sensor to be tested, it is confirmed that the strain gauge adhesive of the sensor has reached the specified curing state and the electrical connection has been completed, and the sensor is placed in a zero-load or reference load state to ensure the consistency of the sensor's stress state during testing.

[0065] In some implementations, before obtaining the product information of the sensor to be tested, the sensor's overlay process version information is also obtained. The overlay refers to the protective adhesive, thick coating, or potting material applied during sensor manufacturing to protect the strain gauges. If the sensor has an overlay, the corresponding thermal reference parameters and model version are retrieved based on the overlay process version information; if the sensor does not have an overlay, the corresponding thermal reference parameters and model version without an overlay are retrieved. The thermal reference parameters are used to correct transient temperature information (including emissivity correction and heat transfer compensation), and the model version is used to perform defect identification on each strain gauge.

[0066] Step 102: Acquire the visible light image of the sensor, and determine the measured geometric information of each strain gauge based on the visible light image.

[0067] Specifically, the detection system uses an industrial camera, telecentric lens or macro lens, and controlled illumination to acquire visible light images from the sensors. During acquisition, the sensors are under a specified zero-load or reference load state, and the clamps act on a rigid mounting area or non-sensitive area to avoid altering local thermal contact or strain gauge resistance due to clamping load. The detection enclosure preferably suppresses changes in ambient airflow and external thermal radiation, and records the ambient temperature. The system can employ template matching, edge / contour detection, key point detection, semantic segmentation, or deep learning methods to identify each strain gauge from the visible light images, obtaining information such as the measured center position, measured direction angle, sensitive grid shape, substrate boundary, weld appearance, and adhesive edge distribution for each strain gauge.

[0068] Visible light results are mainly used to identify surface defects such as positional misalignment, rotational skew, significant deformation of the sensitive grid, abnormal solder joint appearance, adhesive overflow, contamination, and obstruction. For voids, localized debonding, or abnormal interfacial thermal resistance in the adhesive layer beneath the strain gauge, the system does not use "uniform adhesive edges" as a sufficient condition for the internal adhesive layer to be qualified, but instead relies on subsequent transient thermal response for characterization.

[0069] Step 103: Determine the executable excitation object based on the bridge topology information, and apply an electrical excitation signal to the excitation object to generate a Joule thermal response in each strain gauge region.

[0070] Specifically, the testing system reads the actual bridge topology and accessible terminal information of the product to determine the feasible electrical excitation method. Based on the bridge topology and accessible terminal information, the system selects one or more of the following as the excitation object: single strain gauge, single bridge arm, half-bridge, and full bridge. For products with test terminals or electronic switching matrices, single-gauge or single-bridge arm excitation can be selected; for standard full-bridge non-separable structures, full-bridge excitation is used.

[0071] After an electrical excitation signal is applied to the object being excited, the strain gauge itself generates Joule heat as a resistive element. The heat is conducted to the elastomer through the sensitive grid, substrate, and adhesive layer. Under normal bonding conditions, the heat transfer path is unobstructed. However, when there is local debonding, voids, abnormal adhesive layer thickness, or interface contamination, the equivalent thermal resistance, heat capacity, and heat diffusion path change, resulting in differences in transient temperature response.

[0072] The excitation parameters must simultaneously meet the product's electrical safety, permissible self-heating, and detection signal-to-noise ratio requirements. A maximum temperature rise constraint can be set.

[0073] in, Indicates the first Each strain gauge area is located at all positions. And all moments The maximum temperature rise in the middle, To ensure a safe temperature rise threshold, this constraint ensures that the temperature in the strain gauge area does not exceed a preset safe upper limit during the testing process.

[0074] In some implementations, before applying an electrical excitation signal to the object being excited, the waveform parameters of the electrical excitation signal are determined based on the resistance of each strain gauge of the sensor, the adhesive material, the elastomer material, and the maximum allowable temperature rise. The waveform parameters include one or more of peak power, cumulative energy, and maximum temperature rise.

[0075] Step 104: During the application of the electrical excitation signal and after the application is stopped, acquire the transient temperature information of each strain gauge region.

[0076] Specifically, the detection system acquires a baseline temperature frame before electrical excitation, continuously acquires temperature sequences during the heating phase when the electrical excitation signal is applied, and continues to acquire temperature sequences during the cooling phase after the excitation is stopped, thereby obtaining transient temperature information of each strain gauge region over time. Infrared imaging preferably uses a macro or microthermal imaging module that meets the spatial resolution of the target strain gauge region.

[0077] To motivate the previous moment Define the temperature rise field as the baseline:

[0078] in, Indicates the first Each strain gauge region is located at... ,time Infrared temperature field, Indicates the first Each strain gauge region at the moment before excitation The initial temperature field, This represents the temperature rise field relative to the baseline. The calculation of this temperature rise field eliminates the influence of the difference between ambient temperature and the initial temperature of the sensor.

[0079] In some implementations, after the heating phase of the excitation object is completed, the transient temperature information of each strain gauge region during the cooling phase is acquired, and the voltage and current signals of the excitation object are acquired simultaneously.

[0080] It should be noted that for standard full-bridge strain gauges that cannot be accessed piece by piece, "synchronously obtaining the independent resistance value of each strain gauge" is not a necessary condition for detection. When full-bridge excitation is used, discrimination is made by symmetrical regions and timing characteristics, or by comparing corresponding regions with the same bridge topology and the same excitation conditions to obtain effective features that can be used for defect discrimination.

[0081] Step 105: Spatial registration of the visible light image and the transient temperature information.

[0082] Specifically, the detection system spatially registers visible light images with transient temperature information to establish the positional correspondence between the visible light region and the infrared temperature region of each strain gauge. Because the visible light camera and the infrared camera have different fields of view, resolutions, and distortions, the pixel positions of the same strain gauge are different in the two images. Spatial registration involves calculating the mapping relationship between the two coordinate systems.

[0083] In some implementations, spatial registration of the visible light image with transient temperature information includes: mapping the positions of each strain gauge in the visible light image to the corresponding infrared image coordinates of the transient temperature information based on camera calibration parameters and workpiece positioning information. This mapping relationship can generally be expressed as:

[0084] in, Represents the pixel coordinates in a visible light image. This represents the corresponding pixel coordinates in the infrared image. This represents the set of calibration parameters (including camera intrinsic and extrinsic parameters, distortion coefficients, and workpiece pose information). For approximately planar regions, the mapping... Homography can be used; for curved surfaces, beam sides, or multi-height areas, partition mapping, 3D point cloud / geometric model projection, or multi-view calibration can be used.

[0085] Step 106: Based on the spatial registration results, extract the thermal response characteristic information of each strain gauge region from the transient temperature information.

[0086] Specifically, the detection system uses the results of spatial registration to extract the temperature time series data of each strain gauge region from the transient temperature information, and extracts the temperature change characteristics of each strain gauge region over time to obtain thermal response characteristic information.

[0087] For the temperature rise sequence of each strain gauge or bridge section, the following features can be extracted: maximum temperature rise, average temperature rise, initial heating slope, cooling slope, thermal time constant, hot spot area, temperature gradient, spatial uniformity, and hot spot location. The hot spot area refers to the area of ​​the region where the temperature exceeds a set threshold, which can be determined statistically based on the temperature distribution of the normal strain gauge region. The hot spot location refers to the coordinates of the location of the highest temperature point. For an approximate first-order thermal response, the heating stage can be fitted using the following formula:

[0088] in, This represents the steady-state temperature rise during the heating phase, which is the saturation value that the temperature rise approaches when the heating time is long enough. The equivalent thermal time constant for the heating stage is the time required for the temperature to rise from zero to approximately 63.2% of the final steady-state temperature rise. The smaller the value, the faster the thermal response.

[0089] The cooling phase after a power outage can be achieved by:

[0090] in, To cool the initial temperature rise, i.e., the moment of power failure. The temperature rise value; This is the equivalent thermal time constant for the cooling stage, which physically represents the time required for the temperature to drop from its initial value to approximately 36.8% of that value. The smaller the value, the faster the cooling.

[0091] In some implementations, before extracting thermal response feature information, the transient temperature information undergoes one or more of the following processes: baseline subtraction, ambient temperature correction, and emissivity compensation. Emissivity compensation refers to correcting the infrared thermometry results based on the differences in emissivity of different material surfaces to improve the accuracy of temperature measurement.

[0092] For areas with strong surface reflectivity and unstable emissivity, the weight of the corresponding pixels can be reduced, or the measurability of temperature measurement in this area can be improved through validated surface treatments / reference markers. For mass production, it is preferable to use "relative temperature rise, time constant, spatial gradient, energy normalized response, and symmetry difference" as the main criteria, rather than just using the absolute infrared temperature at a single point.

[0093] In some embodiments, after processing the transient temperature information as described above, the processed transient temperature information is further normalized based on the excitation energy of the electrical excitation signal. Each incentive recipient at time Instantaneous power is defined as:

[0094] in, For the first Each incentive recipient at time voltage, For the first Each incentive recipient at time The current.

[0095] The excitation energy for the first heating stage is defined as:

[0096] in, This is the total duration of the heating phase. This represents the total electrical energy accumulated during this heating phase.

[0097] The energy-normalized thermal response can be defined as:

[0098] Energy normalization is used to reduce the impact of differences in normal excitation and dispersion in normal resistance, thereby improving the repeatability of detection.

[0099] Step 107: Determine the geometric error information of each strain gauge based on the designed location information and the measured geometric information.

[0100] Specifically, the detection system compares the measured geometric information obtained from visible light detection with the design position information to determine the geometric error information of each strain gauge. The geometric error information includes one or more of positional and angular errors.

[0101] No. The positional error of a strain gauge is defined as the Euclidean distance between the measured center and the design center:

[0102] in, The Euclidean norm of a vector (i.e., the straight-line distance between two points in space). For the first The design center coordinates of the strain gauge These are the measured center coordinates obtained from visible light detection.

[0103] Angular error is defined as the angular difference between the measured direction and the design direction:

[0104] in, For design direction and angle, For the measured direction and angle, the function Normalize the angle difference to Within the radius of curvature.

[0105] Position and angle thresholds are configured separately for different models, sensitive areas, and surface mount processes.

[0106] Step 108: Perform differential comparison on the geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetrical strain gauge group, and determine the mounting quality test result of each strain gauge based on the differential comparison result.

[0107] Specifically, the detection system performs differential comparison of the geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetrical strain gauge group based on the symmetrical structure information of the sensor, and determines the mounting quality detection result of each strain gauge based on the result of the differential comparison.

[0108] In some implementations, the preset symmetrical strain gauge group is established in advance based on the symmetrical structure information of the sensor. The symmetrical structure information includes one or more of geometric symmetry relationships and bridge circuit functional symmetry relationships. Geometric symmetry refers to pairs of strain gauges arranged symmetrically along spatial positions on the sensor's elastic body, such as two strain gauges symmetrically attached along the positive and negative X-axis directions in a six-dimensional force sensor. Bridge circuit functional symmetry refers to pairs of strain gauges in opposite arms of a Wheatstone bridge, such as R1 and R4, R2 and R3 in opposite arms of a full-bridge circuit. When establishing the preset symmetrical group, the comparability of local heat dissipation boundaries must be confirmed through design or calibration. For areas that are geometrically symmetrical but have significantly different substrate thicknesses, adjacent heat sources, lead heat dissipation, or mounting boundaries, the same symmetry threshold is not directly applied; instead, grouping benchmarks or model compensation are used.

[0109] For the Film and the first Strain gauges belonging to the same preset symmetry group can have their thermal / electrical eigenvectors defined.

[0110] in, For the first Maximum temperature rise of the strain gauge; The heating rate is the rate of change of temperature over time during the heating phase. This is the heating time constant; This is the cooling time constant; For hotspot area; This is the resistance value; This is the zero-point output of the bridge circuit (i.e., the output voltage of the bridge when there is no load). This represents the transpose of a vector.

[0111] And define the symmetry difference index:

[0112] in, This is a symmetrical region feature orientation / scale mapping matrix, used to perform a unified transformation of the orientation and scale of the features of two symmetrical strain gauges; This is the feature weight matrix, used to weight the importance of different features; Represents vector transpose. Symmetric difference index. This reflects the degree of difference in thermal / electrical characteristics between two strain gauges within the same pre-set symmetry group. The larger the value, the more obvious the characteristic difference between the two strain gauges.

[0113] In some implementations, determining the mounting quality inspection results of each strain gauge based on the results of differential comparison further includes: inputting geometric error information, thermal response characteristic information, and the symmetry difference index obtained from differential comparison into the defect identification model to obtain the defect category, defect severity, and model uncertainty of each strain gauge.

[0114] The defect identification model takes multimodal features as input, including visible light geometric features, energy-normalized infrared temporal features, synchronization electrical parameters, bridge topology labels, workpiece region labels, and symmetry difference indices. The model can employ convolutional networks, temporal networks, Transformers, traditional machine learning, or a combination thereof. The model outputs defect category, defect severity, and model uncertainty. Model uncertainty represents the model's confidence level in the prediction result; higher uncertainty indicates a less reliable model judgment.

[0115] In some implementations, after obtaining the defect category, defect severity, and model uncertainty of each strain gauge, the following determination step is also included: If the model uncertainty exceeds a preset threshold, the sensor's detection result will be marked as a retest.

[0116] If the defect category is reworkable and the defect severity is below a preset value, the sensor's detection result will be marked as reworkable. Reworkable categories refer to defect types that can be repaired through rework processes, such as poor solder joint appearance that has not yet caused an electrical break.

[0117] If the defect category is non-repairable or the defect severity exceeds a preset value, the sensor's detection result will be marked as isolated or scrapped. Non-repairable categories refer to defect types that cannot be repaired through rework in the manufacturing process, such as strain gauge sensor breakage or large-area debonding of the elastomer interface.

[0118] In some implementations, a detection validity gating check is performed before the judgment step. This check includes one or more of the following: workstation status check, environmental stability check, excitation waveform validity check, image quality check, and excitation temperature rise safety check. If the detection validity gating check fails, the detection result is marked as invalid, and a re-inspection process is triggered.

[0119] In some embodiments, the method provided in this application further includes: saving detection process data, which includes visible light images, transient temperature information, waveform data of electrical excitation signals, thermal response characteristic information, mounting quality inspection results, and corresponding product serial numbers; and statistically analyzing the trend changes of process indicators based on the detection process data of multiple sensors in the same batch, and issuing a process warning signal when the process indicators exceed the warning range.

[0120] In addition, the method flow in this embodiment one can also refer to Figure 5 The timing relationship between electrical excitation and transient thermal imaging synchronization is shown: Before a certain time, the system is in standby mode and completes forward visual acquisition; the infrared thermal imager continuously acquires baseline temperature frames to establish a temperature reference. At a certain moment, the system triggers continuous infrared thermal imaging acquisition and records the baseline temperature. ;exist At time 1, the system applies an electrical excitation signal to the object being excited, and Joule heating begins to occur in the strain gauge region. The temperature rises over time, and the heating phase lasts until 10:00 AM. At time; in At a certain point, the system stops electrical excitation, and the strain gauge area enters the natural cooling phase, where the temperature decreases over time. This cooling phase lasts until... At time; in At a certain moment, the system stops continuous infrared thermal imaging acquisition, completing one full heating-cooling transient temperature sequence acquisition; After the moment At each step, the system performs a visual verification. Throughout the process, the system synchronously records the voltage waveform of the excitation object. and current waveform Infrared thermal imaging is continuously acquired, while visible light acquisition is triggered.

[0121] Meanwhile, the multimodal fusion and symmetric difference evaluation in this embodiment can be referenced. Figure 6 The process is as follows: after geometric feature extraction from the visible light image, the measured geometric information is obtained, and the geometric error information is obtained by comparing it with the design location information; after preprocessing the infrared temperature sequence, thermal response feature information is extracted; synchronously acquired electrical parameters are used as supplementary feature inputs; the system reads the preset symmetrical strain gauge group information and calculates the symmetry difference index; finally, the geometric error information, thermal response feature information, electrical parameters and symmetry difference index are input together into the multimodal defect identification model, and the defect category, defect severity and model uncertainty of each strain gauge are output.

[0122] The quality assessment and process handling in this embodiment can be referred to... Figure 7 The closed-loop process is as follows: First, a gated inspection for product validity is performed. If the inspection fails, the station parameters are adjusted and the inspection is repeated. If the inspection passes, a hard threshold inspection for product quality is performed (position, angle, resistance / bridge zero point, continuity, etc.). Then, a judgment is made based on the defect type, defect severity, and model uncertainty, and the results are output as re-inspection, rework, isolation / scrap, or qualified release. After re-inspection, the product is returned for judgment, and after rework, the inspection is repeated. All inspection results are bound to the product serial number and uploaded to the database / manufacturing execution system.

[0123] Example 2 Figure 2 This is a schematic diagram of the overall inspection process in the strain gauge mounting quality inspection method provided in this application embodiment. Figure 2 As shown, the testing process includes the following steps: Step 1): Elastomer patching, curing, and soldering are completed.

[0124] The testing station is set after the strain gauge adhesive has reached the specified curing state and the solder joints or lead connections have been completed and stabilized. During testing, the elastomer is under a specified zero-load or reference load state, and the clamps act on the rigid mounting area or non-sensitive area to avoid altering local thermal contact or strain gauge resistance due to clamping load. The testing enclosure preferably suppresses changes in ambient airflow and external heat radiation, and the ambient temperature is recorded.

[0125] Step 2): Establish the workpiece coordinate system and the detection area.

[0126] A workpiece coordinate system is established on the elastic body, and the design center, orientation, sensitive grid area, pad area, and bridge circuit assignment of each strain gauge are read from computer-aided design files, process drawings, or strain gauge design templates. For curved surfaces, beam sides, and other areas, the design template is saved using a local coordinate system.

[0127] Step 3): Visible light image acquisition.

[0128] Visible light images of the sensor are acquired using an industrial camera, telecentric lens, or macro lens, and controlled illumination. During acquisition, the sensor is under specified zero-load or reference load conditions.

[0129] Step 4): Attachment geometry and appearance inspection.

[0130] Visible light imaging is used to detect the center position, rotation angle, sensitive grid / substrate profile, obvious grid wire deformation, solder joint appearance, adhesive overflow, and surface contamination of strain gauges. The uniformity of the internal adhesive layer covered by the strain gauge is not directly determined by visible light results. Visible light results are mainly used to identify surface defects such as positional misalignment, rotational skew, obvious sensitive grid deformation, abnormal solder joint appearance, adhesive overflow, contamination, and obstruction.

[0131] Step 5): Bridge topology identification and electrical excitation configuration.

[0132] Read the actual bridge circuit topology and accessible terminals, and select the excitation method of single chip, bridge arm, half bridge, or full bridge according to the product structure. Single chip / single bridge arm independent excitation is only performed when test terminals or switching matrices are designed; for standard bridges that cannot be independently disconnected, full bridge or achievable bridge circuit excitation is used, and the distinction is made by symmetry region and timing characteristics.

[0133] Step 6): Baseline thermal image acquisition.

[0134] Baseline temperature frames were acquired before electrical excitation to establish a benchmark for subsequent temperature rise calculations. Ambient temperature and initial temperature field distribution were recorded.

[0135] Step 7): Controlled electrical excitation and continuous acquisition of transient thermal images.

[0136] A controlled electrical excitation waveform is generated based on the strain gauge resistance, rated power consumption, adhesive and elastomer materials, allowable temperature rise, and thermal recovery time, limiting peak power, cumulative energy, and maximum temperature rise. Temperature sequences are continuously acquired during the heating and power-off cooling phases, and voltage, current, resistance, or bridge circuit output are recorded simultaneously.

[0137] Detection validity gating check.

[0138] To determine the validity of the detection, the following checks are performed: image quality, environmental stability, excitation waveform, and detection temperature rise limit. If the detection validity gate check fails, the detection is stopped, parameters are adjusted, or the detection is repeated; if it passes, proceed to the next step.

[0139] Step 8): Extract thermal response features.

[0140] Baseline subtraction, ambient temperature correction, emissivity / reflectance compensation, and excitation energy normalization were performed on the infrared sequence to extract features such as maximum temperature rise, heating rate, cooling rate, thermal time constant, hotspot area, temperature gradient, and spatial inhomogeneity. The extracted features include: (Temperature field) (temperature rise) (Temperature change rate, including heating rate and cooling rate) (Thermal time constant, including heating time constant) and cooling time constant ), (Temperature gradient), etc.

[0141] Step 9): Symmetrical strain gauge differential evaluation.

[0142] For geometrically or functionally symmetrical strain gauges, pairing / grouping relationships are established, and their transient thermal response and electrical parameters are compared under the same or normalized excitation conditions to reduce the common-mode effects of ambient temperature, batch thermal conductivity conditions, and partial emissivity errors. A symmetry difference index is constructed to identify anomalous gauges within the same structure.

[0143] Step 10): Multimodal fusion analysis.

[0144] This involves fusing visible light features, infrared temporal / thermal features, electrical parameters, bridge topology, and symmetry difference features. Multimodal fusion can employ convolutional networks, temporal networks, Transformers, traditional machine learning, or combinations thereof, with the inclusion of physical features and rule constraints being preferred.

[0145] Step 11): Output the defect category, severity, and uncertainty.

[0146] The multimodal defect identification model outputs the defect type, severity, and uncertainty. Defect categories include positional offset, angular deviation, sensitive grid deformation, solder joint anomalies, localized debonding, adhesive layer anomalies, microcracks, and surface contamination.

[0147] Step 12): QC judgment and process handling.

[0148] First, check hard thresholds such as location, angle, resistance, and maximum temperature rise. Then, combine these with the defect type, severity, and uncertainty, and make a judgment according to the rules validated by the process. Determine whether the release conditions are met: If the release conditions are not met, a pass / re-inspection / rework / isolation judgment will be generated, and the test will be re-executed after the repair.

[0149] If the release conditions are met, proceed to the next step and end the testing process.

[0150] Example 3 (System Summary Corresponding Example) Figure 3 This is a schematic diagram of a strain gauge mounting quality inspection system provided in an embodiment of this application. Figure 3 As shown, the system includes: The information acquisition module 301 is used to acquire product information of the sensor to be tested, including the design location information of each strain gauge and the bridge topology information. The information acquisition module can interact with the production management system through wired or wireless communication interfaces to read information such as product model, serial number, computer-aided design template, and process version.

[0151] Specifically, the information acquisition module 301 corresponds to the function of step 101 in Embodiment 1, and is used to acquire product information, including the design location information and bridge topology information of each strain gauge, to provide a data basis for subsequent testing. The information acquisition module 301 can integrate barcode scanning or radio frequency identification (RFID) reading functions to automatically identify the current workpiece model and call up the corresponding design data and process parameters.

[0152] The visible light imaging module 302 is used to acquire visible light images from the sensor and determine the measured geometric information of each strain gauge based on the visible light images. The visible light imaging module includes an industrial camera, a telecentric / macro lens, and a controlled illumination unit, and can use template matching, edge / contour detection, key point detection, semantic segmentation, or deep learning methods to achieve strain gauge identification and geometric parameter extraction.

[0153] Specifically, the visible light imaging module 302 corresponds to the function of step 102 in Embodiment 1, used to acquire visible light images and determine the measured geometric information (position, angle, contour, etc.) of each strain gauge. The telecentric / macro lens of this module can reduce perspective distortion and improve the accuracy of geometric measurements. The controlled illumination unit ensures the consistency of image acquisition under different time periods and different ambient light conditions.

[0154] The electrical excitation module 303 is used to determine an executable excitation object based on bridge topology information and apply an electrical excitation signal to the excitation object. The electrical excitation module includes a programmable voltage / current source, a bridge switching matrix, and a waveform generator. In some embodiments, the electrical excitation module is also used to determine the waveform parameters of the electrical excitation signal based on the resistance of each strain gauge of the sensor, the adhesive material, the elastomer material, and the maximum allowable temperature rise. The waveform parameters include one or more of peak power, cumulative energy, and maximum temperature rise.

[0155] Specifically, the electrical excitation module 303 corresponds to the function of step 103 in Embodiment 1, used to determine the excitation object and apply an electrical excitation signal. The bridge switching matrix of this module can be flexibly configured with excitation channels according to the actual bridge topology, compatible with various excitation methods such as single-chip, bridge arm, half-bridge, and full-bridge. The programmable voltage / current source can generate various waveforms such as pulses, square waves, and sine waves to adapt to the detection needs of different materials and structures.

[0156] The infrared thermal imaging module 304 is used to acquire transient temperature information of each strain gauge region during and after the application of the electrical excitation signal. The infrared thermal imaging module includes an infrared thermal imager, a macro / microscope lens, and an optional temperature reference unit.

[0157] Specifically, the infrared thermal imaging module 304 corresponds to the function of step 104 in Embodiment 1, and is used to acquire transient temperature information of each strain gauge region. The macro / microscope lens of this module can resolve temperature distribution at the strain gauge level, which is particularly important for small strain gauges or densely packed applications. A temperature reference unit (such as a blackbody reference) is used to correct infrared temperature measurement deviations in real time, improving the accuracy of temperature data.

[0158] The spatial registration module 305 is used to spatially register visible light images with transient temperature information. The spatial registration module stores the intrinsic parameters, distortion parameters, and relative pose calibration results of the visible light camera and the infrared camera, and establishes a mapping relationship from visible light coordinates to infrared coordinates through workpiece reference, marker points, or fixture reference.

[0159] Specifically, the spatial registration module 305 corresponds to the function of step 105 in Embodiment 1, and is used to spatially register the visible light image with the transient temperature information. Based on pre-calibrated camera parameters and workpiece positioning information, this module establishes a precise mapping between visible light pixel coordinates and infrared pixel coordinates, so that the geometric position of the strain gauge corresponds one-to-one with the thermal response area.

[0160] The feature extraction module 306 is used to extract thermal response feature information of each strain gauge region from the transient temperature information based on the spatial registration results. The feature extraction module performs baseline subtraction, ambient temperature correction, emissivity compensation, and energy normalization on the infrared temperature sequence to extract features such as maximum temperature rise, heating rate, cooling rate, thermal time constant, hot spot area, temperature gradient, and spatial temperature uniformity.

[0161] Specifically, the feature extraction module 306 corresponds to the function of step 106 in Embodiment 1, and is used to extract the thermal response feature information of each strain gauge region. This module can perform baseline subtraction, ambient temperature correction, and emissivity compensation on the temperature time series data of each strain gauge region, and perform energy normalization processing to extract features such as maximum temperature rise, heating rate, cooling rate, thermal time constant, hot spot area, temperature gradient, and spatial temperature uniformity, providing a quantitative basis for subsequent defect identification.

[0162] The geometric error determination module 307 is used to determine the geometric error information of each strain gauge based on the design location information and the measured geometric information. The geometric error information includes one or more of positional errors and angular errors.

[0163] Specifically, the geometric error determination module 307 corresponds to the function of step 107 in Embodiment 1, and is used to determine the geometric error information (position error) of each strain gauge. and angle error This module compares the measured geometric information obtained from visible light detection with the designed position information, and calculates the position error and angle error as important indicators for evaluating the geometric quality of the mounting.

[0164] The defect identification module 308 is used to perform differential comparison of the geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetrical strain gauge group, and determine the mounting quality inspection result of each strain gauge based on the result of the differential comparison.

[0165] Specifically, the defect identification module 308 corresponds to the function of step 108 in Embodiment 1. It is used to perform differential comparison of the geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetrical strain gauge group to determine the mounting quality inspection result. This module utilizes the structural symmetry of the sensor itself to effectively offset common-mode errors such as ambient temperature and material batch through differential comparison, thereby improving the accuracy and reliability of defect detection.

[0166] Output module 309 is used to output the placement quality inspection results. The output module can output the inspection results via a display, indicator light, printer, or communication interface.

[0167] Specifically, the output module 309 is used to output the placement quality inspection results. The inspection results can be output through a display, indicator light, printer or communication interface, so that the operator can understand the inspection status and results in real time.

[0168] In some embodiments, the defect identification module 308 further includes: a symmetry difference evaluation unit, used to compare the geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetric strain gauge group to determine the symmetry difference index; and a multimodal identification unit, used to input the geometric error information, thermal response characteristic information and symmetry difference index into the defect identification model to obtain the defect category, defect severity and model uncertainty of each strain gauge.

[0169] In some embodiments, the system further includes a quality judgment module 310, which performs the following judgments according to preset process rules based on the defect category, defect severity, and model uncertainty output by the defect identification module: if the model uncertainty exceeds a preset threshold, the sensor detection result is marked as re-inspection; if the defect category is reworkable and the defect severity is lower than a preset value, the sensor detection result is marked as rework; if the defect category is non-reworkable or the defect severity exceeds a preset value, the sensor detection result is marked as isolation or scrap; if the above conditions are not triggered, the sensor detection result is marked as qualified and released.

[0170] Specifically, the quality judgment module 310 corresponds to the judgment sub-step of step 108 in Embodiment 1. It is used to make a comprehensive judgment based on the defect category, defect severity and model uncertainty output by the defect identification module, combined with the product quality hard threshold (position error, angle error, resistance / bridge zero point, electrical continuity, etc.), and output the disposal conclusion such as re-inspection, rework, isolation / scrap or qualified release.

[0171] In some implementations, the quality judgment module 310 is also used to perform a detection validity gating check before performing the quality judgment. The detection validity gating check includes one or more of the following: workstation status check, environmental stability check, excitation waveform validity check, image quality check, and excitation temperature rise safety check. If the detection validity gating check fails, the current detection result is marked as invalid and a re-inspection process is triggered.

[0172] Specifically, the detection validity gating check function corresponds to the detection validity gating check sub-step in step 108 of Embodiment 1, ensuring that the detection data is valid before quality judgment is made, and avoiding misjudging invalid detection status as product defects.

[0173] In some implementations, the system also includes a data traceability module 311 for storing detection process data. The detection process data includes visible light images, transient temperature information, waveform data of electrical excitation signals, thermal response characteristic information, mounting quality inspection results, and corresponding product serial numbers; and statistical analysis of process indicator trends based on detection process data from multiple sensors in the same batch, issuing a process warning signal when the process indicators exceed the warning range.

[0174] Specifically, the data traceability module 311 corresponds to the data storage and trend warning sub-step in step 108 of Embodiment 1. It is used to store the detection process data (visible light image, transient temperature information, electrical excitation waveform data, thermal response characteristic information, mounting quality detection results and corresponding product serial numbers), and to statistically analyze the trend changes of process indicators based on the detection process data of multiple sensors in the same batch. When the process indicators exceed the warning range, a process warning signal is issued to realize preventive adjustment of the manufacturing process.

[0175] It should be noted that the specific functions of each module described in this embodiment three can be referred to in the detailed description of the corresponding steps in embodiment one. The modules work together through data flow and control flow to complete the complete process of strain gauge mounting quality inspection.

[0176] Example 4 Figure 4 This is a block diagram of the inspection system components in the strain gauge mounting quality inspection system provided in the embodiments of this application. Figure 4 As shown in the diagram, solid lines represent data flow, and dashed lines represent control flow. The detection system includes the following components: (1) Workpiece loading and positioning fixture. Used to load and repeatedly position the sensor to be tested. During testing, the elastic body is under a specified zero load or reference load state, and the fixture acts on the rigid mounting area or non-sensitive area to establish repeated positioning without compressing the sensitive structure, thus avoiding the clamping load from changing the local thermal contact or strain gauge resistance.

[0177] (2) Six-dimensional force sensor elastomer. The object to be tested has multiple strain gauges attached to it. The adhesive on the strain gauges has reached the specified curing state and the electrical connection has been completed. The strain gauges on the elastomer form multiple Wheatstone bridges according to the structural symmetry and bridge circuit function relationship.

[0178] (3) Visible light imaging module. Includes an industrial camera and a telecentric / macro lens, used to acquire visible light images of the sensor and detect the center position of the strain gauge, rotation angle, sensitive grid / substrate profile, obvious grid wire deformation, solder joint appearance, glue overflow and surface contamination.

[0179] (4) Infrared thermal imaging module. Includes an infrared thermal imager and a macro / microscope lens, used to acquire transient temperature information of each strain gauge region during and after the application of the electrical excitation signal.

[0180] (5) Illumination module. Used to provide controlled illumination for the visible light imaging module to ensure the stability and consistency of image acquisition.

[0181] (6) Electrical excitation and bridge switching module. Includes a programmable voltage / current source and a bridge switching matrix, used to apply controlled voltage / current according to the actual bridge topology and select the accessible single chip, bridge arm, half bridge or full bridge. Select the appropriate excitation method according to the product structure; for standard bridges that cannot be independently disconnected, use full bridge or achievable bridge excitation.

[0182] (7) Temperature Reference / Blackbody Reference Module. Used for temperature calibration and emissivity compensation of the infrared thermal imaging module, eliminating the influence of ambient temperature fluctuations and material emissivity differences on the accuracy of infrared temperature measurement, and improving the accuracy and repeatability of temperature measurement. A blackbody reference is a standard radiation source with a known emissivity (close to 1.0) used to correct the temperature measurement deviation of the infrared thermal imager.

[0183] (8) Image and data synchronous acquisition module. Used to uniformly trigger the synchronous acquisition of visible light images, infrared images, voltage, current, resistance and bridge output. Electrical excitation and infrared thermal imaging acquisition use a unified time reference. Baseline frames are acquired before excitation, temperature sequences are continuously acquired during the heating and power-off cooling phases, and voltage, current, resistance or bridge output is recorded synchronously.

[0184] (9) Coordinate Registration and Preprocessing Module. This module is used to spatially register visible light images with transient temperature information and to preprocess infrared temperature sequences. It uses calibration plates, workpiece features, camera calibration parameters, or 3D / partition mapping to bind visible light, infrared images, and computer-aided design coordinates to a unified workpiece coordinate system; it also performs one or more preprocessing steps on transient temperature information, including baseline subtraction, ambient temperature correction, and emissivity compensation.

[0185] (10) Thermal response feature extraction module. Used to extract thermal response feature information of each strain gauge region from transient temperature information based on the spatial registration results. Features such as maximum temperature rise, heating rate, cooling rate, thermal time constant, hot spot area, temperature gradient and spatial non-uniformity are extracted, and the processed transient temperature information is normalized according to the excitation energy of the electrical excitation signal.

[0186] (11) Detection validity gating module. Used to perform detection validity gating checks before quality judgment. The check contents include one or more of the following: workstation status check, environmental stability check, excitation waveform validity check, image quality check, and excitation temperature rise safety check; if the detection validity gating check fails, the current test result will be marked as invalid and a re-inspection process will be triggered.

[0187] (12) Symmetry difference analysis module. Used for differential evaluation of strain gauges with geometric or bridge circuit functional symmetry. Based on the symmetrical structure information of the sensor, a group of strain gauges with geometric symmetry or bridge circuit functional symmetry is established. The geometric error information and thermal response characteristic information of each strain gauge in the same strain gauge group are compared to determine the symmetry difference index, so as to reduce the common mode influence of ambient temperature, material batch and some emissivity errors.

[0188] (13) Multimodal Feature Recognition Module. This module is used to input geometric error information, thermal response feature information, and symmetry difference index into the defect recognition model to obtain the defect category, defect severity, and model uncertainty for each strain gauge. The model can employ convolutional networks, temporal networks, Transformers, traditional machine learning, or a combination thereof. Defect categories include one or more of the following: positional offset, angular deviation, sensitive gate deformation, solder joint anomaly, local debonding, adhesive layer anomaly, microcracks, and surface contamination.

[0189] (14) QC Judgment and Process Handling Module. Used to perform judgment according to preset process rules based on the defect category, defect severity and model uncertainty output by the defect identification module: if the model uncertainty exceeds the preset threshold, the sensor detection result is marked as re-inspection; if the defect category is reworkable and the defect severity is lower than the preset value, the sensor detection result is marked as rework; if the defect category is non-reworkable or the defect severity exceeds the preset value, the sensor detection result is marked as isolation or scrap; if the above conditions are not triggered, the sensor detection result is marked as qualified release.

[0190] (15) Display and Human-Machine Interaction Module. Used to output and display the placement quality inspection results and provide an interactive interface between the operator and the inspection system. The output module can output the inspection results through a monitor, indicator lights, printer or communication interface.

[0191] (16) Database / MES Traceability Module. Used to store inspection process data. Inspection process data includes visible light images, transient temperature information, waveform data of electrical excitation signals, thermal response characteristic information, mounting quality inspection results, and corresponding product serial numbers; and statistical analysis of process index trends based on inspection process data from multiple sensors in the same batch, issuing process warning signals when process indexes exceed the warning range. For each product serial number, the module stores visible light images, raw infrared sequences, excitation voltage / current waveforms, environmental parameters, registration parameters, feature vectors, defect results, handling conclusions, rework records, algorithm versions, threshold versions, and tooling / equipment numbers.

[0192] Example 5 Figure 5 This is a schematic diagram illustrating the timing synchronization of electrical excitation and transient thermal imaging provided in an embodiment of this application. Figure 5 As shown, the detection process is performed according to a unified time reference. The horizontal axis in the graph represents time, and the vertical axis represents temperature and excitation signal amplitude. The excitation control signal is shown above. The waveform is a rectangular pulse, and the temperature response curve at the corresponding moment is shown below. In the picture The baseline temperature is the initial temperature of the strain gauge region before the application of electrical excitation. Peak temperature, which is the highest temperature reached during the heating phase; This represents the maximum temperature rise. The time constant for the heating stage; This represents the time constant for the cooling phase. Output characteristic quantities include... (Maximum temperature rise) (Rate of temperature change) (Heating time constant) (Cooling time constant) (Hotspot area) and (Temperature gradient).

[0193] exist Before the specified time, the system completes front visual acquisition (triggered acquisition) to obtain visible light images from the sensor, which are used for subsequent geometric information extraction and position benchmark establishment.

[0194] exist At any given moment, the system triggers continuous infrared thermal image acquisition, and the infrared thermal imager continuously acquires temperature frames to establish a baseline temperature reference.

[0195] exist At a certain moment, the system applies an electrical excitation signal to the object being excited. (Such as pulsed voltage or constant current source), Joule heating begins to occur in the strain gauge region, and the temperature rises over time. The heating phase continues until... time.

[0196] exist At a certain point, the system stops electrical excitation, and the strain gauge area enters the natural cooling phase, with the temperature decreasing over time. This cooling phase continues until... time.

[0197] exist At that moment, the system stops continuous infrared thermal imaging acquisition, completing a full heating-cooling transient temperature sequence acquisition.

[0198] exist After the moment At any given moment, the system performs a visual verification (triggered acquisition) to acquire the visible light image from the sensor again. This image is used to verify whether the appearance of the strain gauge has changed during the electrical excitation process (such as displacement caused by softening of the adhesive layer, melting of the solder joints, etc.), ensuring that the detection process does not damage the product.

[0199] Throughout the process, infrared thermal images were acquired continuously (from...). to Continuous acquisition of temperature sequences), visible light acquisition is the trigger for acquisition (in respectively) Previously and (Each data point was collected once afterward). The system synchronously records the voltage waveform of the excitation object. and current waveform Used to calculate instantaneous power and cumulative energy .

[0200] After the test is completed, a cooling recovery time can be set. The next process can only proceed after the bridge circuit zero point or resistance value has returned to the allowable range.

[0201] It should be noted that this application uses a complete transient temperature sequence (heating stage + cooling stage) rather than a single thermal image for defect determination. By comparing the differences in thermal response between the heating and cooling stages, the heat transfer characteristics of the strain gauge mounting interface can be more comprehensively reflected, improving the accuracy and reliability of defect detection.

[0202] Example 6 Figure 6 This is a schematic diagram of the multimodal fusion and symmetric difference evaluation process provided in an embodiment of this application. Figure 6 As shown, the process includes the following steps: Step 1: Input data.

[0203] The system acquires multimodal input data, including: Visible light images This is used to extract the geometric appearance information of the strain gauge; Infrared thermal imaging sequence This is used to extract transient thermal response information from each strain gauge region; Electrical parameters (voltage) Current ,resistance ,power ), used to calculate excitation energy and monitor electrical status; Computer-aided design / design datum serves as a reference benchmark for judging geometric errors.

[0204] Step 2: Coordinate unification and region segmentation.

[0205] Spatial registration unifies visible light images, infrared thermographic sequences, and computer-aided design coordinates to the same workpiece coordinate system. Furthermore, each strain gauge undergoes fine-grained region segmentation, resulting in the following sub-regions: Sensitive grid area: The area where the sensitive grid of the strain gauge is located, used to extract the thermal response and observe the deformation of the grid wire; Solder joint area: The area where the solder joint connects the strain gauge lead to the elastomer pad, used to observe the appearance of the solder joint and detect welding abnormalities; Adhesive layer edge area: The adhesive layer area at the edge of the strain gauge substrate, used to observe adhesive overflow and adhesive layer uniformity; Symmetrical strain gauge groups (SG1, SG2, SG3, SG4): Based on the geometric symmetry of the sensor or the functional symmetry of the bridge circuit, the strain gauges are divided into multiple symmetrical groups. The strain gauges in each group have comparable heat transfer boundary conditions for subsequent differential comparison.

[0206] Step 3: Visible light feature extraction.

[0207] Geometric and morphological features of each strain gauge were extracted based on visible light images, including: Positional deviation , The positional deviations between the measured center and the design center in the x and y directions; Angle deviation : The angular deviation between the measured direction and the design direction; Gate deformation: Whether the geometry of the sensitive gate is deformed or damaged; Solder joint appearance: Check for any abnormalities in the appearance of the solder joint, such as poor soldering, burrs, or oxidation; Contamination / Adhesive Overflow: Whether there are contaminants or adhesive overflow on the surface of the strain gauge.

[0208] Step 4: Extraction of thermal features.

[0209] Based on the spatial registration results, thermal response characteristics are extracted from the transient temperature information of each strain gauge region, including: Maximum temperature rise Peak temperature rise reached during the heating phase; Temperature change rate The rate of temperature change during the heating and cooling phases; Thermal time constant , The equivalent thermal time constants for the heating and cooling stages; Temperature gradient : Spatial temperature distribution gradient within the strain gauge region; Hotspot area Area of ​​the region where the temperature exceeds a set threshold.

[0210] Step 5: Symmetric difference analysis.

[0211] For two strain gauges belonging to the same pre-defined symmetrical strain gauge group, a differential comparison analysis is performed. For strain gauges 1 and 2 (belonging to the same symmetrical group SG1), the following symmetry difference index is calculated:

[0212] in, and These are the maximum temperature rises of the first and second strain gauges, respectively. This represents absolute value operations.

[0213]

[0214] in, and These are the thermal time constants of the two strain gauges, respectively.

[0215] Similarly, multiple comparison pairs are performed on multiple symmetry groups (such as SG1 vs. SG3, SG2 vs. SG4, etc.), and the temperature rise difference and time constant difference between strain gauges within each group are calculated. The larger the symmetry difference index, the more significant the difference in thermal response between the two strain gauges within the same symmetry group, i.e., the higher the degree of asymmetry, and the more likely there are local defects. This differential comparison can effectively offset the common-mode effects of ambient temperature, material batch, and some station system errors in the same test.

[0216] Step 6: Multimodal fusion model.

[0217] The following multimodal features are input into the defect identification model: Visible light characteristics (positional deviation, angular deviation, grid deformation, solder joint appearance, contamination / overflow of adhesive); Thermal characteristics (maximum temperature rise, rate of temperature change, thermal time constant, temperature gradient, hot spot area); Electrical parameter characteristics (voltage, current, resistance, power); Symmetrical difference characteristics (symmetrical difference indices such as temperature rise difference and time constant difference).

[0218] Multimodal fusion models can employ convolutional networks, temporal networks, Transformers, traditional machine learning, or combinations thereof, with the inclusion of physical features and rule constraints being preferred. The model outputs the defect category, defect severity, and model uncertainty for each strain gauge.

[0219] Step 7: Output to QC for judgment.

[0220] The defect category, defect severity, and model uncertainty output by the multimodal fusion model are fed into the QC judgment and process handling procedures (e.g., Figure 7 As shown in the figure, the effective gating check, the hard threshold check of product quality, and the final process handling determination are carried out.

[0221] Step 8: End.

[0222] Complete the multimodal fusion and symmetry difference evaluation process, and output the judgment result to the next process.

[0223] Example 7 Figure 7 This is a schematic diagram of the closed-loop process for online quality assessment and process handling provided in an embodiment of this application. Figure 7 As shown, the closed-loop process includes the following steps: Step 701: Read the current workpiece and product model.

[0224] The system reads information such as the product model and serial number of the sensor to be tested, in preparation for subsequently calling the corresponding testing formula and threshold version.

[0225] Step 702: Call the detection formula and threshold version.

[0226] The rule engine comprehensively considers defect category, severity, uncertainty, product version, and process specifications, and calls upon the detection formula and threshold version that match the current product model. The detection formula includes hard index parameters such as position error threshold, angle error threshold, resistance / bridge zero-point threshold, electrical continuity threshold, and maximum allowable temperature rise threshold, as well as the version information of the defect identification model.

[0227] Step 703: Perform visible light detection and electrically excited transient thermal detection.

[0228] According to the detection formula, visible light detection and electrically excited transient thermal detection are performed to acquire visible light images, transient temperature sequences and synchronous electrical parameters.

[0229] Step 704: Was this test valid? Perform a detection validity gate check to determine if the detection is valid. The check includes: image quality, environmental stability, excitation waveform, and detection temperature rise limit. If the detection validity gate check fails, adjust the station parameters and re-detect, returning to step 703; if it passes, proceed to the next step.

[0230] Step 705: Check the hard threshold of product quality.

[0231] Read the model output (defect type, severity, uncertainty, symmetry difference), and check the hard thresholds of product quality, including: position error, angle error, resistance / bridge zero point, electrical continuity and other hard indicators.

[0232] Step 706: Are the release conditions met? A comprehensive assessment is made to determine whether the release conditions are met: If the release conditions are met, a qualified record is generated, uploaded to the database / manufacturing execution system, the next process is released, and the process ends.

[0233] If the release conditions are not met, proceed to the next step of the judgment.

[0234] Step 707: Is manual review / re-inspection required? Determine whether manual review or re-examination is needed. The criteria for this determination include: whether the model uncertainty exceeds a preset threshold, whether there are contradictions between multiple modalities, and whether the input exceeds the training distribution.

[0235] If manual review / re-inspection is required, a manual review / re-inspection instruction will be output. After review, the process will be returned to step 706 to re-determine whether the release conditions are met.

[0236] If no manual review / re-inspection is required, proceed to the next step of the judgment.

[0237] Step 708: Is it a reworkable defect? Determine whether the current defect belongs to the reworkable category. The determination criteria include: whether the defect category belongs to the preset reworkable category (such as poor solder joint appearance, abnormal glue amount, slight positional misalignment, etc.) and whether the defect severity is lower than the preset reworkable threshold.

[0238] If the defect is repairable, a repair instruction is output. After repair, the test is re-executed, and the process returns to step 703 for re-testing.

[0239] If the defect is not reworkable, an isolation / scrap determination is performed, the product is marked as isolated or scrapped, the data is uploaded to the database / manufacturing execution system, and the process ends.

[0240] Throughout the closed-loop process, all test results are linked to the product serial number and uploaded to the database / manufacturing execution system to form a single-item quality file. For reworked products, the test is repeated before entering the closed-loop decision process again. This closed-loop process ensures that scrap / isolation decisions are not directly determined by a single neural network result, but are jointly determined by test validity gating, product quality hard thresholds, model recognition, uncertainty, and process rules, reducing the risk of black-box misjudgments.

[0241] Example 8 Figure 8 This is a schematic diagram of an electronic device provided in an embodiment of this application. The electronic device can be a data processing device in the aforementioned detection system, an industrial control computer, an embedded industrial control computer, or the main control unit of a dedicated detection device. Figure 8 As shown, the electronic device includes: The processor 801 may be a central processing unit, a digital signal processor, a graphics processor, a field-programmable gate array or other programmable logic device, a general-purpose processor, an application-specific integrated circuit, etc. The processor 801 is used to execute computer programs to implement the various steps of the strain gauge mounting quality inspection method described in Embodiment 1 of this application.

[0242] The memory 802 is used to store computer programs. The memory 802 can be a read-only memory, random access memory, flash memory, hard disk, solid-state drive, or other storage media. The computer program stored in the memory 802 includes program code, which, when called by the processor 801, implements all the steps of the strain gauge mounting quality inspection method described in Embodiment 1. Specifically, the memory 802 can be divided into a program storage area and a data storage area. The program storage area is used to store the operating system, application program, and detection algorithm program; the data storage area is used to store detection process data such as visible light images, infrared thermographic sequences, electrical excitation waveform data, thermal response characteristic information, and mounting quality inspection results generated during the detection process.

[0243] Communication interface 803 is used for data communication with external devices. Communication interface 803 may include wired communication interfaces (such as Ethernet interfaces, serial interfaces, USB interfaces, etc.) and / or wireless communication interfaces (such as Wi-Fi interfaces, Bluetooth interfaces, 4G / 5G communication interfaces, etc.). Through communication interface 803, electronic devices can interact with visible light cameras, infrared thermal imagers, electrically excited units, manufacturing execution systems, database servers, etc. For example, communication interface 803 receives image data acquired by a visible light camera and thermal image sequence data acquired by an infrared thermal imager, and uploads the detection results to the MES system.

[0244] Bus 804 is used to connect processor 801, memory 802, and communication interface 803, enabling the transmission of data and control signals between these components. Bus 804 can be an industry-standard architecture bus, a peripheral component interconnection standard bus, an extended industry-standard architecture bus, or an advanced scalable interface bus. Bus 804 can be divided into address bus, data bus, and control bus.

[0245] In some implementations, the electronic device may also include an input / output interface (...) Figure 8 (Not shown in the image) is used to connect to human-computer interaction devices such as monitors, keyboards, mice, touch screens, indicator lights, and printers to enable the display of test results and interactive operation by operators.

[0246] The processor 801 is used to call and run the computer program stored in the memory 802 to perform the strain gauge mounting quality inspection method as described in Embodiment 1. Specifically, the processor 801 performs the following operations: acquires product information of the sensor to be tested through the communication interface 803; receives visible light images acquired by a visible light camera through the communication interface 803, and determines the measured geometric information of each strain gauge based on the visible light images; determines the executable excitation object according to the bridge topology information, and controls the electrical excitation module to apply an electrical excitation signal to the excitation object through the communication interface 803; receives transient temperature information of each strain gauge region acquired by an infrared thermal imager through the communication interface 803; performs spatial registration between the visible light images and the transient temperature information; extracts thermal response characteristic information of each strain gauge region from the transient temperature information based on the spatial registration result; determines the geometric error information of each strain gauge based on the design location information and the measured geometric information; and performs differential comparison between the geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetrical strain gauge group, determines the mounting quality detection result of each strain gauge based on the differential comparison result, and outputs the detection result through the output module or the communication interface 803.

[0247] The specific functions of the electronic device described in this embodiment can be referred to the detailed description of each step in Embodiment 1. When its processor 801 executes the computer program stored in the memory 802, it implements all the steps of the strain gauge mounting quality detection method described in Embodiment 1.

[0248] Example 9 This application also provides a computer-readable storage medium for storing a computer program. The computer-readable storage medium can be any medium capable of storing data, including but not limited to: a USB flash drive, a portable hard drive, a read-only memory, a random access memory, a magnetic disk, an optical disk (such as a CD-ROM, DVD-ROM), a solid-state drive, or other semiconductor storage devices. When the computer program stored in the storage medium is executed by a processor, it causes the computer to perform the method described in Embodiment 1. Specifically, the computer program includes program code, which includes computer-executable instructions for implementing the steps in Embodiment 1. When the computer program runs on the processor of a computer or embedded device, the processor performs the following operations: acquires product information of the sensor to be tested; acquires a visible light image of the sensor and determines the measured geometric information of each strain gauge; determines the executable excitation object based on the bridge topology information and applies an electrical excitation signal; acquires transient temperature information of each strain gauge region; spatially registers the visible light image with the transient temperature information; extracts thermal response feature information from the transient temperature information; determines the geometric error information of each strain gauge; and performs differential comparison of the geometric error information and thermal response feature information of each strain gauge within the same preset symmetrical strain gauge group, determining the mounting quality inspection result of each strain gauge based on the result of the differential comparison. This computer-readable storage medium is independent of temporary signal storage, enabling long-term data storage and is suitable for data recording and program deployment in the manufacturing environment.

[0249] The specific implementation of the computer-readable storage medium described in this embodiment can be found in the detailed description of each step in Embodiment 1. When the computer program stored therein is executed, it implements all the steps of the strain gauge mounting quality inspection method described in Embodiment 1.

[0250] Example 10 This application also provides a computer program product, including computer program instructions. The computer program product can be directly loaded into the internal memory of a computer, such as into the RAM of the control computer of the detection system. The computer program product can be in the form of a software installation package, downloaded and installed into the detection system via a wired or wireless network; it can also be pre-programmed into a ROM chip as part of the firmware of the detection system. When executed by a processor, the computer program instructions cause the computer to perform the method described in Embodiment 1. Specifically, the program instructions included in the computer program product implement the following functions: controlling a visible light camera to perform image acquisition and strain gauge identification; controlling an electrical excitation module to generate and apply electrical excitation signals; controlling an infrared thermal imager to perform transient temperature sequence acquisition; performing spatial registration, feature extraction, symmetry difference analysis, and defect identification on the acquired multimodal data; and outputting the mounting quality inspection results. The computer program product can be stored in a computer-readable storage medium and can also be transmitted and distributed via a network. The instructions included in the computer program product can be in the form of source code, object code, or executable file. The computer program product can run on a general-purpose computer, an industrial control computer, or an embedded system to achieve automated online inspection of strain gauge mounting quality. By using this computer program product, manufacturing enterprises can quickly deploy the testing method of this application without making large-scale modifications to existing hardware equipment, and it has good compatibility and promotional value.

[0251] The specific implementation of the computer program product described in this embodiment can be referred to the detailed description of each step in Embodiment 1. When the computer program instructions contained therein are executed, they implement all the steps of the strain gauge mounting quality inspection method described in Embodiment 1.

[0252] The technical solution of this application has the following beneficial effects: 1. This application obtains the measured geometric information of the strain gauge through visible light detection and obtains transient thermal response information by electrically exciting the strain gauge to generate Joule heat. This comprehensively characterizes the geometric appearance quality, internal bonding interface quality, and electrical connection status of the strain gauge mounting, solving the technical problem that single-mode detection methods in related technologies cannot simultaneously cover three types of quality problems. 2. This application achieves precise correspondence between visible light and infrared images through spatial registration, enabling accurate location of thermal anomalies to specific strain gauges and sensitive grid areas, providing precise location basis for defect location and rework. 3. This application effectively offsets the common-mode effects of ambient temperature, material batch, and station system errors by utilizing the structural symmetry of the sensor itself for differential comparison, improving the robustness and batch repeatability of detection. 4. This application selects an executable excitation method based on the actual bridge circuit topology, compatible with the mass production structure of the standard Wheatstone bridge, avoiding the impracticality problem caused by assuming that all strain gauges can be independently excited in related technologies. 5. This application controls the electrical excitation parameters through safe temperature rise and cumulative energy constraints, ensuring that the detection process does not damage the adhesive layer, zero point, or long-term stability. 6. This application simultaneously outputs defect category, severity, and uncertainty through a defect identification model, and performs tiered judgment by combining detection effectiveness gating and product quality hard thresholds, avoiding the black-box risk of a single neural network directly determining scrap, and improving the interpretability and reliability of the detection results. 7. This application enables traceability of the entire detection process data and batch process trend monitoring, providing a data foundation for preventative adjustments in the manufacturing process, and has good standardization promotion value and commercial application prospects.

[0253] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of protection of the claims. The various features in the embodiments of this application can be appropriately combined, split, deleted, or added according to actual needs, and the resulting technical solutions all fall within the scope of protection of this application. The hardware modules such as processors, memory, and communication interfaces described in the embodiments of this application can be implemented by general-purpose integrated circuits, application-specific integrated circuits, field-programmable gate arrays, or other programmable logic devices, and the corresponding computer programs can be stored in RAM, ROM, EPROM, EEPROM, flash memory, hard disk, solid-state drive, or other storage media. The methods in the embodiments of this application can be implemented by a computer program product, which includes computer program instructions that implement the methods in the embodiments of this application when executed by a processor.

Claims

1. A method for inspecting the mounting quality of strain gauges, characterized in that, Applied to a detection system, the method includes: Obtain product information of the sensor to be tested, including the design location information of each strain gauge and the bridge topology information; Acquire visible light images of the sensor and determine the measured geometric information of each strain gauge based on the visible light images; Based on the bridge topology information, an executable excitation object is determined, and an electrical excitation signal is applied to the excitation object to induce a Joule thermal response in each strain gauge region. During the application of the electrical excitation signal and after the application is stopped, the transient temperature information of each strain gauge region is acquired; Spatial registration is performed between the visible light image and the transient temperature information; Based on the spatial registration results, thermal response characteristic information of each strain gauge region is extracted from the transient temperature information; Based on the designed location information and the measured geometric information, the geometric error information of each strain gauge is determined; The geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetrical strain gauge group are compared differentially, and the mounting quality test result of each strain gauge is determined based on the result of the differential comparison.

2. The method according to claim 1, characterized in that, Before extracting the thermal response characteristic information of each strain gauge region from the transient temperature information based on the spatial registration result, the method further includes: The transient temperature information is subjected to one or more of the following processes: baseline subtraction, ambient temperature correction, and emissivity compensation.

3. The method according to claim 2, characterized in that, After performing one or more of the following processes on the transient temperature information: baseline subtraction, ambient temperature correction, and emissivity compensation, the method further includes: The processed transient temperature information is normalized based on the excitation energy of the electrical excitation signal.

4. The method according to claim 1, characterized in that, The determination of the mounting quality inspection results of each strain gauge based on the differential comparison results also includes: The geometric error information, the thermal response characteristic information, and the symmetry difference index obtained by differential comparison are input into the defect identification model to obtain the defect category, defect severity, and model uncertainty of each strain gauge.

5. The method according to claim 4, characterized in that, After obtaining the defect category, defect severity, and model uncertainty of each strain gauge, the following determination step is also included: If the uncertainty of the model exceeds a preset threshold, the detection result of the sensor will be marked as a retest. If the defect category belongs to the reworkable category and the defect severity is lower than a preset value, then the sensor detection result is marked as reworkable. If the defect category is non-repairable or the defect severity exceeds a preset value, the sensor's detection result will be marked as isolated or scrapped.

6. The method according to claim 5, characterized in that, Before performing the determination step, the method further includes: Perform a gated check for detection effectiveness, which includes one or more of the following: workstation status check, environmental stability check, excitation waveform effectiveness check, image quality check, and excitation temperature rise safety check; If the detection validity gating check fails, the current detection result will be marked as invalid, and a re-inspection process will be triggered.

7. A strain gauge mounting quality inspection system, characterized in that, include: The information acquisition module is used to acquire product information of the sensor to be tested, including the design location information of each strain gauge and the bridge topology information; The visible light imaging module is used to acquire the visible light image of the sensor and determine the measured geometric information of each strain gauge based on the visible light image; An electrical excitation module is used to determine an executable excitation object based on the bridge topology information and to apply an electrical excitation signal to the excitation object; An infrared thermal imaging module is used to acquire transient temperature information of each strain gauge region during and after the application of the electrical excitation signal. A spatial registration module is used to spatially register the visible light image with the transient temperature information; The feature extraction module is used to extract the thermal response feature information of each strain gauge region from the transient temperature information based on the spatial registration result. The geometric error determination module is used to determine the geometric error information of each strain gauge based on the design location information and the measured geometric information. The defect identification module is used to perform differential comparison of the geometric error information and thermal response characteristic information of each strain gauge in the same preset symmetrical strain gauge group, and determine the mounting quality inspection result of each strain gauge based on the result of the differential comparison. The output module is used to output the mounting quality inspection results.

8. An electronic device, characterized in that, include: A processor and a memory, the memory being used to store a computer program, and the processor being used to invoke and run the computer program stored in the memory to perform the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that, Used to store a computer program that causes a computer to perform the method as described in any one of claims 1 to 6.

10. A computer program product, characterized in that, It includes computer program instructions that cause a computer to perform the method as described in any one of claims 1 to 6.

Images