A method for on-line detection and rejection of defective gloves

Abstract

This invention relates to the field of glove manufacturing and discloses a method for online inspection and rejection of defective gloves. The method includes: introducing the demolded finished glove into an inspection carrier and circumferentially sealing the wrist opening to form a cavity to be inspected; using a segmented support core to segment and shape each finger and palm area, so that the finished glove forms a preset inspection posture; sequentially applying pre-charge, stabilizing, pulsed pressurization, and pressure attenuation, and acquiring reflection images, transmission images, deformation response images, and cavity pressure attenuation data respectively; extracting surface defect features, abnormal light transmission features, abnormal pressure response features, and sealing defect features based on various inspection data, and performing fusion judgment; binding the inspection results with carrier identification information, inspection position, and conveying cycle time, and performing secondary identification confirmation and timing correction before the rejection station to achieve rejection of defective products or diversion of verified products. This invention can improve the ability to identify latent defects and the accuracy of rejecting defective products.

Description

Technical Field

[0001] This invention relates to the field of online quality inspection of finished gloves, specifically to a method for online inspection and rejection of defective finished gloves. Background Technology

[0002] After dipping, solidification, vulcanization, drying, and demolding, glove production lines typically require online quality inspection of the finished products. Defective items are promptly separated from the continuous conveyor path to prevent them from entering subsequent counting, packaging, or boxing processes. Because finished gloves are flexible, thin-walled products, their shape after demolding is easily affected by their own weight, residual stress, and conveyor disturbances. Therefore, the finished product inspection process not only relates to surface appearance quality but also directly impacts the judgment of sealing performance and the accuracy of subsequent rejection.

[0003] In existing technologies, a common approach is to visually sample the finished product in its naturally drooping state directly after the gloves are demolded, and identify anomalies such as damage, stains, missing edges, or adhesion based on the images. Another approach involves placing the gloves on the inspection piece and inflating or pressurizing them to determine if there are any leaks. After inspection, abnormal gloves are removed from the conveyor path using air blowing, pushing, or diversion mechanisms. While these approaches can achieve online screening to some extent, the methods for establishing the inspection status, displaying defects, and corresponding the inspection results with the rejection actions remain relatively rudimentary.

[0004] However, existing structures mostly complete the inspection when the gloves are hanging naturally or simply inflated. It is difficult to form a consistent inspection boundary in the finger, palm, and wrist areas of the gloves. During use, local folds, wrinkles, or uneven stress may occur, making it difficult to fully detect micropores, weak areas, and local adhesion areas. At the same time, if the rejection action is triggered only at a fixed rhythm after the inspection is completed, fluctuations in the conveyor speed or deviations in the carrier position can easily disrupt the correspondence between the inspected and rejected items, ultimately affecting the integrity of the inspection and the accuracy of rejecting non-conforming products. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a method for online inspection and rejection of defective finished gloves, thereby resolving the technical problems existing in the prior art.

[0006] The above-mentioned technical objective of the present invention is achieved through the following technical solution: A method for online inspection and rejection of defective finished gloves includes the following steps: S1: The demolded finished glove is transported to the testing carrier. The wrist opening of the finished glove is guided into the sealing socket through the wrist opening guide structure, and the edge of the wrist opening is circumferentially pressed and sealed using the annular pressing structure to form a cavity to be tested inside the finished glove. At the same time, the carrier identification information of the testing carrier is recorded. S2: Insert the segmented support core into the cavity to be tested and perform the support action. The segmented support core includes a palm support segment and multiple finger support segments corresponding to each finger of the glove. Each finger support segment unfolds independently according to a preset displacement threshold or contact feedback. The palm support segment flattens and supports the palm area so that the finished glove forms a preset testing posture. S3: Apply pre-pressure, stabilization, pulse boosting and pressure holding attenuation sequentially to the cavity to be tested, so that the finished glove eliminates wrinkles and completes initial shaping in the pre-pressure stage, maintains a stable testing state in the stabilization stage, induces abnormal deformation response in weak areas, microporous areas or adhesion areas in the pulse boosting stage, and forms corresponding cavity pressure change data in the pressure holding attenuation stage. S4: During the pressure stabilization phase, reflective and transmissive images of the finished glove are acquired; during the pulse boosting phase, deformation response images at least at two moments are acquired; and during the pressure holding and decay phase, cavity pressure decay data are acquired. S5: Extract apparent defect features from the reflected image, extract light transmission anomaly features from the transmitted image, extract pressure response anomaly features from the deformation response image, extract sealing defect features from the cavity pressure attenuation data, and perform fusion judgment on the apparent defect features, light transmission anomaly features, pressure response anomaly features and sealing defect features according to preset judgment rules, so as to output the detection results of the corresponding finished glove. S6: Bind the detection results with the corresponding vehicle identification information, detection position and conveying cycle to generate target rejection information; S7: When the corresponding detection carrier runs to the rejection station, the detection carrier is identified and confirmed a second time according to the target rejection information, and the target rejection information is corrected according to the second identification and confirmation result. At the corrected target rejection time, the corresponding rejection execution mechanism is controlled to perform diversion processing on the corresponding finished glove.

[0007] Preferably, the detection carrier includes a carrier base, a sealing socket disposed on the carrier base, a positioning and limiting part disposed on the outer periphery of the sealing socket, and a carrier identification part disposed on the carrier base; the wrist-mouth guide structure includes a guide port and a complete port component. S1 specifically includes: expanding and guiding the wrist opening of the finished glove using the inlet, circumferentially tidying the edge of the wrist opening using the straightening component, fitting the wrist opening onto the outer periphery of the sealing socket, and controlling the annular pressing structure to move axially along the sealing socket so that the edge of the wrist opening is clamped between the annular pressing structure and the sealing socket to form a continuous circumferential seal. The carrier identification information is provided by the carrier identification unit.

[0008] Preferably, S2 specifically includes: controlling multiple finger support segments to move one by one toward the inner side of the corresponding finger; stopping the finger support segment from continuing to unfold after any finger support segment reaches the corresponding displacement threshold or receives the corresponding contact feedback; and controlling the palm support segment to compensate and flatten the palm area after each finger support segment has finished unfolding.

[0009] Preferably, S3 specifically includes: in the pre-pressurization stage, inflating the cavity to be tested with a first pressure value; in the pressure stabilization stage, adjusting the pressure in the cavity to be tested to a second pressure value and maintaining it for a predetermined time; in the pulse pressurization stage, increasing the pressure in the cavity to be tested from the second pressure value to a third pressure value and then dropping it back to the second pressure value, wherein the third pressure value is greater than the second pressure value; and in the pressure holding and decay stage, stopping the continued air intake and collecting the pressure decay process in the cavity to be tested. In the above process, pressure data in the chamber to be tested is collected in real time using a pressure sensor, and the intake valve and pressure relief valve are controlled according to the pressure data to perform closed-loop regulation of the pressure in the chamber to be tested.

[0010] Preferably, S4 specifically includes: setting a reflective light source and a reflective imaging unit on the outside of the finished glove to acquire a reflective image of the outer surface of the finished glove; Transmitted light sources and transmission imaging units are set on opposite sides of the finished glove to acquire transmission images of corresponding areas of the finished glove. Images were captured from the palm and back of the hand areas of the finished glove. The reflected and transmitted images are acquired in a time-division triggering manner and stored in association with the vehicle identification information of the same detection vehicle.

[0011] Preferably, step S5 involves extracting abnormal features of the pressure response based on the deformation response image, specifically including: registering deformation response images acquired at least at two times during the pulse boosting phase, calculating the contour change, local bulging, or area change of the corresponding region at different acquisition times, and identifying regions exceeding the response threshold as abnormal regions of the pressure response.

[0012] Preferably, in step S5, the apparent defect features, light transmission abnormal features, pressure response abnormal features and sealing defect features are fused and judged according to preset judgment rules, specifically including: identifying surface defects based on reflection images, identifying light transmission abnormal defects based on transmission images, identifying pressure abnormal defects based on deformation response images, and identifying sealing defects based on cavity pressure attenuation data. When any defect feature meets the corresponding non-conformance judgment condition, the corresponding finished glove will be judged as a non-conformance product; If the defect characteristics meet the preset verification conditions but do not meet the non-conformity judgment conditions, the corresponding finished glove will be judged as a verification product; the remaining finished glove will be judged as a qualified product.

[0013] Preferably, after extracting the apparent defect features, light transmission abnormal features, and pressure response abnormal features, the method further includes: mapping the apparent defect features, light transmission abnormal features, and pressure response abnormal features to the finger area, palm area, or wrist area of ​​the finished glove to generate defect area location information.

[0014] Preferably, the target rejection information generated in S6 includes carrier identification information, defect type, rejection station number and target rejection time, wherein the target rejection time is calculated based on the detection location, conveying cycle time and the running path of the detection carrier to the rejection station; The secondary identification and confirmation of the detection vehicle in S7 specifically includes: reading the vehicle identification information of the detection vehicle before the rejection station, collecting the real-time position information of the detection vehicle, correcting the deviation of the target rejection time according to the vehicle identification information and the real-time position information, and controlling the corresponding rejection execution mechanism to act at the corrected target rejection time.

[0015] Preferably, in step S7, when the test result of the corresponding finished glove is a non-conforming result, the rejection mechanism is controlled according to the defect type and the product is introduced into the non-conforming product collection channel. When the test result of the corresponding finished glove is a review judgment result, the control rejection execution mechanism will import it into the review channel; Within multiple consecutive detection cycles, when the number of occurrences of the same type of defect in the same area reaches the warning threshold, a corresponding process warning message is generated and sent to the front-end production station for adjusting the operating parameters of the corresponding process.

[0016] In summary, the present invention has the following main beneficial effects: This application, by setting up a wrist opening guide structure, a sealing socket, an annular pressing structure, and a segmented support core, ensures that the finished glove, after demolding, undergoes circumferential pressing and sealing of the wrist opening before entering the inspection stage, followed by segmented support shaping of the finger and palm areas. This prevents the finished glove from being in an unstable state of natural drooping, partial folding, or random bulging during inspection, instead forming a preset inspection posture with consistent boundaries, uniform posture, and repeatability. This achieves the effects of providing a unified inspection benchmark for subsequent inspection areas, reducing image fluctuations and misjudgment risks caused by the flexible deformation of the finished product, and improving the inspection consistency of the palm, finger, and wrist areas.

[0017] By sequentially applying pre-pressure, stabilizing pressure, pulsed pressure increase, and pressure attenuation to the chamber under test, and combining this with the acquisition of reflection images, transmission images, deformation response images, and chamber pressure attenuation data, the finished glove can not only detect static surface defects but also reveal dynamic abnormal responses in weak areas, micropore adjacent areas, and local adhesion areas under controlled differential pressure. This further exposes latent defects that are difficult to detect by a single photograph or ordinary inflation leak detection, achieving the effect of simultaneously identifying surface defects, light transmission abnormalities, pressure abnormalities, and sealing defects, improving the detection capability of minute and early defects, and enhancing the overall detection integrity and judgment reliability.

[0018] By binding the detection results with carrier identification information, detection location, and conveying cycle time, and correcting the target rejection time by combining carrier re-identification and real-time position detection before the rejection station, the rejection execution mechanism is controlled to guide non-conforming products into the non-conforming product collection channel and the verification products into the verification channel. At the same time, the same type of defects in the same area within the continuous detection cycle are statistically analyzed and process early warning information is generated. This not only avoids the problems of incorrect rejection, missed rejection, and defective product mixing caused by simply relying on fixed delay control, but also enables the back-end detection results to have a reverse effect on the front-end dipping, drying, demolding, and conveying processes, thereby improving the accuracy of non-conforming product rejection, enhancing the online quality control capability of the production line, and promoting the synergistic optimization of front and back processes. Attached Figure Description

[0019] Figure 1 This is a flowchart of the method of the present invention.

[0020] Figure 2 This is a block diagram of the overall structure of the online inspection and defective product rejection system for finished gloves of the present invention. Detailed Implementation

[0021] 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.

[0022] Example 1 refer to Figure 1 A method for online inspection and rejection of defective finished gloves, comprising the following steps: S1: The demolded finished glove is transported to the testing carrier. The wrist opening of the finished glove is guided into the sealing socket through the wrist opening guide structure, and the edge of the wrist opening is circumferentially pressed and sealed using the annular pressing structure to form a cavity to be tested inside the finished glove. At the same time, the carrier identification information of the testing carrier is recorded. S2: Insert the segmented support core into the cavity to be tested and perform the support action. The segmented support core includes a palm support segment and multiple finger support segments corresponding to each finger of the glove. Each finger support segment unfolds independently according to a preset displacement threshold or contact feedback. The palm support segment flattens and supports the palm area so that the finished glove forms a preset testing posture. S3: Apply pre-pressure, stabilization, pulse boosting and pressure holding attenuation sequentially to the cavity to be tested, so that the finished glove eliminates wrinkles and completes initial shaping in the pre-pressure stage, maintains a stable testing state in the stabilization stage, induces abnormal deformation response in weak areas, microporous areas or adhesion areas in the pulse boosting stage, and forms corresponding cavity pressure change data in the pressure holding attenuation stage. S4: During the pressure stabilization phase, reflective and transmissive images of the finished glove are acquired; during the pulse boosting phase, deformation response images at least at two moments are acquired; and during the pressure holding and decay phase, cavity pressure decay data are acquired. S5: Extract apparent defect features from the reflected image, extract light transmission anomaly features from the transmitted image, extract pressure response anomaly features from the deformation response image, extract sealing defect features from the cavity pressure attenuation data, and perform fusion judgment on the apparent defect features, light transmission anomaly features, pressure response anomaly features and sealing defect features according to preset judgment rules, so as to output the detection results of the corresponding finished glove. S6: Bind the detection results with the corresponding vehicle identification information, detection position and conveying cycle to generate target rejection information; S7: When the corresponding detection carrier runs to the rejection station, the detection carrier is identified and confirmed a second time according to the target rejection information, and the target rejection information is corrected according to the second identification and confirmation result. At the corrected target rejection time, the corresponding rejection execution mechanism is controlled to perform diversion processing on the corresponding finished glove.

[0023] To implement this method, an online detection and rejection system is set up and connected to the end of the glove production line. The online detection and rejection system, along the finished glove conveying direction, sequentially includes a feeding conveyor unit, a detection carrier circulation conveyor unit, a wrist opening guide structure, a ring-shaped pressing structure, a segmented support core, a pressure adjustment component, a reflection imaging component, a transmission imaging component, a deformation response acquisition component, a carrier identification component, a position detection component, a control processing unit, and a rejection execution mechanism.

[0024] The circulating conveying unit for the testing carriers is used to carry multiple testing carriers and allow each carrier to sequentially pass through the loading station, testing station, and rejection station along a predetermined path. Each testing carrier includes a carrier base, a sealing socket disposed on the carrier base, a positioning and limiting part disposed on the outer periphery of the sealing socket, and a carrier identification part disposed on the carrier base. The carrier identification part is preferably an coded carrier fixedly disposed on the carrier base. The coded carrier can be any of a barcode, QR code, or RFID tag; in this embodiment, a QR code is preferred to uniquely identify the corresponding testing carrier. The positioning and limiting part is used to limit the insertion depth of the glove's wrist opening and maintain the axial reference position of the glove relative to the sealing socket, thus avoiding excessively deep or shallow insertion or unilateral offset of the wrist opening, thereby ensuring that subsequent shaping, imaging, and pressure testing are performed under uniform boundary conditions.

[0025] The wrist opening guide structure is located at the loading station and is used to stably guide the wrist opening of the demolded glove into the sealing socket. The wrist opening guide structure includes an inlet and a straightening component. The inlet cross-section of the inlet is larger than the outlet cross-section, which is used to guide the expansion of the wrist opening edge in its naturally closed state. The straightening component is located on the outlet side of the inlet and is used to circumferentially straighten the wrist opening edge, so that the wrist opening edge is evenly spread along the outer periphery of the sealing socket, avoiding local folding, flanging, or unilateral creases of the wrist opening. The annular pressing structure is located on the outer periphery of the sealing socket and can reciprocate along the axial direction of the sealing socket. When the wrist opening edge is fitted onto the outer periphery of the sealing socket after being straightened by the straightening component, the annular pressing structure moves towards the sealing socket, so that the wrist opening edge is clamped between the annular pressing structure and the sealing socket, thereby forming a continuous circumferential seal to form a cavity to be tested inside the finished glove.

[0026] In this embodiment, after the cavity to be tested is formed, the finished glove needs to further achieve a preset testing posture. The preset testing posture refers to the stable unfolding of each finger and palm under the action of the segmented support core, while maintaining continuous circumferential sealing at the wrist edge, and satisfying the following conditions: each finger maintains basic extension along its corresponding extension direction, without significant folding or adhesion / obstruction between adjacent fingers; the local undulations of the palm area relative to the imaging plane remain within a preset allowable range, not affecting the effective acquisition of reflected and transmitted images; and the wrist edge maintains continuous fit with the sealing socket, without loosening or obvious wrinkles. Through the above state definition, the preset testing posture has a clear implementation basis, rather than being a simple result description.

[0027] The segmented support core is located at the testing station, on the axial inner side of the sealing socket. The segmented support core includes a palm support segment and five finger support segments corresponding to the thumb, index finger, middle finger, ring finger, and little finger, respectively. Each finger support segment is driven by an independent drive unit, preferably a miniature electric push rod with displacement feedback. Each finger support segment is equipped with a displacement acquisition element, and a current monitoring module is installed in the control circuit of the corresponding drive unit to simultaneously acquire displacement and contact feedback information when the finger support segment extends into the corresponding finger cavity.

[0028] The contact feedback is preferably determined by the change in the output current of the driving component. That is, when the driving current of a finger segment continues to advance and reaches a preset contact threshold, it indicates that the finger segment has formed a restricted contact with the inner wall of the corresponding finger, and the unfolding of the finger segment is stopped. If a finger segment reaches the corresponding displacement threshold before reaching the contact threshold, the unfolding of the finger segment is also stopped. The displacement threshold is not arbitrarily set, but is pre-calibrated based on the finger length, finger width, wall thickness, and normal unfolding margin of the corresponding glove model, and stored in the control processing unit. Specifically, qualified samples of the same model can be selected first, and multiple rounds of unfolding tests can be conducted on each finger under the condition of not causing local abnormal swelling and not producing obvious wrinkles. The displacement range when each finger is stably unfolded is taken as the reference range of the corresponding finger displacement threshold, and then the final displacement threshold is determined by combining it with the safety margin. Thus, each finger segment can complete the unfolding according to the actual structural characteristics of the glove model, rather than relying on a single fixed extension for rough unfolding.

[0029] In this embodiment, the five finger support segments are preferably unfolded sequentially in the order of thumb, index finger, middle finger, ring finger, and little finger. After each finger support segment reaches its corresponding displacement threshold or receives corresponding contact feedback, the palm support segment then compensates and flattens the palm area. The reason for adopting the order of finger support before palm support is that if the palm area is flattened as a whole first, it is easy to cause secondary folding or local traction between the base of the thumb and adjacent fingers, affecting subsequent transmission sampling and deformation response sampling. By completing the independent positioning of each finger first, and then performing palm compensation and flattening, the entire finished glove can form a stable and repeatable preset detection posture.

[0030] The pressure regulating assembly is connected to the chamber to be tested and preferably includes an intake valve, a pressure relief valve, a pressure regulating valve, a pressure source, and a pressure sensor. The pressure sensor is located near the sealing socket and is used to collect pressure data from the chamber to be tested in real time. Based on the real-time pressure data collected by the pressure sensor, the control processing unit performs closed-loop regulation on the opening degree and action sequence of the intake valve and the pressure relief valve, thereby sequentially realizing four stages: pre-pressurization, pressure stabilization, pulse pressurization, and pressure holding decay.

[0031] The target pressure during the pre-charging stage is the first pressure value, the target pressure during the stabilization stage is the second pressure value, and the peak pressure during the pulse boosting stage is the third pressure value. All three must satisfy the condition that the third pressure value is greater than the second pressure value, and the second pressure value is greater than the first pressure value. The first pressure value is used to eliminate wrinkles remaining in the finished glove during the introduction and shaping process, transforming it from a partially folded state to an initially unfolded state. The second pressure value is used to maintain a stable contour in the palm and finger areas of the finished glove without causing abnormal local swelling in a normally qualified glove. The third pressure value is used to amplify the dynamic response of weak areas, areas adjacent to micropores, or areas of local adhesion under short-term pressure conditions, thereby making latent defects that are not easily visible in a static state identifiable.

[0032] The first, second, and third pressure values ​​are preferably obtained through sample calibration. Specifically, for finished gloves of different materials, sizes, and thicknesses, qualified samples, typical micro-defect samples, and typical unqualified samples of the same model are pre-selected. Multiple rounds of testing are conducted under the same equipment and cycle time conditions. The lowest stable pressure when wrinkles are essentially eliminated is recorded as the reference range for the first pressure value. The pressure when the contour remains stable without significant excessive bulging is recorded as the reference range for the second pressure value. Then, the pressure is gradually increased based on the second pressure value, and the pressure that significantly amplifies the response to latent defects without causing the qualified sample to break or permanently deform is selected as the reference range for the third pressure value. During actual operation, the parameter group corresponding to the current product model can be called.

[0033] The duration of the pressure stabilization phase is not arbitrarily set, but determined based on the glove material, specifications, wall thickness, the stabilization time of the outline after shaping, and the image acquisition cycle. Preferably, in tests on samples of the same model, the time required for the intracavity pressure fluctuation to stabilize and the change in the outline of the palm and finger areas to decrease to within a preset allowable range is used as the basis for determining the pressure stabilization duration. This setting ensures that the pressure state within the chamber under test has stabilized when acquiring reflective and transmissive images, avoiding image judgment fluctuations caused by pressure imbalance.

[0034] After completing the support and differential pressure control, the reflection imaging component, transmission imaging component, and deformation response acquisition component begin to work collaboratively. The reflection imaging component, located on the outside of the finished glove, includes a reflection light source and a reflection imaging unit, used to acquire reflection images of the outer surface of the finished glove. The transmission imaging component, located on opposite sides of the finished glove, includes a transmission light source and a transmission imaging unit, used to acquire transmission images of the corresponding areas of the finished glove. The deformation response acquisition component preferably shares an imaging unit with the reflection imaging component, triggering sampling only at different pressure levels. To ensure that the images of the palm and back of the hand stably correspond to the same detection object, in this embodiment, after completing the palm detection on the same detection carrier, the detection carrier is rotated around its longitudinal axis, causing the back of the hand to rotate to the same imaging position. Reflection and transmission sampling are then repeated, and the same sealing socket and the same support state are maintained during the rotation process, ensuring that the palm and back of the hand images correspond to the same state of the cavity to be detected.

[0035] To avoid optical crosstalk between reflection and transmission imaging, this embodiment employs a time-division triggering method for data acquisition. Specifically, during the voltage stabilization phase, the transmission light source is first turned off, and only the reflection light source is turned on to acquire reflection images; subsequently, the reflection light source is turned off, and the transmission light source is turned on to acquire transmission images. After completing image acquisition during the voltage stabilization phase, the pulse boosting phase begins. Before the pulse boosting begins, a deformation response image is acquired at the first moment; when the intracavity pressure approaches the third pressure value, a deformation response image is acquired at the second moment. If necessary, a third moment deformation response image can be acquired after the pressure drops back to the second pressure value for further analysis of the rebound state in local areas. Finally, during the pressure holding and decay phase, the data sequence of cavity pressure changes over time is recorded.

[0036] In this embodiment, to facilitate subsequent defect localization and registration analysis, the finished glove is divided into multiple fixed detection areas, including the thumb area, index finger area, middle finger area, ring finger area, little finger area, palm area, palm heel area, and wrist area. Image registration for each area can be achieved based on the wrist edge baseline, the center position of the sealing socket, and fixed reference points on the detection carrier. Since the detection carriers are structurally consistent, and the finished glove forms a unified detection boundary under the combined action of the positioning and limiting part, the sealing socket, and the segmented support core, deformation response images acquired at different time points can maintain a high degree of consistency in the corresponding areas.

[0037] Based on the above sampling results, the control processing unit processes different types of data separately. For reflection images, it mainly extracts apparent defect features. For transmission images, it mainly extracts light transmission anomaly features. For deformation response images, it mainly extracts pressure response anomaly features. For pressure data acquired during the pressure holding and decay stage, it mainly extracts sealing defect features.

[0038] To ensure a clear method for obtaining abnormal characteristics of the pressure response, this embodiment employs a combination of region registration and response quantity calculation. Let the... The projected area of ​​each detection region in the image at the first moment before pulse boosting is: The projected area in the image at the second time step near the peak of the pulse boost is Let the first The perimeter of the contour of each detection region in the image at the first moment is: The perimeter of the contour in the image at the second time step is Then define the first... The deformation response coefficient of each detection area is: In the formula, Indicates the first Deformation response coefficient of each detection area; Indicates the first The projected area of ​​each detection region in the image at the first moment; Indicates the first The projected area of ​​each detection region in the image at the second time step; Indicates the first The perimeter of the contour of each detection region in the first-time image; Indicates the first The perimeter of the contour of each detection region in the image at the second time step; and These are the weighting coefficients, and In this embodiment, Used to characterize the contribution of regional area changes to anomalous responses. Used to characterize the contribution of contour changes to the anomalous response. The deformation response threshold. This is achieved through multiple rounds of testing on qualified samples of the same model under the same pressure program. Specifically, the maximum deformation response coefficient for each test area of ​​the qualified sample can be calculated first, and then a preset safety factor can be introduced based on this maximum value to obtain the response threshold for the corresponding glove model. When a certain detection area meets the requirements When this occurs, the detection area is identified as an abnormal pressure response area. Through the above method, the abnormal deformation response has a clear source of judgment, rather than being a purely empirical assessment. For sealing defect characteristics, this embodiment uses the pressure holding decay rate for characterization. Let the intracavity pressure at the start of the pressure holding decay stage be... The intracavitary pressure at the end is The corresponding times are respectively and The pressure decay rate can then be expressed as: In the formula, Indicates the rate of pressure decay; This indicates the intracavitary pressure at the start of the pressure-holding decay phase; This indicates the intracavitary pressure at the end of the pressure holding and decay phase; Indicates the start time of the pressure holding and decay phase; This indicates the end time of the pressure holding and decay phase. greater than the preset leakage threshold If the product is found to have a sealing defect, the resulting glove is deemed to have an air leakage threshold. Similarly, it can be obtained through comparison and calibration of qualified samples and samples with known micro-leakage.

[0039] After acquiring features of apparent defects, abnormal light transmission, abnormal pressure response, and sealing defects, the control processing unit performs a fusion judgment on these various defect features according to preset judgment rules. To ensure the feasibility of this fusion judgment method, this embodiment preferably establishes a comprehensive defect scoring system. Its expression is: In the formula, This indicates the overall defect score; This represents the apparent defect score extracted based on the reflection image; This represents the transmittance anomaly score extracted from the transmission image; This represents the score of abnormal compressive response extracted based on the deformation response coefficient; This represents the sealing defect score extracted based on the pressure decay rate; For the corresponding weight coefficients, and in, , , and All of these can be obtained by normalizing the deviation of each corresponding feature quantity from the threshold of the qualified sample. That is, firstly, benchmark thresholds are established for qualified samples of the same model in four dimensions: surface defects, abnormal light transmission, pressure response, and sealing attenuation. Then, the deviation of the glove under test in the corresponding dimensions is mapped to a score value with a unified dimension. This allows four types of features from different sources to be included in the same evaluation framework. The non-compliance threshold is obtained through sample calibration. and review and judgment threshold and satisfy .when When this happens, the corresponding finished gloves will be judged as defective; when When the corresponding finished gloves are identified as verification products, they are considered as such. At that time, the corresponding finished gloves will be judged as qualified products.

[0040] In actual operation, priority rules can also be set to directly determine obvious fatal defects. For example, when a penetrating pinhole appears in the transmission image, or when the pressure attenuation rate is much higher than the leakage threshold, it can be directly determined as a non-conforming product without waiting for the comprehensive defect score to be calculated. The above priority rules are still part of the preset judgment rules and do not change the overall idea of ​​multi-source feature fusion judgment in this embodiment.

[0041] To ensure accurate correspondence with subsequent rejection actions, target rejection information must be generated after the detection results are determined. This target rejection information includes at least vehicle identification information, defect type, defect area location information, rejection station number, and target rejection time. The defect area location information is obtained by mapping each defect feature to a pre-defined thumb area, index finger area, middle finger area, ring finger area, little finger area, palm area, palm base area, or wrist area.

[0042] The method for determining the rejection station number is as follows: When the system has only one rejection station, the rejection station number is a fixed number for that rejection station; when the system has multiple rejection stations, the control processing unit assigns a target rejection station number to the corresponding finished glove based on the running path from the detection position to each rejection station, the current conveying cycle time, and the load status of each rejection station's channel. This ensures that the rejection station number has a clear basis for generation.

[0043] The target rejection time is not simply estimated using a fixed delay, but is calculated based on the detection completion time, the path length of each conveying section between the detection position and the rejection station, and the actual operating speed. Let the detection completion time be... The path from the detection location to the rejection station is divided into Section 1, No. The path length of each segment is , No. The vehicle speed at each segment at the corresponding time is The system control compensation time is The target removal time It can be represented as: In the formula, Indicates the time when the target was removed; Indicates the time when the test was completed; Indicates the first The path length of each segment; Indicates the first Vehicle operating speed in each section; This indicates the control compensation time. Before the corresponding detection vehicle reaches the rejection station, the vehicle identification component reads the vehicle identification information of the corresponding detection vehicle again, while the position detection component acquires the real-time position information of the detection vehicle. If there is a deviation between the actual position and the theoretical position, the control processing unit corrects the target rejection time based on the deviation. Let the rejection reference position be... The current real-time location of the detection vehicle is The current operating speed of the excluded segment is Then the time deviation correction amount for: Corrected target removal time for: In the formula, This indicates the amount of time deviation correction; Indicates the removal of the reference position; Indicates the real-time location of the current detection vehicle; Indicates the current operating speed of the excluded segment; This indicates the corrected target rejection time. Using this method, accurate sorting of the corresponding finished glove kits can still be achieved even when the detection vehicle exhibits speed fluctuations, minor cumulative errors in chain links, or beat disturbances.

[0044] The rejection mechanism is located at the rejection station and electrically connected to the control processing unit. In this embodiment, the rejection mechanism preferably adopts a flow-dividing structure combining a guide vane and auxiliary airflow. For gloves determined to be defective, at the corrected target rejection time, the guide vane is switched to the defective product collection channel side, and auxiliary airflow is applied simultaneously to detach the corresponding gloves from the corresponding inspection carrier and guide them into the defective product collection channel. For gloves determined to be valid, the guide vane is switched to the valid product channel side, guiding the gloves into the valid product channel. For gloves determined to be valid, the guide vane remains in its original position, and the gloves continue to run along the original conveying path and enter the valid product collection station, thereby completing the three-way diversion processing of valid, valid, and defective products.

[0045] Furthermore, within multiple consecutive detection cycles, the control processing unit statistically analyzes the occurrence of defects of the same type and in the same area. The warning threshold can be determined based on the cumulative number of occurrences or the proportion of defects of the same type and in the same area within a preset statistical window. The statistical window and the warning threshold are preset based on historical quality data, typical defect distribution, and production cycle time for the corresponding product model. In other words, "multiple consecutive detection cycles" is not a general statement, but rather uses a preset statistical window as the unit of analysis. Within this statistical window, when the cumulative number of occurrences or the proportion of defects of a certain type and in a certain area reaches the warning threshold, corresponding process warning information is generated.

[0046] For example, within a preset statistical window, if a high proportion of abnormal light transmission defects continuously appear in the palm area, process warning information related to the control of the dip coating thickness is generated; if edge loss or abnormal flanging continuously appears in the wrist area, process warning information related to the demolding process, transfer process, or finishing process is generated; if the same finger area continuously shows an abnormal increase in pressure response, process warning information related to local molding uniformity, drying conditions, or front-end stretching state is generated. After the process warning information is sent to the front-end production station, the operating parameters of the dip coating process, drying process, demolding process, or conveying process can be adjusted by the upper control system or on-site operators. Through the above settings, this embodiment not only realizes online detection and rejection of single-piece finished gloves, but also can use the detection results to correct the front-end process, thereby forming a complete online quality closed loop.

[0047] The specific operation process of this embodiment is as follows: First, the demolded gloves are transferred to the loading station, where the wrist opening guide structure expands and circumferentially adjusts it, ensuring the wrist opening fits evenly onto the outer circumference of the sealing socket of the corresponding testing carrier. Then, the annular pressing structure moves axially, continuously pressing the edge of the wrist opening against the outer circumference of the sealing socket to form the cavity to be tested, while simultaneously recording the carrier identification information corresponding to the testing carrier.

[0048] Then, the segmented support core is inserted into the cavity to be tested, and the five finger support segments move sequentially into their respective finger areas. Each finger support segment stops unfolding after reaching its corresponding displacement threshold or receiving contact feedback, avoiding excessive stretching of the fingers of a normal glove. After all the finger support segments have been positioned, the palm support segment compensates for and flattens the palm area, allowing the finished glove to enter a stable preset testing posture.

[0049] Based on this, the pressure regulating component sequentially applies pre-charge pressure, stabilization pressure, pulsed pressure boosting, and pressure holding attenuation to the cavity under test. The pre-charge pressure stage mainly eliminates residual wrinkles, the stabilization pressure stage provides stable imaging conditions, the pulsed pressure boosting stage actively amplifies the dynamic response of weak areas, areas adjacent to micropores, or localized adhesion areas, and the pressure holding attenuation stage is used to extract sealing defect information. Compared with schemes that only acquire images under natural suspension or only determine leakage after a single overall inflation, this embodiment transforms latent defects from a difficult-to-discover state to an identifiable state through standardized support and staged differential pressure excitation.

[0050] During the pressure stabilization phase, reflective and transmissive images of the palm and back of the hand are acquired respectively. During the pulse pressure boosting phase, deformation response images at least at two time points are acquired, and the deformation response coefficient is calculated by combining region registration. Subsequently, the pressure decay rate is obtained during the pressure holding and decay phase. The control processing unit extracts apparent defect features, light transmission anomaly features, pressure response anomaly features, and sealing defect features respectively, and performs fusion judgment according to preset judgment rules to obtain the detection results of qualified products, verification products, or unqualified products, while generating defect area location information.

[0051] The control processing unit then binds the detection results with the corresponding carrier identification information, detection position, and conveying cycle time to generate target rejection information and calculate the target rejection time. Before the detection carrier reaches the rejection station, its carrier identification information is read again and real-time position information is collected to correct the deviation of the target rejection time. Subsequently, at the corrected target rejection time, the rejection execution mechanism is controlled to achieve precise diversion of non-conforming and verification products, while qualified products enter the qualified product collection station along the original conveying path. This avoids the problems of incorrect rejection and missed rejection caused by simply controlling rejection based on a fixed delay.

[0052] It should be noted that the displacement threshold, contact threshold, first pressure value, second pressure value, third pressure value, pressure holding time, deformation response threshold, leakage threshold, review judgment threshold, non-conformity judgment threshold, and early warning threshold in this embodiment can all be determined through sample calibration for finished gloves of different materials, sizes, and thicknesses. Sample calibration refers to selecting a predetermined number of qualified samples, slightly defective samples, and typical non-conforming samples before the official production of gloves of the same model, conducting multiple rounds of testing under the same equipment and cycle time conditions, and determining the corresponding thresholds by combining the false positive rate, false negative rate, rejection accuracy rate, and the effectiveness of process early warnings. Once the above thresholds are determined, they can be stored in the control processing unit as operating parameters for that model of glove, and the corresponding parameters can be retrieved when switching product models.

[0053] Furthermore, in this embodiment, the apparent defect features, light transmission anomaly features, pressure response anomaly features, and sealing defect features are not obtained in isolation, but rather are obtained around the same detection vehicle, the same cavity to be detected, the same preset detection posture, and the same timing chain. Wrist-mouth standard import and circumferential pressure sealing ensure consistent detection boundaries; segmented support shape ensures consistent detection morphology; staged differential pressure excitation ensures consistent defect manifestation conditions; multi-source image and pressure data fusion ensures consistent judgment criteria; and vehicle identification binding and time correction ensure consistent rejection targets. It is precisely because the above-mentioned links are implemented collaboratively under the same object, the same boundary, and the same timing that this embodiment is fundamentally different from existing schemes that rely solely on single visual detection, single leak detection, or simple delayed rejection.

[0054] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for online inspection and rejection of defective finished gloves, characterized in that, Includes the following steps: S1: The demolded finished glove is transported to the testing carrier. The wrist opening of the finished glove is guided into the sealing socket through the wrist opening guide structure, and the edge of the wrist opening is circumferentially pressed and sealed using the annular pressing structure to form a cavity to be tested inside the finished glove. At the same time, the carrier identification information of the testing carrier is recorded. S2: Insert the segmented support core into the cavity to be tested and perform the support action. The segmented support core includes a palm support segment and multiple finger support segments corresponding to each finger of the glove. Each finger support segment unfolds independently according to a preset displacement threshold or contact feedback. The palm support segment flattens and supports the palm area so that the finished glove forms a preset testing posture. S3: Apply pre-pressure, stabilize pressure, pulse pressure increase and pressure holding attenuation sequentially to the cavity to be tested, so that the finished glove eliminates wrinkles and completes initial shaping in the pre-pressure stage, maintains a stable testing state in the stabilize pressure stage, induces abnormal deformation response in weak areas, microporous areas or adhesion areas in the pulse pressure increase stage, and forms corresponding cavity pressure change data in the pressure holding attenuation stage. S4: During the pressure stabilization phase, reflective and transmissive images of the finished glove are acquired; during the pulse boosting phase, deformation response images at least at two moments are acquired; and during the pressure holding and decay phase, cavity pressure decay data are acquired. S5: Extract apparent defect features from the reflected image, extract light transmission anomaly features from the transmitted image, extract pressure response anomaly features from the deformation response image, extract sealing defect features from the cavity pressure attenuation data, and perform fusion judgment on the apparent defect features, light transmission anomaly features, pressure response anomaly features and sealing defect features according to preset judgment rules, so as to output the detection results of the corresponding finished glove. S6: Bind the detection results with the corresponding vehicle identification information, detection position and conveying cycle to generate target rejection information; S7: When the corresponding detection carrier runs to the rejection station, the detection carrier is re-identified and confirmed according to the target rejection information, and the target rejection information is corrected according to the re-identification and confirmation results. At the corrected target rejection time, the corresponding rejection execution mechanism is controlled to perform diversion processing on the corresponding finished glove.

2. The method for online inspection and rejection of defective finished gloves according to claim 1, characterized in that, The detection carrier includes a carrier base, a sealing socket on the carrier base, a positioning and limiting part on the outer periphery of the sealing socket, and a carrier identification part on the carrier base. The wrist opening guide structure includes an inlet and a complete opening component. S1 specifically includes: expanding and guiding the wrist opening of the finished glove using the inlet, circumferentially tidying the edge of the wrist opening using the straightening component, fitting the wrist opening onto the outer periphery of the sealing socket, and controlling the annular pressing structure to move axially along the sealing socket so that the edge of the wrist opening is clamped between the annular pressing structure and the sealing socket to form a continuous circumferential seal. The carrier identification information is provided by the carrier identification unit.

3. The method for online inspection and rejection of defective finished gloves according to claim 2, characterized in that, S2 specifically includes: controlling multiple finger support segments to move one by one toward the inner side of the corresponding finger; stopping the finger support segment from continuing to unfold after any finger support segment reaches the corresponding displacement threshold or receives the corresponding contact feedback; and controlling the palm support segment to compensate and flatten the palm area after each finger support segment has finished unfolding.

4. The method for online inspection and rejection of defective finished gloves according to claim 3, characterized in that, S3 specifically includes: in the pre-pressurization stage, air is introduced into the chamber to be tested at a first pressure value; in the pressure stabilization stage, the pressure in the chamber to be tested is adjusted to a second pressure value and maintained for a predetermined time; in the pulse pressurization stage, the pressure in the chamber to be tested is increased from the second pressure value to a third pressure value and then dropped back to the second pressure value, wherein the third pressure value is greater than the second pressure value; in the pressure holding and decay stage, air intake is stopped and the pressure decay process in the chamber to be tested is collected. In the above process, pressure data in the chamber to be tested is collected in real time using a pressure sensor, and the intake valve and pressure relief valve are controlled according to the pressure data to perform closed-loop regulation of the pressure in the chamber to be tested.

5. The method for online inspection and rejection of defective finished gloves according to claim 4, characterized in that, S4 specifically includes: setting a reflective light source and a reflective imaging unit on the outside of the finished glove to acquire a reflective image of the outer surface of the finished glove; Transmitted light sources and transmission imaging units are set on opposite sides of the finished glove to acquire transmission images of corresponding areas of the finished glove. Images were captured from the palm and back of the hand areas of the finished glove. The reflected and transmitted images are acquired in a time-division triggering manner and stored in association with the vehicle identification information of the same detection vehicle.

6. The method for online inspection and rejection of defective finished gloves according to claim 5, characterized in that, In step S5, extracting abnormal features of the pressure response based on the deformation response image specifically includes: registering deformation response images acquired at least two times during the pulse boosting stage, calculating the contour change, local bulging, or area change of the corresponding region at different acquisition times, and identifying regions exceeding the response threshold as abnormal regions of the pressure response.

7. The method for online inspection and rejection of defective finished gloves according to claim 6, characterized in that, In S5, the apparent defect features, light transmission abnormal features, pressure response abnormal features and sealing defect features are fused and judged according to the preset judgment rules. Specifically, this includes: identifying surface defects based on reflection images, identifying light transmission abnormal defects based on transmission images, identifying pressure abnormal defects based on deformation response images, and identifying sealing defects based on cavity pressure attenuation data. When any defect feature meets the corresponding non-conformance judgment condition, the corresponding finished glove will be judged as a non-conformance product; If the defect characteristics meet the preset verification conditions but do not meet the non-conformity judgment conditions, the corresponding finished glove will be judged as a verification product; the remaining finished glove will be judged as a qualified product.

8. The method for online inspection and rejection of defective finished gloves according to claim 7, characterized in that, After extracting the apparent defect features, light transmission abnormality features, and pressure response abnormality features, the method further includes: mapping the apparent defect features, light transmission abnormality features, and pressure response abnormality features to the finger area, palm area, or wrist area of ​​the finished glove to generate defect area location information.

9. The method for online inspection and rejection of defective finished gloves according to claim 8, characterized in that, The target rejection information generated in S6 includes vehicle identification information, defect type, rejection station number and target rejection time. The target rejection time is calculated based on the detection location, conveying cycle time and the running path of the detection vehicle to the rejection station. The secondary identification and confirmation of the detection vehicle in S7 specifically includes: reading the vehicle identification information of the detection vehicle before the rejection station, collecting the real-time position information of the detection vehicle, correcting the deviation of the target rejection time according to the vehicle identification information and the real-time position information, and controlling the corresponding rejection execution mechanism to act at the corrected target rejection time.

10. The method for online inspection and rejection of defective finished gloves according to claim 9, characterized in that, In step S7, when the test result of the corresponding finished glove is a non-conforming result, the rejection mechanism is controlled according to the defect type and the product is introduced into the non-conforming product collection channel. When the test result of the corresponding finished glove is a review judgment result, the control rejection execution mechanism will import it into the review channel; Within multiple consecutive detection cycles, when the number of occurrences of the same type of defect in the same area reaches the warning threshold, a corresponding process warning message is generated and sent to the front-end production station for adjusting the operating parameters of the corresponding process.

Images