A plug-in kit multi-size point consistency verification anti-misjudgment method

Abstract

The present application relates to the technical field of visual detection, and discloses a kind of multi-size point consistency verification error prevention methods of plug-in kit, the method includes obtaining plug-in kit frame-by-frame image sequence and conversion alignment as multi-point physical size dataset, using rigid body geometric projection distortion law to solve space deflection angle value and to size measurement value Projection compensation is executed to generate multi-size joint physical consistency deviation value, with single-item tolerance limit, consistency determination threshold upper limit and aperture independent excess sign are carried out joint logic voting and drive pneumatic rejection valve;The present application eliminates false excess caused by attitude error by projection compensation, prevents qualified plug-in kit from being mis-rejected under the premise of not increasing three-dimensional scanning hardware, and ensures that hole size real defects are not missed and released.

Description

Technical Field

[0001] This invention relates to the field of visual inspection technology, and in particular to a method for preventing false judgments in the consistency verification of multiple size points of fitting kits. Background Technology

[0002] In the automated production process of connector kits, non-contact optical vision inspection systems are typically used to measure critical dimensions of the product online. This technology acquires images of the connector kits using an industrial camera, and employs image processing algorithms to extract data on feature points such as the size of the connector opening, the width of the bend, the size of the hole, and the height of the contact point. Each dimensional measurement is then compared with a preset tolerance range to achieve automatic sorting of good and defective products. This technology features high inspection speed and quantifiable evaluation capabilities.

[0003] In actual production, when a component experiences a slight tilt or rotation at the inspection station, the vision inspection system may experience measurement distortion, leading to qualified products being incorrectly classified as dimensionally defective. For example, a component with perfectly acceptable physical dimensions may slightly tilt forward as it enters the inspection area. In this case, the measured value of the contact point height in the system's optical image shifts upward, exceeding the tolerance limit, while the measured value of the bend width remains within the acceptable range, and the insertion / removal port size and hole size show no abnormalities. However, the inspection system, based solely on this single out-of-tolerance measurement of the contact point height, directly classifies the component as defective and removes it from the production line. During subsequent re-inspection, operators may find that the rejected component is actually in good condition, with all dimensions meeting specifications, but the system record shows an unacceptable contact point height. Repeated occurrences of this situation result in an abnormally low yield rate on the production line, leading to material waste and a continuous increase in manual re-inspection costs. Therefore, reducing misjudgments caused by posture errors during the visual inspection of components is a pressing issue that needs to be addressed in this field. Summary of the Invention

[0004] To address the shortcomings of existing technologies, the present invention aims to provide a method for preventing misjudgment in multi-dimensional consistency verification of insert kits. By eliminating spatial attitude projection distortion of insert kits and calculating the joint physical consistency deviation value of multiple dimensions, the method achieves single-point out-of-tolerance trigger type identification, effectively avoiding misjudgment caused by attitude errors and improving the reliability and accuracy of dimensional verification.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: In a first aspect, the present invention provides a method for preventing false judgments in multi-size point consistency verification of fitting kits, comprising: The image sequence of the interposer under online motion is acquired frame by frame using an industrial camera and a trigger photoelectric sensor, and then converted and output as a multi-point physical dimension dataset of the interposer by a dimensional measuring instrument. Spatial attitude projection distortion is eliminated using the multi-point physical dimension dataset of the plug kit output by the dimensional measuring instrument, and a multi-dimensional joint physical consistency deviation value is calculated and generated. Based on the multi-size joint physical consistency deviation value, the single-point out-of-tolerance trigger type is identified, and a qualified status mark or pneumatic rejection control signal is output to the pneumatic rejection valve or system display screen.

[0006] Furthermore, the step of acquiring frame-by-frame image sequences of the interposer kit under online motion conditions using an industrial camera and a trigger photoelectric sensor, and then converting and outputting a multi-point physical dimension dataset of the interposer kit via a dimensional measuring instrument, includes: As the insert kit passes through the conveyor belt and triggers the photoelectric sensor, the industrial camera is controlled to capture optical images of the insert kit surface at a fixed exposure time, generating a frame-by-frame image sequence of the insert kit. The image sequence of the insert kit is acquired frame by frame and input into the size measuring instrument. The pixel calibration function of the size measuring instrument is used to extract the physical feature edges. The size measuring instrument is used to extract four key feature dimensions of the insert kit on the two-dimensional projection plane, including the first measurement value of the insertion port size, the second measurement value of the bend width, the third measurement value of the hole size, and the fourth measurement value of the contact point height. The first measurement value of the plug-in port size, the second measurement value of the bend width, the third measurement value of the hole size, and the fourth measurement value of the contact point height are aligned and stitched together in the built-in storage module of the dimension measuring instrument according to the hardware arrival timestamp, and the multi-point physical dimension dataset of the plug kit is output.

[0007] Furthermore, the step of controlling the industrial camera to capture optical images of the insert's surface at a fixed exposure time when the insert passes through the conveyor belt and triggers the photoelectric sensor, and generating a frame-by-frame image sequence of the insert, includes: The trigger photoelectric sensor outputs a rising edge trigger level when the insert kit blocks the through-beam optical path, and the hardware clock of the control core of the verification system adds an incrementing hardware arrival timestamp to the rising edge since the system starts. After receiving the trigger level, the industrial camera immediately performs a single-frame exposure imaging of the insert surface with a fixed exposure time. The upper limit of the fixed exposure time is calculated using the allowable number of motion blur pixels, the physical size corresponding to a single pixel, and the conveyor belt linear speed. A sequence of grayscale image frames is generated, arranged in order of the hardware arrival timestamps. Each frame carries the corresponding hardware arrival timestamp, and the frames are used as the interpolation kit to acquire image sequences one by one.

[0008] Further, the step of aligning and stitching the first measured value of the insertion port size, the second measured value of the bend width, the third measured value of the hole size, and the fourth measured value of the contact point height in the built-in storage module of the dimensional measuring instrument according to the hardware arrival timestamp, and outputting the multi-point physical dimension dataset of the insertion kit includes: The size measuring instrument reads the hardware arrival timestamp inherited by each of the four key feature dimensions, and determines the measurement results of multiple corresponding features with equal hardware arrival timestamps as the same insert kit. The size measuring instrument concatenates four key feature dimensions with the same hardware arrival timestamp into a quintuple record that sequentially contains the corresponding hardware time and each dimension. The dimensional measuring instrument writes each quintuple record into the built-in storage module in ascending order of hardware arrival timestamp, and outputs a multi-point physical dimension dataset of the plug kit arranged in sequence.

[0009] Furthermore, the process of eliminating spatial attitude projection distortion and calculating the multi-dimensional joint physical consistency deviation value using the multi-point physical dimension dataset of the fitting output by the dimensional measuring instrument includes: The first item of the insertion port size measurement value, the second item of the bend width measurement value, the third item of the hole size measurement value, and the fourth item of the contact point height measurement value are read from the multi-point physical size dataset of the insert kit output by the size measuring instrument. Based on the inherent spatial rigid body geometry of the plug-in kit, the two-dimensional projection variation of four physical dimensions under small spatial tilting attitude errors is calculated. Based on the projection distortion law of the inherent rigid body geometry of the plug kit when it is deflected in space, the spatial deflection angle value is inversely calculated using the width measurement value at the second bend, and the projection compensation is performed on the first plug-in port size measurement value and the fourth contact point height measurement value to generate the multi-size joint physical consistency deviation value.

[0010] Furthermore, based on the projection distortion law of the inherent rigid body geometry of the insert when it deflects in space, the spatial deflection angle value is inferred from the second measurement value of the width at the bend, and projection compensation is performed on the first measurement value of the insert opening size and the fourth measurement value of the contact point height to generate the multi-size joint physical consistency deviation value, including: Based on the deviation ratio between the second measured value of the bend width and the preset standard bend width, the spatial deflection angle value of the current insert is calculated; The projection length is compensated by using the cosine function relationship corresponding to the spatial deflection angle value, and the reconstructed size value of the plug-in port and the reconstructed height value of the contact point after attitude error elimination are generated. Calculate the first absolute difference between the reconstructed size of the plug-in port after attitude error elimination and the preset standard plug-in port size; calculate the second absolute difference between the reconstructed height of the contact point after attitude error elimination and the preset standard contact point height; add the two absolute differences to generate a multi-dimensional joint physical consistency deviation value.

[0011] Further, the step of using the cosine function relationship corresponding to the spatial deflection angle value to perform projection length compensation, and generating the reconstructed size value of the plug-in port and the reconstructed height value of the contact point after attitude error elimination, includes: For the first measured value of the plug-in port size, the value is stretched and restored by dividing by the cosine of the deflection angle, and the value is equal to the value after the attitude error is eliminated by dividing by the cosine of the spatial deflection angle. For the fourth contact point height measurement value, it is compressed and restored by multiplying by the amplified deflection angle cosine, and the resulting value is equal to the contact point reconstructed height value after attitude error elimination by multiplying by the amplified deflection angle cosine.

[0012] Furthermore, the step of identifying the single-point out-of-tolerance trigger type based on the multi-dimensional joint physical consistency deviation value, and outputting a pass status mark or pneumatic rejection control signal to the pneumatic rejection valve or system display screen includes: Set the upper limit of the consistency judgment threshold and the individual tolerance limit in the control kernel of the verification system; Read the third hole size measurement value, determine whether the measurement value exceeds the single tolerance limit, and at the same time determine whether the physical consistency deviation value of the multi-size combination is greater than the upper limit of the consistency judgment threshold; If a single dimension exceeds the single tolerance limit, but the combined physical consistency deviation of multiple dimensions is less than or equal to the upper limit of the consistency judgment threshold, and the third dimension, the hole size measurement value, does not exceed the limit, then a qualified status mark is output and the process is allowed to continue. If a single dimension exceeds the single tolerance limit and the combined physical consistency deviation of multiple dimensions exceeds the upper limit of the consistency judgment threshold, or if the third dimension, the hole size measurement value, exceeds the limit, a pneumatic rejection control signal is issued to drive the pneumatic rejection valve to perform physical isolation rejection.

[0013] Furthermore, the step of reading the third hole size measurement value, determining whether the measurement value exceeds the individual tolerance limit, and simultaneously determining whether the combined physical consistency deviation value of multiple dimensions is greater than the upper limit of the consistency determination threshold includes: If any of the following measurements—the first (insert port size), the second (bend width), the third (hole size), and the fourth (contact point height)—exceeds the corresponding individual tolerance limit, then the individual tolerance trigger flag is set to 1; otherwise, it is set to 0. If the measured value of the third hole size exceeds the single tolerance limit, the hole diameter independent out-of-tolerance flag is set to 1; otherwise, it is set to 0. If the multi-size joint physical consistency deviation value is greater than the upper limit of the consistency judgment threshold, then the consistency deviation exceeding the limit flag is set to 1; otherwise, it is set to 0.

[0014] Furthermore, setting the upper limit of the consistency judgment threshold and the individual tolerance limit in the control kernel of the verification system includes: The individual tolerance limits are the upper and lower limits of the allowable tolerance range for all physical dimensions pre-defined based on the product specifications; Select a fixed number of insert kits whose dimensions are known to be qualified, image each kit one by one within the normal attitude fluctuation range of the conveyor belt, and calculate the sample multi-dimensional joint physical consistency deviation value corresponding to each insert kit. Extract the statistical mean and standard deviation of the corresponding deviation values ​​of all samples, obtain the value calculated by adding a preset multiple of the statistical mean to the standard deviation, and store and retrieve it as the upper limit of the consistency judgment threshold.

[0015] Furthermore, the calculation of the spatial deflection angle value of the current fitting assembly based on the deviation ratio between the measured width at the second bend and the preset standard bend width includes: Extract the ratio of the second item's measured width at the bend to the preset standard bend width, and use it as the corresponding deviation ratio; Compare the constant 1 with the calculated deviation ratio, and take the smaller value of the two as the cutoff ratio. An inverse cosine function is applied to the cutoff ratio, and the resulting inverse cosine angle is output as the spatial deflection angle value of the current insert kit for use in the next processing stage.

[0016] Secondly, the present invention provides a system for preventing false judgments in multi-size point consistency verification of fitting kits, comprising: The data acquisition and alignment conversion module is used to acquire frame-by-frame image sequences of the interposer under online motion conditions through an industrial camera and a trigger photoelectric sensor, and then converts and outputs a multi-point physical dimension dataset of the interposer through a dimensional measuring instrument. The projection distortion elimination and deviation calculation module is used to eliminate spatial attitude projection distortion using the multi-point physical dimension dataset of the plug kit output by the dimension measuring instrument, and calculate and generate a multi-dimensional joint physical consistency deviation value. The trigger type identification and control output module is used to identify the single-point out-of-tolerance trigger type based on the multi-size joint physical consistency deviation value, and output a qualified status mark or pneumatic rejection control signal to the pneumatic rejection valve or system display screen.

[0017] Compared to existing technologies, the advantages of this invention are as follows: By establishing a consistency correlation verification mechanism for multi-point data, this invention effectively solves the misjudgment problem caused by the inability of traditional independent single-point size threshold logic to distinguish between attitude projection distortion and actual physical deformation. Based on the projection distortion law of the inherent rigid body geometry of the insert kit during spatial deflection, the spatial deflection angle value is inferred from the width measurement value at the bend. Projection compression compensation and projection stretching compensation are performed on the insert size and contact point height, respectively, to generate a multi-dimensional joint physical consistency deviation value reflecting the actual deformation after deducting attitude distortion. Combined with the three-marker joint logic voting reinforced by the independent out-of-bounds marker of hole size, the system can accurately separate false single-point deviations induced by hardware problems such as conveyor belt vibration and tooling loosening or caused by the pose fluctuation of the insert kit itself from the multi-point linkage deformation caused by actual processing defects. Classification and rejection can be completed with only one-dimensional pneumatic control without the need to add three-dimensional scanning hardware. For hidden scenarios where plating foreign matter adheres to the plug-in port and the width of the bend does not have a corresponding compression, the system can still calculate the actual deformation residue and trigger rejection by joint logic voting. This avoids missing the actual hole size defect due to the compensation residual being accidentally too small, thus reducing the false rejection rate of qualified products while ensuring the electrical connection reliability of the plug-in kit. It also improves the anti-posture interference capability and detection accuracy of the dimensional visual inspection method in the field of geometric measurement. Attached Figure Description

[0018] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0019] Figure 1 This is a flowchart of a method for preventing false judgments in multi-size point consistency verification of insert kits according to the present invention; Figure 2 This is a schematic diagram of the layout of an online automated visual inspection production line in an embodiment of the present invention; Figure 3 This is a schematic diagram showing the distribution of key feature dimension measurement locations in an embodiment of the present invention; Figure 4 This is a schematic diagram of the multi-point physical size dataset splicing structure of the plug-in kit in an embodiment of the present invention; Figure 5 This is a schematic diagram of the spatial deflection projection distortion geometric constraints in an embodiment of the present invention; Figure 6 This is a schematic diagram illustrating the projection length compensation logic effect in an embodiment of the present invention; Figure 7This is a schematic diagram of the logic structure for multi-size joint physical consistency deviation in an embodiment of the present invention; Figure 8 This is a schematic diagram of the static module interface for joint logic voting in an embodiment of the present invention; Figure 9 This is a schematic diagram of a physical isolation and culling scenario in an embodiment of the present invention; Figure 10 This is a functional block diagram of a multi-size point consistency verification and anti-false judgment system for insert kits in this invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Example 1: Please see Figure 1 As shown, this embodiment provides a method for verifying the consistency of multiple dimensions of an insert kit, including: This embodiment provides a method for preventing false judgments in multi-dimensional consistency verification of insert kits, applied to an online automated visual inspection production line for insert kits. The insert kit refers to a metal stamping and bending electronic component used for electrical connection, structurally comprising four key features: a insertion port, a bend, a through hole, and an electrical contact point. The online automated visual inspection production line sequentially deploys a conveyor belt, a trigger photoelectric sensor, an industrial camera, a dimensional measuring instrument, a verification system control core, a system display screen, and a pneumatic rejection valve along the insert kit flow direction. The trigger photoelectric sensor is a through-beam photoelectric switch installed on the side of the conveyor belt, at the entrance of the industrial camera's field of view, to output a trigger level when the insert kit moves to the shooting station. The industrial camera is an area-array industrial camera with a global shutter, its optical axis installed at a fixed angle tilted relative to the normal of the conveyor belt surface, diagonally above the shooting station, so that the four structural features—the insertion port, bend, through hole, and contact point—are simultaneously imaged in the same two-dimensional projection image. The dimensional measuring instrument is connected to the industrial camera via an image data interface and has internal pixel calibration function and built-in storage module. The pneumatic rejection valve is installed at the rejection station at the end of the conveyor belt, with its high-pressure air nozzle facing the flow path of the insert kit on the conveyor belt. The control core of the verification system is connected to the dimensional measuring instrument, the system display screen, and the pneumatic rejection valve via an industrial communication bus. See also Figure 2 This is a schematic diagram of the layout of an online automated visual inspection production line provided in an embodiment of this application. Figure 2As shown, the online automated visual inspection production line sequentially deploys a conveyor belt, trigger photoelectric sensor, industrial camera, dimensional measuring instrument, verification system control core, system display screen, and pneumatic rejection valve along the flow direction of the insertion kit. The trigger photoelectric sensor uses a through-beam photoelectric switch, installed on the side of the conveyor belt at the entrance of the industrial camera's field of view. The industrial camera has a global shutter, and its optical axis is installed at a fixed angle relative to the normal of the conveyor belt table surface, diagonally above the shooting station. The dimensional measuring instrument is connected to the industrial camera via an image data interface. The pneumatic rejection valve is installed at the rejection station at the end of the conveyor belt, with its high-pressure air nozzle facing the flow path of the insertion kit on the conveyor belt. The verification system control core is connected to the dimensional measuring instrument, system display screen, and pneumatic rejection valve via an industrial communication bus. This layout allows four structural features—the insertion port, bend, through hole, and contact point—to be simultaneously imaged in the same two-dimensional projection image, providing a unified data acquisition and communication foundation for subsequent dimensional measurement, consistency verification, and rejection control.

[0022] S1: Acquire frame-by-frame image sequences of the interposer kit under online motion conditions using an industrial camera and trigger photoelectric sensors, and output a multi-point physical dimension dataset of the interposer kit via a dimensional measuring instrument.

[0023] The subsequent step S2 requires spatial attitude projection distortion elimination using a complete dimensional combination consisting of four physical dimensions of the same insert kit at the same time as input. However, there is an inherent geometric and mechanical coupling relationship between the insert kit's insertion / removal port size, bend width, hole size, and contact point height; the actual deformation of any dimension will synchronously change with the other dimensions. If the four dimensions are not extracted and time-aligned into a unified data structure at the data acquisition source, the dimensions will be scattered in an isolated manner with inconsistent time bases. It will be impossible to determine whether the four values ​​participating in the consistency calculation come from the same rigid body. Therefore, this step must first establish a unified temporal dataset of multi-point physical dimensions of the insert kit.

[0024] Further, step S1 includes: S11: When the insert kit passes through the conveyor belt and triggers the photoelectric sensor, control the industrial camera to capture optical images of the insert kit surface at a fixed exposure time, and generate a frame-by-frame image sequence of the insert kit.

[0025] The insert kit is in an online motion state. The conveyor belt drives the insert kit to continuously pass through the shooting station at a constant linear speed. If long exposure imaging is used, the displacement of the insert kit during the exposure will cause edge ghosting, causing the subsequent edge extraction position to shift. Therefore, it is necessary to freeze the motion with a fixed exposure time.

[0026] Specifically, the trigger photoelectric sensor outputs a rising edge trigger level when the insert kit blocks the through-beam optical path, and the hardware clock of the control core of the verification system marks the rising edge with a hardware arrival timestamp. The hardware arrival timestamp is a single-adjustment count value that increases in nanoseconds since the system starts up.

[0027] Upon receiving the trigger level, the industrial camera immediately performs a single-frame exposure imaging of the insert surface at a fixed exposure time.

[0028] The method for determining the fixed exposure time is as follows: Take the product of the allowable number of motion blur pixels and the physical size corresponding to a single pixel, divide by the conveyor belt speed as the upper limit of the fixed exposure time. For example, when the conveyor belt speed is 0.2 meters per second, the physical size corresponding to a single pixel is 0.02 millimeters, and the allowable motion blur is 1 pixel, the fixed exposure time value is no greater than... That is, the fixed exposure time is set to 100 microseconds.

[0029] The data structure of the frame-by-frame image sequence acquired by the plug-in kit is a sequence of grayscale image frames arranged in order of hardware arrival timestamps. Each frame is a two-dimensional pixel grayscale value matrix, and each frame carries its corresponding hardware arrival timestamp.

[0030] S12: The frame-by-frame image sequence of the insert kit is acquired and input into the dimension measuring instrument. The pixel calibration function of the dimension measuring instrument is used to extract the physical feature edges.

[0031] The S11 outputs a frame-by-frame image sequence in the form of a pixel grayscale matrix, which contains only pixel coordinate information and no physical scale information. If the pixel spacing is used directly for subsequent size determination, it will not be able to be compared with the tolerance range defined in millimeters. Therefore, this step requires first establishing a mapping from pixels to physical scale and extracting physical feature edges.

[0032] Specifically, the pixel calibration function of the size measuring instrument pre-obtains the physical size conversion ratio corresponding to a single pixel by imaging a standard gauge block of known length. For example, when a standard gauge block with a length of 10.000 mm spans 500 pixels in the image, the physical size conversion ratio corresponding to a single pixel is: .

[0033] The method for extracting physical feature edges is sub-pixel edge extraction based on gray-level gradient: the gray-level gradient magnitude is calculated pixel by pixel along the search direction perpendicular to the structural feature contour, and the sub-pixel position where the gray-level gradient magnitude reaches its maximum value is taken as the physical feature edge point.

[0034] The data structure of the physical feature edge is as follows: for each frame in the image sequence acquired frame by frame for the plug-in kit, output the sub-pixel edge point coordinate list corresponding to the plug-in port feature, bending point feature, through hole feature and contact point feature respectively. The elements in each coordinate list are two-dimensional coordinate pairs in pixels and inherit the hardware arrival timestamp of the frame.

[0035] S13: Specifically, the four key feature dimensions of the insert kit on the two-dimensional projection plane are extracted using a dimensional measuring instrument, including the first measurement value of the insertion port size, the second measurement value of the bend width, the third measurement value of the hole size, and the fourth measurement value of the contact point height.

[0036] The subsequent step S14 requires four dimensions with clear physical dimensions as the splicing objects, while the physical feature edges output by S12 are still a list of edge point coordinates in pixels, which have not yet been converted into physical dimension values ​​that can be compared with the tolerance range. Therefore, this step converts the physical feature edges into four key feature dimensions.

[0037] Specifically, the first measured value of the plug-in port size is the product of the pixel pitch between the two opposite inner edges corresponding to the plug-in port feature and the conversion ratio of the physical size of a single pixel. For example, the plug-in port size tolerance range is 0.63 to 0.66 mm. The plug-in port size is used to detect whether there are foreign objects stuck in the plug-in port due to electroplating.

[0038] The second measurement of the bend width is the product of the pixel spacing between the two lateral edges corresponding to the bend feature and the physical size conversion ratio of a single pixel. The bend width tolerance range is 5.27 to 5.37 mm.

[0039] The third measurement value for hole size is the product of the diameter of the fitted circle in the list of coordinates of the feature edge points of the through hole and the conversion ratio of the physical size corresponding to a single pixel. The hole size tolerance range is 0.95 to 1.05 mm.

[0040] The fourth contact point height measurement is the product of the pixel pitch along the height direction between the top edge of the contact point feature and the bottom edge of the fitting reference, and the physical size conversion ratio of a single pixel. The contact point height tolerance range is 0.25 to 0.28 mm. See also Figure 3 This is a schematic diagram showing the distribution of key feature dimension measurement locations provided in the embodiments of this application. For example... Figure 3As shown, the measurement locations of four key feature dimensions are marked on the main outline of the insert kit: the first measurement, the insertion port size, is taken from the pixel spacing between the two opposite inner edges corresponding to the insertion port feature; the second measurement, the bend width, is taken from the pixel spacing between the two lateral edges corresponding to the bend feature; the third measurement, the hole size, is taken from the diameter of the fitted circle in the coordinate list of the edge points of the through hole feature; and the fourth measurement, the contact point height, is taken from the pixel spacing along the height direction between the top edge point of the contact point feature and the reference bottom edge of the insert kit. This distribution of measurement locations allows the four key structural features simultaneously acquired by the industrial camera in the same two-dimensional projection image to be converted into dimensional values ​​with clear physical dimensions, providing a corresponding measurement basis for subsequent alignment and stitching according to the hardware arrival time stamp.

[0041] The first measurement value of the plug-in port size, the second measurement value of the bend width, the third measurement value of the hole size, and the fourth measurement value of the contact point height are all in millimeters, and all inherit the hardware arrival timestamp of the frame in which they are located.

[0042] S14: Align and stitch the first insertion port size measurement value, the second bend width measurement value, the third hole size measurement value, and the fourth contact point height measurement value in the built-in storage module of the dimension measuring instrument according to the hardware arrival timestamp, and output the multi-point physical dimension dataset of the insertion kit.

[0043] The four key dimensions of the insert kit exhibit linked deformation under actual machining defects. That is, the deviations of the insertion port size, the width of the bend, the hole size, and the contact point height are related to each other. Only by binding the four dimensions of the same insert kit into a single record can subsequent consistency verification be applied to the linked features of the same rigid body. Therefore, the hardware arrival timestamp must be used as the basis for alignment.

[0044] Specifically, the dimension measuring instrument reads the hardware arrival timestamps inherited by the four key feature dimensions output by S13, and determines the first plug-in port size measurement value, the second bend width measurement value, the third hole size measurement value, and the fourth contact point height measurement value with the same hardware arrival timestamp value as the measurement results of the same trigger event and the same plug-in kit.

[0045] The dimensional measuring instrument stitches together four key feature dimensions with the same hardware arrival timestamp into a quintuple record. The quintuple record sequentially includes the hardware arrival timestamp, the first item (plug-in port size measurement), the second item (bend width measurement), the third item (hole size measurement), and the fourth item (contact point height measurement).

[0046] The dimensional measuring instrument writes each five-tuple record into the built-in storage module in ascending order of hardware arrival timestamp, and outputs a multi-point physical dimension dataset of the plug kit. The data structure of the multi-point physical dimension dataset of the plug kit is an ordered array with five-tuple records as elements and arranged in ascending order of hardware arrival timestamp.

[0047] See Figure 4 This is a schematic diagram of the multi-point physical size dataset splicing structure of the plug-in kit provided in the embodiments of this application. For example... Figure 4 As shown, the left side contains scattered data, including five independent data blocks: hardware arrival timestamp, first item (insert port size measurement), second item (bend width measurement), third item (hole size measurement), and fourth item (contact point height measurement). The right side is an ordered array, where each row is a 5-tuple record. Each record contains five columns: timestamp, insert port size measurement, bend width measurement, hole size measurement, and contact point height measurement. These 5-tuple records are arranged in ascending order of hardware arrival timestamp. This splicing structure binds four key feature dimensions with the same hardware arrival timestamp value into the same 5-tuple record, ensuring that the four dimensions participating in subsequent consistency checks strictly originate from the rigid body state of the same insert kit at the same exposure time, thus cutting off the path to cross-component size mismatches at the data source.

[0048] The structure of the multi-point physical dimension dataset of the insert kit, aligned and stitched according to the hardware arrival timestamp, ensures that the four dimensions participating in subsequent consistency verification strictly originate from the rigid body state of the same insert kit at the same exposure time. In scenarios where the conveyor belt continuously flows at high speed and the time interval between adjacent insert kits within the field of view of the shooting station is short, if the four dimensions are not aligned according to the hardware arrival timestamp but are only stacked according to the order of extraction, the dimensions of two adjacent insert kits may be incorrectly stitched into the same record. This would cause the multi-dimensional joint physical consistency deviation value calculated by S233 to reflect the manufacturing differences between two insert kits rather than the attitude and deformation state of the same insert kit, thus rendering the judgment criteria of S33 and S34 meaningless physically. The alignment and stitching according to the hardware arrival timestamp cuts off the path of cross-part size mismatch from the data source, ensuring that the multi-dimensional joint physical consistency deviation value only carries the attitude and deformation information of a single insert kit.

[0049] S2: Use the multi-point physical dimension dataset of the plug-in kit output by the dimension measuring instrument to eliminate spatial attitude projection distortion and calculate the multi-dimensional joint physical consistency deviation value.

[0050] This step addresses the issue of synchronous two-dimensional projection distortion of data points across various dimensions when the fitting assembly experiences slight tilting or rotational attitude errors at the inspection station. The fundamental mechanism of this attitude error lies in the fact that after the fitting assembly undergoes spatial deflection relative to the industrial camera lens, its inherent rigid body geometry is compressed or stretched on the two-dimensional projection plane, causing the measured value of a certain dimension to exceed the tolerance range. However, this deviation does not originate from actual physical deformation and therefore does not accompany synchronous changes in other related dimensions according to the mechanical coupling law. If the independent single-point threshold logic is used, this false out-of-tolerance will be directly judged as defective. To numerically isolate attitude projection distortion and restore the true physical dimensions, this step utilizes the projection law of inherent rigid body geometry to first calculate the spatial deflection, then performs projection compensation on dimensions susceptible to attitude influence, and aggregates the compensated residuals into a multi-dimensional joint physical consistency deviation value.

[0051] Further, step S2 includes: S21: Read the first item of the insertion port size measurement value, the second item of the bend width measurement value, the third item of the hole size measurement value, and the fourth item of the contact point height measurement value corresponding to each insert from the multi-point physical dimension dataset of the insert output by the dimension measuring instrument.

[0052] The multi-point physical dimension dataset of the plug kit output by S14 is packaged into an ordered array to encapsulate the dimensions of each plug kit. The geometric calculations of S22 and S23 require the four dimensions of a single plug kit as direct input. If it is not disassembled by record first, the geometric calculation cannot obtain the specific values ​​of the object being processed. Therefore, this step performs record-by-record reading of the dataset.

[0053] Specifically, the verification system control kernel traverses each quintuple record in the multi-point physical size dataset of the interpolation kit in ascending order of hardware arrival timestamp.

[0054] For the currently traversed quintuple record, the verification system control kernel reads four values ​​from it: the first value of the plug-in port size measurement, the second value of the bend width measurement, the third value of the hole size measurement, and the fourth value of the contact point height measurement.

[0055] The verification system control kernel reads the four values ​​from the currently traversed quintuple record as the current interpolation kit size combination to be processed, and submits them to steps S22 and S23 in sequence.

[0056] S22: Based on the inherent spatial rigid body geometry of the plug-in kit, calculate the two-dimensional projection variation of four physical dimensions when a small spatial tilting attitude error occurs.

[0057] The insert kit is a rigid body structure, and its four key dimensions have different orientations in three-dimensional space: the width of the bend and the size of the insertion port are lateral dimensions located within the same reference plane, while the height of the contact point is a dimension protruding from the reference plane in the vertical direction. This difference in orientation determines that, under the same spatial deflection angle, the two-dimensional projection of the lateral dimension is compressed, while the two-dimensional projection of the protruding dimension is stretched. Furthermore, the sensitivity of the projection changes of each dimension to the deflection angle varies. Therefore, it is necessary to first quantify the two-dimensional projection change patterns of the four physical dimensions.

[0058] Specifically, the two-dimensional projection variation law is obtained through a calibration process: a known qualified insert kit with all four dimensions in the standard value is installed on a precision angle measuring turntable and placed within the field of view of an industrial camera. The insert kit is deflected relative to the industrial camera lens in a step of 1 degree within a range of -10 degrees to +10 degrees. At each deflection angle, the two-dimensional projection values ​​of the four physical dimensions are measured by a dimension measuring instrument.

[0059] For the second measurement of the width at the bend, the calibration data shows that its two-dimensional projection value decreases monotonically as the deflection angle increases, and the fitting result satisfies... ,in This refers to the width measurement at the second bend. To preset the standard bending width, Since the spatial deflection angle value is used, the second item, the width measurement at the bend, is determined as a compressed projection and used as the reference dimension for calculating the deflection angle.

[0060] For the first measurement of the insertion / removal port dimension, the calibration data shows that it and the second measurement of the bend width are on the same reference plane and both undergo projection compression, with the compression law satisfying... ,in This refers to the first measurement of the plug-in port dimensions. This is a preset standard plug-in port size.

[0061] For the fourth contact point height measurement, the calibration data shows that its two-dimensional projection value increases monotonically with the increase of the deflection angle, i.e., projection stretching occurs, and the fitting law satisfies... ,in This is the fourth measurement value for the contact point height. To preset the standard contact point height, The contact point height projection sensitivity coefficient; the contact point height projection sensitivity coefficient Using the projection stretch ratio at each deflection angle in the calibration process as the target variable and the corresponding deflection angle as the independent variable, the solution with the minimum sum of squared residuals is taken as the estimated value. For example, the contact point height projection sensitivity coefficient... The value is 3.9, which indicates that the rate of change of the projected length of the contact point as a protruding structure with the deflection angle is approximately 3.9 times the lateral dimension.

[0062] The verification system control kernel outputs a two-dimensional projection variation law. The data structure of the two-dimensional projection variation law is a mapping table that records the projection type and projection coefficient of each of the four physical dimensions. It includes the following: the width at the bend is a compression type reference; the insertion / extraction port size is a compression type with a coefficient of 1; the contact point height is a stretch type with a coefficient equal to the contact point height projection sensitivity coefficient. The four points are: the aperture size remains approximately unchanged under slight deflection; and the projection of the aperture remains approximately constant. (See also...) Figure 5 This is a schematic diagram of spatial deflection projection distortion geometric constraints provided in an embodiment of this application. For example... Figure 5 As shown in the figure, the standard state rigid body geometry of the insert assembly is represented by dashed lines, and the deflected state rigid body geometry is represented by solid lines. Both are projected onto the two-dimensional projection plane below. Projection compression is marked in the lateral dimension range, and projection stretching is marked in the corresponding range of the protruding structure. This schematic diagram illustrates the two-dimensional projection changes of the lateral dimensions (bend width and insert / pull-out size) under deflection, and the projection stretching of the contact point height (the height dimension protruding from the reference surface), providing a geometric basis for subsequently using the bend width measurement to infer the spatial deflection angle and for projection compensation of the insert / pull-out size and contact point height measurements.

[0063] If the width of the bend itself has a genuine manufacturing defect that causes its measured value to deviate significantly from the tolerance range, this deviation will be calculated as a large spatial deflection angle value. This will lead to excessive projection compensation on the insertion / removal port dimensions and contact point height, causing... and Significant deviation from their respective standard values, multi-size joint physical consistency deviation value With the corresponding increase, and in conjunction with the S32's judgment of exceeding the single tolerance limit of the bend width, the actual deformation defect at the bend can still be eliminated through joint logic voting.

[0064] S23: Based on the projection distortion law of the inherent rigid body geometry of the plug kit when it deflects in space, the spatial deflection angle value is inferred by using the width measurement value at the bend, and the projection compensation is performed on the plug-in port size measurement value and the contact point height measurement value to generate a multi-dimensional joint physical consistency deviation value.

[0065] The subsequent step S3 requires a scalar input to characterize whether true physical deformation still exists after subtracting attitude projection distortion. However, the four measurements output by S21 simultaneously mix true size information and attitude projection distortion, making direct differentiation impossible. This step utilizes the two-dimensional projection variation law determined in S22. First, it calculates the spatial deflection angle value from the compressed reference size. Then, it performs projection compensation on the plug-in port size and contact point height, which are susceptible to attitude influence, to restore the true physical size. Finally, it constructs a multi-size joint physical consistency deviation value by summing the residuals between the compensated reconstructed value and the standard value, providing S3 with a deformation metric after removing attitude distortion.

[0066] Further, step S23 includes: S231: Based on the deviation ratio between the measured width at the second bend and the preset standard bend width, calculate the spatial deflection angle value of the current fitting relative to the industrial camera lens.

[0067] Among the four physical dimensions, the bend width has the largest value and the narrowest tolerance band. Under the same spatial deflection, its absolute projection compression is the largest and the proportion submerged by noise is the smallest. Therefore, its projection compression has the highest signal-to-noise ratio for encoding the spatial deflection angle. Thus, the second measurement value of the bend width is selected as the reference dimension for calculating the spatial deflection angle value.

[0068] Specifically, the deviation ratio is the ratio of the measured width at the second bend to the preset standard bend width, i.e. .

[0069] The preset standard bending width The median value of the width tolerance range at the bend is taken, determined by the product design specifications of the insert kit. For example, a preset standard bend width is used. The value is 5.32 mm.

[0070] The spatial deflection angle value The calculation expression is: The design principle of the spatial deflection angle calculation expression is as follows: the measured width of the second bend is used as the lateral dimension in the reference plane. Its length on the two-dimensional projection plane is compressed according to the cosine law of the rigid body deflecting around the optical axis of the lens. Therefore, the ratio of the measured value to the standard value is equal to the cosine of the spatial deflection angle. Considering that the positive tolerance in the actual processing of the insert kit may cause the measured width of the bend to be slightly larger than the preset standard bend width, the following is introduced: Ratios greater than 1 are truncated to 1, resulting in a spatial deflection angle of 0 degrees. This indicates that the system determines no observable projection compression has occurred, and the subsequent projection compensation coefficient degenerates to 1. All dimensional measurements are output as is, and the physical logic is completely self-consistent. Taking the inverse cosine of the ratio (or the truncated value) yields the spatial deflection angle that caused the compression from the compressed observation length. The selection of the inverse cosine function directly corresponds to the physical law of cosine projection of lateral length with viewing angle deflection in optical imaging. The design goal of this expression is to make the deviation ratio smaller, i.e., the more the width at the bend is compressed, the larger the calculated spatial deflection angle value, thus maintaining monotonic consistency with the physical fact that the more severe the actual deflection and the more severe the projection distortion.

[0071] For example, when the width measurement at the second bend... millimeters, preset standard bending width When the diameter is millimeters, the deviation ratio is , Spatial deflection angle value Degrees; at this point, the second measurement of the bend width, 5.29 mm, still falls within the tolerance range of 5.27 to 5.37 mm, but its slight compression relative to the standard value has been resolved into a spatial deflection angle of approximately 6.09 degrees.

[0072] S232: Using the cosine function relationship corresponding to the spatial deflection angle value, the first plug-in port size measurement value and the fourth contact point height measurement value are projected length compensated to generate the plug-in port reconstructed size value and the contact point reconstructed height value after attitude error elimination.

[0073] The subsequent step S233 requires the dimensions of the plug-in port and contact point, after deducting attitude projection distortion, as input to calculate the residual that only reflects the actual deformation. However, the first plug-in port dimension measurement value and the fourth contact point height measurement value directly read from S21 still have compression and stretch distortion caused by spatial deflection superimposed. Therefore, this step performs back projection compensation on the two based on the spatial deflection angle value calculated by S231.

[0074] Specifically, for the first measured value of the plug-in port size, based on the rule determined in S22 that the plug-in port size is a compressed projection with a coefficient of 1, the size is stretched and restored by dividing by the cosine of the deflection angle to generate the reconstructed plug-in port size value after attitude error elimination. The calculation expression is as follows: in After eliminating attitude errors, the dimensions of the plug-in port are reconstructed. This refers to the first measurement of the plug-in port dimensions. The spatial deflection angle value generated for S231.

[0075] Regarding the fourth contact point height measurement value, the contact point height determined according to S22 is a stretched projection and the coefficient is the contact point height projection sensitivity coefficient. Based on the pattern, the contact point reconstructed height value is generated by multiplying by the cosine of the amplified deflection angle for compression and restoration. The calculation expression is as follows: in This represents the reconstructed height of the contact point after attitude error elimination. This is the fourth measurement value for the contact point height. The contact point height projection sensitivity coefficient obtained from S22 calibration. See [link / reference]. Figure 6 This is a schematic diagram illustrating the projection length compensation logic effect provided in an embodiment of this application. For example... Figure 6 As shown, the left area represents the reconstructed dimensions of the plug-in port after stretching and restoring the first measured value, thus eliminating the attitude error. The rectangle expands significantly from its initial short size after stretching and restoring. The right area represents the reconstructed height of the contact point after compression and restoring the fourth measured value, thus eliminating the attitude error. The rectangle shortens from its initial slender size after compression and restoring. This compensation effect corresponds to the calculation method of stretching and restoring the plug-in port dimensions by dividing the measured value by the cosine of the deflection angle and compressing and restoring the measured value by multiplying the measured value by the amplified cosine of the deflection angle. This makes the reconstructed value after compensation approximate the true physical dimensions that the plug-in assembly should present when there is no attitude error.

[0076] The design principle of the projection length compensation expression is as follows: The first term, the measured value of the insertion / removal port size, serves as the lateral dimension within the reference plane. Under deflection, this lateral dimension is cosine-compressed to a smaller observed value. Therefore, dividing this value by the cosine of the deflection angle stretches the compressed observed value back to the true opening size. Division is used instead of multiplication because the compression projection makes the measured value smaller than the true value. The fourth term, the measured value of the contact point height, serves as the protruding structure. Under deflection, as the protruding surface tends to face the lens, this contact point height is stretched to a larger observed value. Therefore, multiplying this value by the cosine of the amplified deflection angle compresses the stretched observed value back to the true height. Multiplication is used because the stretched projection makes the measured value larger than the true value. A contact point height projection sensitivity coefficient is introduced within the cosine. This is because the rate of change of the projected length of the convex structure with deflection is several times that of the lateral dimension; when the spatial deflection angle calculated by S231 is 0 degrees... The two equations degenerate into: and That is, when the system determines that no observable projection compression has occurred, no compensation is applied to any size, and the measured value is output as is, with complete physical logic self-consistency; the overall design goal of the two formulas is to make the reconstructed value after compensation approximate the true physical size that the plug-in kit should present when there is no attitude error, regardless of whether the deflection is positive or negative.

[0077] For example, when the first plug-in port size measurement value Millimeters, spatial deflection angle value When, After eliminating the attitude error, the dimensions of the plug-in port are reconstructed. Millimeters; when the fourth contact point height measurement value Millimeters, contact point height projection sensitivity coefficient At that time, the magnified deflection angle is Spend, The reconstructed height value of the contact point after attitude error elimination The fourth contact point height measurement, which was originally 0.29 mm and exceeded the tolerance of 0.25 to 0.28 mm, was restored to 0.2654 mm after projection compensation.

[0078] S233: Calculate the first absolute difference between the reconstructed size of the plug-in port after attitude error elimination and the preset standard plug-in port size. At the same time, calculate the second absolute difference between the reconstructed height of the contact point after attitude error elimination and the preset standard contact point height. Add the first absolute difference and the second absolute difference to generate a multi-dimensional joint physical consistency deviation value.

[0079] The reconstructed dimensions of the plug-in port and the reconstructed height of the contact point after attitude error elimination output by S232 are two independent scalars, which have not yet been converged into a comprehensive metric that can be directly compared with a single threshold. S3 needs to characterize the overall deformation degree with a single scalar, so this step sums the absolute values ​​of the deviations of the two reconstructed values ​​relative to their respective standard values.

[0080] Specifically, the first absolute difference The absolute value of the difference between the reconstructed socket size after attitude error elimination and the preset standard socket size is calculated using the following expression: ,in To preset the standard plug-in port size, the median value of the plug-in port size tolerance range is taken, which is determined by the plug kit product design specifications. An example value is 0.645 mm.

[0081] Second absolute difference The absolute value of the difference between the reconstructed contact point height after attitude error elimination and the preset standard contact point height is calculated as follows: ,in The standard contact point height is preset, and the median value of the contact point height tolerance range is taken, which is determined by the product design specifications of the plug kit. An example value is 0.265 mm.

[0082] The multi-size joint physical consistency deviation value The sum of the first absolute difference and the second absolute difference is calculated using the following expression: (See below) Figure 7 This is a schematic diagram of the logic structure for multi-size joint physical consistency deviation provided in an embodiment of this application. For example... Figure 7 As shown, the left half compares the reconstructed size of the plug-in port after attitude error elimination with the preset standard plug-in port size to generate a first absolute difference; the right half compares the reconstructed height of the contact point after attitude error elimination with the preset standard contact point height to generate a second absolute difference; the first and second absolute differences are converged and added together to obtain the multi-dimensional joint physical consistency deviation value at the bottom. This logic structure linearly accumulates the residual deviations dispersed on the two dimensions of plug-in port size and contact point height into a single scalar, enabling the multi-dimensional joint physical consistency deviation value to approach zero under pure attitude error but significantly increase under actual deformation, providing a deformation metric after removing attitude distortion for the S3 determination.

[0083] The design principle of the multi-dimensional joint physical consistency deviation value expression is as follows: the first absolute difference reflects the residual deviation of the insertion port size relative to the standard after deducting attitude distortion, and the second absolute difference reflects the residual deviation of the contact point height relative to the standard after deducting attitude distortion. The absolute values ​​of the two are taken because whether the reconstructed value is too large or too small, it represents the actual deformation. The two are added instead of taking the maximum value because the actual processing defects of the insertion kit will generate residual deviations in multiple dimensions along the mechanical coupling relationship. Summation can linearly accumulate the deformation evidence dispersed in different dimensions, so that the linkage deformation is fully included. The overall design goal of this expression is to make both residuals approach zero under pure attitude error, so that the multi-dimensional joint physical consistency deviation value approaches zero, while at least one residual under actual deformation is significantly increased, so that the multi-dimensional joint physical consistency deviation value is significantly increased.

[0084] For example, following the compensation result of S232, the size value of the plug-in port is reconstructed after the attitude error is eliminated. Millimeters, first absolute difference Millimeters; Contact point reconstructed height value after attitude error elimination millimeters, second absolute difference Millimeters; Multi-size joint physical consistency deviation value The millimeter indicates that after deducting a spatial deflection of approximately 6.09 degrees, all dimensions of the insert kit have returned to standard, and the out-of-tolerance height of the contact point is actually caused by attitude projection distortion rather than actual deformation.

[0085] Step S23 uses the measured width at the bend to infer the spatial deflection angle value, and then performs projection compensation on the plug-in port size and contact point height. Its internal mechanism calculates the deflection angle based solely on the width compression observed at the bend in the image, without making any assumptions about the physical source of the deflection. Based on this mechanism, it can be inferred that when the tilt of the plug-in assembly image is not due to the pose fluctuation of the plug-in assembly itself on the conveyor belt, but rather to hidden hardware problems such as mechanical vibration, belt end jump, or minor tooling loosening caused by long-term operation of the detection conveyor belt, these hardware problems are also manifested as the compression of the width at the bend in the image. Therefore, the same angle calculation based on the width at the bend will include the tilt induced by this hardware in the spatial deflection angle value, and the projection compensation will cancel it out in the digital plane. Therefore, in addition to its original goal of eliminating the attitude error of the plug-in kit itself, this design naturally gains the ability to structurally avoid hardware-induced false out-of-tolerance: without adding any additional modules, the system automatically absorbs the projection distortion superimposed by belt vibration or tooling loosening, overcoming the hidden defect of fixed single-point threshold logic that is forced to frequently stop for calibration or batch misjudgment due to batch false out-of-tolerance when there is a slight drift in the hardware tooling. This defect is usually not associated with the attitude projection compensation mechanism by those skilled in the art.

[0086] S3: Identify the single-point out-of-tolerance trigger type based on the multi-dimensional joint physical consistency deviation value, and output a qualified status mark or pneumatic rejection control signal to the pneumatic rejection valve or system display screen.

[0087] This step addresses the problem of existing independent single-point dimensional threshold verification logic directly classifying false single-point deviations caused by attitude errors as true dimensional defects, leading to the erroneous rejection of qualified kits. The fundamental mechanism lies in the fact that relying solely on whether a single dimension exceeds its limits cannot identify the cause of the deviation. In contrast, the multi-dimensional joint physical consistency deviation value obtained from S233 has already removed attitude projection distortion and specifically carries true deformation information. To translate this discriminative capability into specific control of the pneumatic rejection valve, this step first sets geometric tolerance limits and consistency constraint boundaries, then performs joint logic voting on single-point deviation events and multi-dimensional joint physical consistency deviation values, outputting either a pass / fail status flag or a pneumatic rejection control signal.

[0088] Further, step S3 includes: S31: Set the upper limit of the consistency judgment threshold and the limit of individual tolerances in the control kernel of the verification system.

[0089] Subsequent steps S32 to S34 require both a mathematically defined geometric out-of-bounds boundary and a consistency out-of-bounds boundary representing rigid body mechanical constraints as criteria in order to classify the causes of out-of-bounds errors. However, the geometric tolerance alone cannot answer the question of "how much residual belongs to the purely attitude-interpretable range". Therefore, this step must set two types of boundaries in the control kernel: a single-item tolerance limit and a consistency judgment threshold upper limit.

[0090] Specifically, the individual tolerance limits are the upper and lower limits of the tolerance range of each of the four physical dimensions, which are directly determined by the product design specifications of the plug kit, including the plug-in port size of 0.63 to 0.66 mm, the bending width of 5.27 to 5.37 mm, the hole size of 0.95 to 1.05 mm, and the contact point height of 0.25 to 0.28 mm.

[0091] The upper limit of the consistency determination threshold The determination method is as follows: Select no less than 100 insert kits whose dimensions are all known to be qualified. Image each kit one by one within the normal attitude fluctuation range of the conveyor belt and calculate the multi-dimensional joint physical consistency deviation value of each kit according to S231 to S233. Take the mean of the multi-dimensional joint physical consistency deviation value sample plus 3 times the standard deviation as the upper limit of the consistency judgment threshold. For example, the upper limit of the consistency judgment threshold is... The value is 0.03 mm. The physical basis of this determination method is: it is known that the residual of a qualified insert kit after projection compensation under pure attitude fluctuation is contributed only by measurement noise and compensation error. The deviation value corresponding to its mean plus 3 times the standard deviation represents the upper limit of the multi-size joint physical consistency deviation value that pure attitude error can produce in a statistical sense. Deviations exceeding this upper limit must contain real deformation components.

[0092] The control kernel of the verification system loads the individual tolerance limits and the upper limit of the consistency judgment threshold together and keeps them resident in memory for S32 to S34 to call.

[0093] The individual tolerance limits are derived from the product's geometric design and consist of allowable dimensional ranges defined by pure mathematics. The upper limit of the consistency judgment threshold is derived from the residual statistics of qualified parts under actual posture fluctuations and consists of the natural upper limit of rigid mechanical constraints on the compensated residuals. In scenarios where the posture error of the insert kit and the actual manufacturing defect are similar in single-point data and cannot be distinguished by geometric tolerance alone, relying solely on the individual tolerance limits can only answer whether the dimensions have exceeded the limits, but cannot answer whether the exceedance is caused by posture. The upper limit of the consistency judgment threshold precisely uses the residual distribution of qualified parts as a benchmark to quantify the physical boundary of "pure posture-explainable residuals," allowing S32 to S34 to cross-reference mathematical geometric exceedances with rigid mechanical constraint exceedances. This provides a physical classification standard for the classification of posture defects and manufacturing defects that is directly bound to the residual statistics of this production line and the tolerance specifications of this product.

[0094] S32: The control kernel of the verification system reads the third hole size measurement value, determines whether the third hole size measurement value exceeds the single tolerance limit, and at the same time determines whether the multi-dimensional joint physical consistency deviation value is greater than the upper limit of the consistency judgment threshold.

[0095] S21 has read out the original measurement values ​​from the dataset, and S233 has generated the multi-dimensional joint physical consistency deviation value. However, the two are respectively above the single-point geometric datum and the multi-point compensation residual datum, and their dimensions and physical meanings are different. They cannot be directly merged within the same criterion. Therefore, this step must first compare the original single-point measurement value with the single-item tolerance limit and the multi-dimensional joint physical consistency deviation value with the upper limit of the consistency judgment threshold to generate Boolean judgment results respectively.

[0096] Specifically, the control kernel of the verification system reads the third hole size measurement value of the current fitting kit and determines whether the third hole size measurement value falls outside the single tolerance limit of 0.95 to 1.05 mm for hole size; the hole size is approximately unchanged under slight spatial deflection, so its out-of-bounds is an independent geometric out-of-bounds index that is not affected by attitude.

[0097] The control kernel of the verification system simultaneously checks the first measurement value of the insertion port size, the second measurement value of the bend width, and the fourth measurement value of the contact point height of the current fitting kit against their respective individual tolerance limits to determine if any of these measurements exceeds the corresponding individual tolerance limit. If any of these measurements exceeds the corresponding individual tolerance limit, the control kernel of the verification system sets the individual tolerance trigger flag to 1; otherwise, it sets it to 0.

[0098] Since the projection of the third hole size measurement remains approximately unchanged under slight spatial deflection, its deviation is not affected by attitude projection distortion and is an independent geometric out-of-tolerance indicator. The control kernel of the calibration system additionally sets a separate independent hole diameter out-of-tolerance flag for the third hole size measurement: if the third hole size measurement exceeds the single-item tolerance limit of 0.95 to 1.05 mm, the independent hole diameter out-of-tolerance flag is set to 1; if the third hole size measurement does not exceed the single-item tolerance limit, the independent hole diameter out-of-tolerance flag is set to 0. The independent hole diameter out-of-tolerance flag and the consistency deviation out-of-limit flag are independent of each other and are not affected by the multi-dimensional joint physical consistency deviation value.

[0099] The control kernel of the verification system compares the multi-size joint physical consistency deviation value of the current plug-in kit with the upper limit of the consistency judgment threshold. If the multi-size joint physical consistency deviation value is greater than the upper limit of the consistency judgment threshold, the consistency deviation exceeding the limit flag is set to 1; otherwise, it is set to 0.

[0100] The control kernel of the verification system submits the single-item out-of-tolerance trigger flag, the consistency deviation out-of-limit flag, and the aperture independent out-of-tolerance flag to S33 and S34 for joint voting.

[0101] S33: Specifically, if a single dimension exceeds the single tolerance limit, but the combined physical consistency deviation of multiple dimensions is less than or equal to the upper limit of the consistency judgment threshold, and the third hole size measurement value does not exceed the single tolerance limit, then the current out-of-tolerance is determined to be a false out-of-tolerance caused by the detection posture error, and a qualified status mark is output to the system display screen and the insert kit is allowed to continue to flow on the conveyor belt.

[0102] If a single-point out-of-tolerance condition is not identified in conjunction with the combined physical consistency deviation of multiple dimensions, false single-point out-of-tolerance caused by attitude projection distortion will be directly judged as defective and rejected. Qualified inserts will be mistakenly judged as defective, leading to an increase in the error rejection rate and production costs. Therefore, this step adds a consistency not exceeding limit condition to identify false out-of-tolerance, provided that a single-point out-of-tolerance condition is met. Since the projection of the third item, hole size measurement, is approximately unchanged under slight spatial deflection, its exceeding the limit must originate from a real geometric defect rather than attitude projection distortion, and therefore cannot be included in the false out-of-tolerance exemption range. Thus, a condition of 0 for the independent hole diameter out-of-tolerance flag is also added to ensure that real defects in hole size are not mistakenly judged as false out-of-tolerance and released.

[0103] Specifically, when the single-item out-of-tolerance trigger flag submitted by S32 is 1, the consistency deviation out-of-limit flag is 0, and the aperture independent out-of-tolerance flag is 0, that is, when a certain single dimension exceeds the single-item tolerance limit, the multi-dimensional joint physical consistency deviation value is less than or equal to the upper limit of the consistency judgment threshold, and the third item aperture size measurement value does not exceed the single-item tolerance limit, the control kernel of the verification system determines that the current out-of-tolerance is a false out-of-tolerance caused by the detection attitude error.

[0104] The control kernel of the verification system outputs a pass status mark to the system display screen, but does not send any rejection signal to the pneumatic rejection valve, allowing the current insertion kit to continue flowing on the conveyor belt to the next process.

[0105] For a single out-of-tolerance trigger flag of the plug-in kit, the control kernel of the verification system directly outputs a qualified status flag to the system display screen.

[0106] For example, following the numerical examples in S231 to S233, the current fitting kit's fourth contact point height measurement value of 0.29 mm exceeds the single-item tolerance limit of 0.25 to 0.28 mm, causing the single-item out-of-tolerance trigger flag to be 1; while its multi-dimensional joint physical consistency deviation value is 0.0008 mm, which is less than the upper limit of the consistency judgment threshold of 0.03 mm, causing the consistency deviation out-of-limit flag to be 0; at the same time, the third hole size measurement value does not exceed the single-item tolerance limit of 0.95 to 1.05 mm, and the hole diameter independent out-of-tolerance flag is 0; based on this, the control kernel of the verification system determines that the contact point height out-of-tolerance is a false out-of-tolerance caused by a spatial deflection of approximately 6.09 degrees, outputs a qualified status mark and releases the kit, thus preventing the qualified fitting kit from being mistakenly rejected.

[0107] S34: If a single dimension exceeds the single tolerance limit, and the combined physical consistency deviation of multiple dimensions is greater than the upper limit of the consistency judgment threshold, or the third hole size measurement value exceeds the single tolerance limit, then it is determined that the real physical deformation triggering condition is met, a pneumatic rejection control signal is generated and sent to the pneumatic rejection valve, driving the pneumatic rejection valve to physically isolate and reject the current insert.

[0108] Under actual manufacturing defects, physical deformation of a certain dimension will leave residual deviations in the insertion / removal port size and contact point height along the rigid body mechanical coupling relationship, which cannot be explained by a single spatial deflection. After the accumulation of S233, the combined physical consistency deviation value of multiple dimensions is raised to exceed the upper limit of pure attitude interpretation. Therefore, when a single point out-of-tolerance and consistency exceeding the limit are both met, the actual physical deformation can be determined. Since the out-of-tolerance of the third term, the hole size measurement value, is not affected by attitude projection distortion, its exceeding of the single-term tolerance limit constitutes an independent path for triggering the actual physical deformation condition, without relying on the coordination of the combined physical consistency deviation value of multiple dimensions.

[0109] Specifically, when the single-item out-of-tolerance trigger flag submitted by S32 is 1 and the consistency deviation exceeding the limit flag is 1, that is, when a single dimension exceeds the single-item tolerance limit and the combined physical consistency deviation value of multiple dimensions is greater than the upper limit of the consistency judgment threshold, the control kernel of the verification system determines that the current fitting meets the real physical deformation trigger condition; or when the aperture independent out-of-tolerance flag is 1, that is, when the third aperture size measurement value exceeds the single-item tolerance limit, the control kernel of the verification system also determines that the current fitting meets the real physical deformation trigger condition. See also Figure 8 This is a schematic diagram of the static module interface for joint logic voting provided in an embodiment of this application. For example... Figure 8As shown, the three nodes—single-item out-of-tolerance trigger flag, consistency deviation exceeding limit flag, and aperture independent out-of-tolerance flag—are connected to the central real physical deformation trigger condition block via wiring. One output of this block is connected to the system display screen via a bus, outputting a pass / fail status mark; the other output is connected downwards to the pneumatic rejection valve, outputting a pneumatic rejection control signal. This module interface corresponds to a joint voting logic that determines the real physical deformation trigger condition is met when the single-item out-of-tolerance trigger flag is 1 and the consistency deviation exceeding limit flag is 1, or the aperture independent out-of-tolerance flag is 1. This ensures that the pneumatic rejection control signal is generated only when the out-of-bounds trigger and the real deformation determination are both met, or when the aperture size independently exceeds the limit. This directly presents the drive transmission relationship of the three determination conditions to the downstream pneumatic rejection valve.

[0110] The control kernel of the verification system generates a pneumatic rejection control signal and sends the pneumatic rejection control signal to the pneumatic rejection valve at the end of the conveyor belt via the industrial communication bus.

[0111] Upon receiving a pneumatic rejection control signal, the pneumatic rejection valve opens the high-pressure air circuit, ejecting high-pressure airflow from the nozzle to blow the currently inserted kit away from the conveyor belt's flow path, completing physical isolation and rejection. (See also...) Figure 9 This is a schematic diagram of a physical isolation and exclusion scenario provided in an embodiment of this application. For example... Figure 9 As shown, the conveyor belt is marked with a flow path. A pneumatic rejection valve is located above the rejection station at the end of the conveyor belt, with its high-pressure air nozzle facing the flow path of the insert on the conveyor belt. In the figure, the current insert is rotated off the track under the action of high-pressure airflow, blown away from the flow path, and falls into the physical isolation rejection area marked by the dashed box in the lower right corner. This scenario corresponds to the action process of the pneumatic rejection valve opening the high-pressure air path after receiving the pneumatic rejection control signal, and spraying high-pressure airflow from the nozzle to blow the current insert away from the conveyor belt flow path. This process transforms the judgment result of the verification system control kernel into the physical isolation rejection of the unqualified insert.

[0112] For example, when electroplated foreign matter adheres to the insertion port of a certain connector, and the first dimension measurement of the insertion port is obstructed at 0.60 mm, exceeding the single tolerance limit of 0.63 to 0.66 mm, the single tolerance trigger flag is 1; at this time, the second dimension measurement of the bend width is 5.31 mm. Solve for the spatial deflection angle value Degree, after eliminating attitude error, the reconstructed dimensions of the plug-in port. Millimeters, first absolute difference Millimeters; the fourth item, the contact point height measurement, is 0.271 millimeters, and the reconstructed contact point height value after attitude error elimination. millimeters, second absolute difference Millimeters; Multi-size joint physical consistency deviation value The first measurement is 0.03 mm, which is greater than the upper limit of the consistency judgment threshold. The consistency deviation exceeds the limit and the flag is 1. The third measurement, the hole size, does not exceed the single tolerance limit of 0.95 to 1.05 mm. The hole diameter exceeds the independent tolerance flag and the flag is 0. Based on this, the control kernel of the verification system determines that the real physical deformation trigger condition is met through the dual-condition joint path of the single tolerance trigger flag and the consistency deviation exceeds the limit flag. It then generates a pneumatic rejection control signal and drives the pneumatic rejection valve to reject the insert.

[0113] For example, when a fitting has a reduced through-hole diameter due to stamping burrs, and the third hole size measurement is 0.92 mm, exceeding the single tolerance limit of 0.95 to 1.05 mm, the hole diameter independent out-of-tolerance flag is set to 1. Regardless of whether the multi-dimensional joint physical consistency deviation value exceeds the upper limit of the consistency judgment threshold, the control kernel of the verification system determines the true physical deformation trigger condition through the independent path of the hole diameter independent out-of-tolerance flag, generates a pneumatic rejection control signal, and drives the pneumatic rejection valve to reject the fitting, thus avoiding the true hole size defect being mistakenly judged as a false out-of-tolerance release because the residual compensation of the insertion port and contact point height is just small.

[0114] The single-item out-of-tolerance judgment in step S32, the consistency judgment in steps S33 and S34 which are based on the multi-dimensional joint physical consistency deviation value, and the independent geometric out-of-bounds judgment represented by the aperture independent out-of-tolerance flag together constitute a set of three-flag joint logical voting. The three judgment conditions target different physical levels and act on different links in the judgment chain. The single-item out-of-tolerance judgment acts on the original single-point measurement value and is responsible for capturing the triggering event of size out-of-bounds. The consistency judgment acts on the compensation residual after deducting attitude projection distortion and is responsible for determining the cause attribute of the out-of-bounds event at the level of plug-in port size and contact point height. The aperture independent out-of-tolerance flag judgment is specifically for the aperture size index that is not affected by attitude and is responsible for extracting the real aperture size defect from the false out-of-tolerance exemption mechanism and directly triggering its removal. If only a single-item out-of-tolerance judgment is performed without a consistency judgment, then it is impossible to distinguish between false single-point out-of-tolerance caused by attitude projection distortion and single-point out-of-tolerance caused by real deformation. Qualified inserts will be rejected along with false out-of-tolerance, degenerating into the independent single-point threshold logic described in the background problem. If only a consistency judgment is performed without a single-item out-of-tolerance judgment, then although the consistency judgment can reflect the overall residual deformation, it cannot locate which specific dimension has exceeded the limit, nor can it provide a definite trigger for boundary situations where the residual is just close to the threshold and a certain dimension has clearly exceeded the limit. The judgment result cannot be directly mapped to the rejection action of a specific insert. If an independent out-of-tolerance mark judgment for aperture is missing, then the real defect in aperture size will be misjudged as a false out-of-tolerance and released by S33 when the residual of the insertion port and contact point height compensation is just small, which contradicts the specification's characterization of aperture size as "not affected by attitude, and exceeding the limit is a real geometric defect." The structural coupling of the three conditions makes the classification results of "false attitude out-of-tolerance" and "real manufacturing defects" logically irreducible to any single condition. The joint logic voting mechanism drives the downstream pneumatic rejection valve in the following way: a pneumatic rejection control signal is generated only when the over-limit trigger and the actual deformation determination are both established simultaneously, or when the orifice size exceeds the limit independently. In the hidden scenario where tiny electroplated foreign objects or electroplating nodules occasionally adhere to the insertion port of the fitting, the insertion port size exceeds the limit due to foreign object obstruction, but without corresponding compression of the bending width. The calculated spatial deflection angle value is too small, and the projection compensation cannot restore the over-limit. The multi-dimensional joint physical consistency deviation value remains high. Based on this, the system distinguishes between foreign object obstruction and simple attitude tilt and accurately triggers rejection. The above distinction can be completed with only one-dimensional pneumatic control without the need to add three-dimensional laser scanning hardware, ensuring the electrical connection reliability of the fittings shipped from the factory.

[0115] This embodiment uses an industrial camera and a trigger photoelectric sensor to acquire frame-by-frame image sequences of the insert kit online. These images are then converted by a dimensional measuring instrument and aligned according to the hardware arrival timestamp to form a multi-point physical dimension dataset for the insert kit. Based on this, the spatial deflection angle is calculated from the width measurement value at the bend by utilizing the projection distortion law of the inherent rigid body geometry of the insert kit during spatial deflection. Projection compensation is then performed on the insertion port dimension measurement value and the contact point height measurement value to restore the true physical dimensions, which are then aggregated into a multi-dimensional joint physical consistency deviation value. Finally, a three-marker joint logic voting system consisting of a single-item tolerance limit, a consistency judgment threshold upper limit, and an independent aperture out-of-tolerance mark is used to accurately separate false single-point out-of-tolerance caused by posture from multi-point linkage deformation caused by real processing defects. The identification results are directly converted into a one-dimensional pneumatic control action for the pneumatic rejection valve. This effectively prevents qualified insert kits from being misjudged and rejected without adding 3D scanning hardware, while ensuring that real defects in aperture size are not missed due to accidental underestimation of compensation residuals.

[0116] Example 2: This embodiment, based on Embodiment 1, provides a multi-size point consistency verification system for insert kits, such as... Figure 10 As shown, it includes: Data acquisition and alignment conversion module: used to acquire frame-by-frame image sequences of the interposer kit under online motion conditions through an industrial camera and trigger photoelectric sensor, and output a multi-point physical dimension dataset of the interposer kit after conversion by a dimensional measuring instrument; Projection distortion elimination and deviation calculation module: used to eliminate spatial attitude projection distortion using the multi-point physical dimension dataset of the plug kit output by the dimension measuring instrument, and calculate and generate multi-dimensional joint physical consistency deviation value; Trigger type identification and control output module: used to identify the single-point out-of-tolerance trigger type based on the multi-size joint physical consistency deviation value, and output a qualified status mark or pneumatic rejection control signal to the pneumatic rejection valve or system display screen.

Claims

1. A method for preventing false judgments in multi-size point consistency verification of fitting kits, characterized in that, include: The image sequence of the interposer under online motion is acquired frame by frame using an industrial camera and a trigger photoelectric sensor, and then converted and output as a multi-point physical dimension dataset of the interposer by a dimensional measuring instrument. Spatial attitude projection distortion is eliminated using the multi-point physical dimension dataset of the plug kit output by the dimensional measuring instrument, and a multi-dimensional joint physical consistency deviation value is calculated and generated. Based on the multi-size joint physical consistency deviation value, the single-point out-of-tolerance trigger type is identified, and a qualified status mark or pneumatic rejection control signal is output to the pneumatic rejection valve or system display screen.

2. The method for preventing false judgments in multi-size point consistency verification of fittings according to claim 1, characterized in that, The step involves acquiring frame-by-frame image sequences of the insert kit under online motion conditions using an industrial camera and a trigger photoelectric sensor, and then converting and outputting a multi-point physical dimension dataset of the insert kit via a dimensional measuring instrument. This dataset includes: As the insert kit passes through the conveyor belt and triggers the photoelectric sensor, the industrial camera is controlled to capture optical images of the insert kit surface at a fixed exposure time, generating a frame-by-frame image sequence of the insert kit. The image sequence of the insert kit is acquired frame by frame and input into the size measuring instrument. The pixel calibration function of the size measuring instrument is used to extract the physical feature edges. The size measuring instrument is used to extract four key feature dimensions of the insert kit on the two-dimensional projection plane, including the first measurement value of the insertion port size, the second measurement value of the bend width, the third measurement value of the hole size, and the fourth measurement value of the contact point height. The first measurement value of the plug-in port size, the second measurement value of the bend width, the third measurement value of the hole size, and the fourth measurement value of the contact point height are aligned and stitched together in the built-in storage module of the dimension measuring instrument according to the hardware arrival timestamp, and the multi-point physical dimension dataset of the plug kit is output.

3. The method for preventing false judgments in multi-size point consistency verification of fittings according to claim 2, characterized in that, The step of controlling the industrial camera to capture optical images of the insert's surface at a fixed exposure time when the insert passes through the conveyor belt and triggers the photoelectric sensor, and generating a frame-by-frame image sequence of the insert, includes: The trigger photoelectric sensor outputs a rising edge trigger level when the insert kit blocks the through-beam optical path, and the hardware clock of the control core of the verification system adds an incrementing hardware arrival timestamp to the rising edge since the system starts. After receiving the trigger level, the industrial camera immediately performs a single-frame exposure imaging of the insert surface with a fixed exposure time. The upper limit of the fixed exposure time is calculated using the allowable number of motion blur pixels, the physical size corresponding to a single pixel, and the conveyor belt linear speed. A sequence of grayscale image frames is generated, arranged in order of the hardware arrival timestamps. Each frame carries the corresponding hardware arrival timestamp, and the frames are used as the interpolation kit to acquire image sequences one by one.

4. The method for preventing false judgments in multi-size point consistency verification of insert kits according to claim 2, characterized in that, The step of aligning and stitching the first measured value of the insertion port size, the second measured value of the bend width, the third measured value of the hole size, and the fourth measured value of the contact point height in the built-in storage module of the dimensional measuring instrument according to the hardware arrival timestamp, and outputting the multi-point physical dimension dataset of the insertion kit includes: The size measuring instrument reads the hardware arrival timestamp inherited by each of the four key feature dimensions, and determines the measurement results of multiple corresponding features with equal hardware arrival timestamps as the same insert kit. The size measuring instrument concatenates four key feature dimensions with the same hardware arrival timestamp into a quintuple record that sequentially contains the corresponding hardware time and each dimension. The dimensional measuring instrument writes each quintuple record into the built-in storage module in ascending order of hardware arrival timestamp, and outputs a multi-point physical dimension dataset of the plug kit arranged in sequence.

5. The method for preventing false judgments in multi-size point consistency verification of fittings according to claim 1, characterized in that, The process of eliminating spatial attitude projection distortion by using the multi-point physical dimension dataset of the fitting kit output by the dimensional measuring instrument, and calculating the multi-dimensional joint physical consistency deviation value includes: The first item of the insertion port size measurement value, the second item of the bend width measurement value, the third item of the hole size measurement value, and the fourth item of the contact point height measurement value are read from the multi-point physical size dataset of the insert kit output by the size measuring instrument. Based on the inherent spatial rigid body geometry of the plug-in kit, the two-dimensional projection variation of four physical dimensions under small spatial tilting attitude errors is calculated. Based on the projection distortion law of the inherent rigid body geometry of the plug kit when it is deflected in space, the spatial deflection angle value is inversely calculated using the width measurement value at the second bend, and the projection compensation is performed on the first plug-in port size measurement value and the fourth contact point height measurement value to generate the multi-size joint physical consistency deviation value.

6. The method for preventing false judgments in multi-size point consistency verification of insert kits according to claim 5, characterized in that, The projection distortion law based on the inherent rigid body geometry of the insert kit when it deflects in space is used to inversely deduce the spatial deflection angle value using the second item of the width measurement at the bend, and projection compensation is performed on the first item of the insert / pull-out size measurement value and the fourth item of the contact point height measurement value to generate the multi-dimensional joint physical consistency deviation value, including: Based on the deviation ratio between the second measured value of the bend width and the preset standard bend width, the spatial deflection angle value of the current insert is calculated; The projection length is compensated by using the cosine function relationship corresponding to the spatial deflection angle value, and the reconstructed size value of the plug-in port and the reconstructed height value of the contact point after attitude error elimination are generated. Calculate the first absolute difference between the reconstructed size of the plug-in port after attitude error elimination and the preset standard plug-in port size; calculate the second absolute difference between the reconstructed height of the contact point after attitude error elimination and the preset standard contact point height; add the two absolute differences to generate a multi-dimensional joint physical consistency deviation value.

7. The method for preventing false judgments in multi-size point consistency verification of insert kits according to claim 6, characterized in that, The step of using the cosine function relationship corresponding to the spatial deflection angle value to perform projection length compensation, and generating the reconstructed size value of the plug-in port and the reconstructed height value of the contact point after attitude error elimination, includes: For the first measured value of the plug-in port size, the value is stretched and restored by dividing by the cosine of the deflection angle, and the value is equal to the value after the attitude error is eliminated by dividing by the cosine of the spatial deflection angle. For the fourth contact point height measurement value, it is compressed and restored by multiplying by the amplified deflection angle cosine, and the resulting value is equal to the contact point reconstructed height value after attitude error elimination by multiplying by the amplified deflection angle cosine.

8. The method for preventing false judgments in multi-size point consistency verification of fittings according to claim 1, characterized in that, The step of identifying the single-point out-of-tolerance trigger type based on the multi-dimensional joint physical consistency deviation value and outputting a qualified status mark or pneumatic rejection control signal to the pneumatic rejection valve or system display screen includes: Set the upper limit of the consistency judgment threshold and the individual tolerance limit in the control kernel of the verification system; Read the third hole size measurement value, determine whether the measurement value exceeds the single tolerance limit, and at the same time determine whether the physical consistency deviation value of the multi-size combination is greater than the upper limit of the consistency judgment threshold; If a single dimension exceeds the single tolerance limit, but the combined physical consistency deviation of multiple dimensions is less than or equal to the upper limit of the consistency judgment threshold, and the third dimension, the hole size measurement value, does not exceed the limit, then a qualified status mark is output and the process is allowed to continue. If a single dimension exceeds the single tolerance limit and the combined physical consistency deviation of multiple dimensions exceeds the upper limit of the consistency judgment threshold, or if the third dimension, the hole size measurement value, exceeds the limit, a pneumatic rejection control signal is issued to drive the pneumatic rejection valve to perform physical isolation rejection.

9. The method for preventing false judgments in multi-size point consistency verification of fittings according to claim 8, characterized in that, The process of reading the third hole size measurement value and determining whether the measurement value exceeds the individual tolerance limit, while simultaneously determining whether the combined physical consistency deviation value of multiple dimensions is greater than the upper limit of the consistency determination threshold, includes: If any of the following measurements—the first (insert port size), the second (bend width), the third (hole size), and the fourth (contact point height)—exceeds the corresponding individual tolerance limit, then the individual tolerance trigger flag is set to 1; otherwise, it is set to 0. If the measured value of the third hole size exceeds the single tolerance limit, the hole diameter independent out-of-tolerance flag is set to 1; otherwise, it is set to 0. If the multi-size joint physical consistency deviation value is greater than the upper limit of the consistency judgment threshold, then the consistency deviation exceeding the limit flag is set to 1; otherwise, it is set to 0.

10. The method for preventing false judgments in multi-size point consistency verification of fittings according to claim 8, characterized in that, Setting the upper limit of the consistency judgment threshold and the individual tolerance limit in the control kernel of the verification system includes: The individual tolerance limits are the upper and lower limits of the allowable tolerance range for all physical dimensions pre-defined based on the product specifications; Select a fixed number of insert kits whose dimensions are known to be qualified, image each kit one by one within the normal attitude fluctuation range of the conveyor belt, and calculate the sample multi-dimensional joint physical consistency deviation value corresponding to each insert kit. Extract the statistical mean and standard deviation of the corresponding deviation values ​​of all samples, obtain the value calculated by adding a preset multiple of the statistical mean to the standard deviation, and store and retrieve it as the upper limit of the consistency judgment threshold.

11. The method for preventing false judgments in multi-size point consistency verification of insert kits according to claim 6, characterized in that, The spatial deflection angle value of the current fitting is calculated based on the deviation ratio between the measured width at the second bend and the preset standard bend width, including: Extract the ratio of the measured width at the second bend to the preset standard bend width, and use it as the corresponding deviation ratio; Compare the constant 1 with the calculated deviation ratio, and take the smaller value of the two as the cutoff ratio. An inverse cosine function is applied to the cutoff ratio, and the resulting inverse cosine angle is output as the spatial deflection angle value of the current insert kit for use in the next processing stage.

12. A system for preventing false judgments in multi-size point consistency verification of fitting components, used to implement the method for preventing false judgments in multi-size point consistency verification of fitting components as described in any one of claims 1 to 11, characterized in that, The system includes: The data acquisition and alignment conversion module is used to acquire frame-by-frame image sequences of the interposer under online motion conditions through an industrial camera and a trigger photoelectric sensor, and then convert and output a multi-point physical dimension dataset of the interposer through a dimensional measuring instrument. The projection distortion elimination and deviation calculation module is used to eliminate spatial attitude projection distortion using the multi-point physical dimension dataset of the plug kit output by the dimension measuring instrument, and calculate and generate a multi-dimensional joint physical consistency deviation value. The trigger type identification and control output module is used to identify the single-point out-of-tolerance trigger type based on the multi-size joint physical consistency deviation value, and output a qualified status mark or pneumatic rejection control signal to the pneumatic rejection valve or system display screen.

Images