Optical non-destructive detection device and method for the tightness of the bottle opening of blueberries

Abstract

The present application relates to blueberry bottle mouth sealing optical nondestructive testing equipment technical field, especially in blueberry bottle mouth sealing optical nondestructive testing device and method;Technical scheme: blueberry bottle mouth sealing optical nondestructive testing device, including optical detection device, installation groove, screw rod, mounting bracket, mounting ring, fresnel lens, screw micrometer, double color laser line light source and rotating motor, the inside both sides of optical detection device are provided with installation groove;Compared with the traditional blueberry bottle mouth sealing optical nondestructive testing device, more single light source or simple lens group is used, it is difficult to cover 360 ° annular area of bottle mouth, leading to insufficient illumination of bottle mouth edge or complex curved surface area, small defects are easy to be missed, the blueberry bottle mouth sealing optical nondestructive testing device is through double color laser line light source cooperation with fresnel lens array, and the multispectral reflection data is synchronously collected, and the wavelength difference is used to enhance the contrast of surface micro mark, and the monochromatic light detection sensitivity is effectively improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of blueberry bottle mouth sealing optical nondestructive testing equipment, in particular to a blueberry bottle mouth sealing optical nondestructive testing device and method. BACKGROUND

[0002] The blueberry bottle mouth sealing optical nondestructive testing device is a device specially used for detecting the sealing performance of the bottle mouth of a blueberry bottle. It adopts optical nondestructive testing technology and can quickly and accurately detect the sealing performance of the bottle mouth without damaging the bottle mouth.

[0003] The existing blueberry bottle mouth sealing optical nondestructive testing device mostly uses a single light source or a simple lens group, which is difficult to cover the 360° annular area of the bottle mouth, resulting in insufficient light in the edge or complex curved surface area of the bottle mouth, and small defects are easy to be missed. At the same time, it relies on monochromatic light or low-resolution sensors, and cannot obtain the multi-spectral reflection characteristics of the bottle mouth surface, making it difficult to distinguish the texture and defect characteristics of the material itself, and the false detection rate is high.

[0004] In view of the problem that the existing blueberry bottle mouth sealing optical nondestructive testing device mostly uses a single light source or a simple lens group, which is difficult to cover the 360° annular area of the bottle mouth, resulting in insufficient light in the edge or complex curved surface area of the bottle mouth, and small defects are easy to be missed, the blueberry bottle mouth sealing optical nondestructive testing device utilizes wafer defect / grain monitoring related technology, synchronously collects multi-spectral reflection data through a double-color laser line light source combined with a Fresnel lens array, enhances the contrast of surface marks by wavelength difference, effectively improves the detection sensitivity of monochromatic light, and combines structured light triangulation and screw micrometer encoding data to reconstruct the three-dimensional point cloud of the bottle mouth sealing surface, realizes quantitative judgment of the flatness multi-parameters, and greatly improves the detection accuracy. SUMMARY

[0005] In order to overcome the problem that the existing blueberry bottle mouth sealing optical nondestructive testing device mostly uses a single light source or a simple lens group, which is difficult to cover the 360° annular area of the bottle mouth, resulting in insufficient light in the edge or complex curved surface area of the bottle mouth, and small defects are easy to be missed, and relies on monochromatic light or low-resolution sensors, and cannot obtain the multi-spectral reflection characteristics of the bottle mouth surface, making it difficult to distinguish the texture and defect characteristics of the material itself, and the false detection rate is high.

[0006] The technical scheme of the present application is: the blueberry bottle mouth sealing optical nondestructive testing device, including optical detection device, reflection component, mounting groove, screw rod, mounting bracket, mounting ring, Fresnel lens, screw micrometer, double-color laser line light source, rotating motor and analysis component, the top surface of the optical detection device is provided with the reflection component, the inside of the optical detection device is provided with the mounting groove on both sides, the inside of the mounting groove is provided with the screw rod, the outside of the screw rod is provided with the mounting bracket, both ends of the mounting bracket are provided with the mounting ring, the surface of the mounting ring is provided with a plurality of Fresnel lenses, the Fresnel lenses are arrayed, one side of the mounting bracket is provided with the screw micrometer, the upper side of the screw micrometer is provided with the double-color laser line light source, the top surface of the optical detection device is provided with the rotating motor, one side of the optical detection device is provided with the analysis component.

[0007] Preferably, the detection light source inside the optical detection device is refracted through the reflection component to ensure that the light uniformly covers the 360° annular area of the bottle mouth, the screw rod is installed through the mounting groove, the screw rod is rotated through the rotating motor, the mounting bracket is moved up and down through the screw rod, the mounting ring is installed through the mounting bracket, the Fresnel lens is installed through the mounting ring, the light source inside the optical detection device is refracted through the Fresnel lens, the Fresnel lens can effectively collect and converge light, the small scratches on the surface of the blueberry bottle mouth are irradiated and refracted, the position distance between the mounting ring and the blueberry bottle is detected through the screw micrometer, stable and uniform light is provided through the double-color laser line light source, the double-color laser line is uniformly irradiated to the bottle mouth, and the refracted light is analyzed through the analysis component to determine whether the bottle mouth has defects.

[0008] As a preferred, the reflection component includes a mounting box, a replaceable top plate, a light-transmitting mirror, an electric telescopic rod, a reflective mounting plate and an annular micro-prism array, the top surface of the optical detection device is provided with the mounting box, the inside of the mounting box is provided with the replaceable top plate, the bottom surface of the replaceable top plate is provided with the light-transmitting mirror, the bottom surface of the replaceable top plate is provided with the electric telescopic rod, the electric telescopic rod is provided with a plurality of groups, the bottom end of the electric telescopic rod is provided with the reflective mounting plate, and the bottom surface of the reflective mounting plate is provided with the annular micro-prism array.

[0009] As a preferred, the analysis component includes a light-transmitting bottom plate, a CMOS sensor array, a light source amplification channel, a light source receiver and a nondestructive testing analyzer, the inside of the optical detection device is provided with the light-transmitting bottom plate, the surface of the light-transmitting bottom plate is provided with the CMOS sensor array, the CMOS sensor array is provided with a plurality of groups, one side of the optical detection device is provided with the light source amplification channel, one end of the light source amplification channel is provided with the light source receiver, and the other side of the optical detection device is provided with the nondestructive testing analyzer.

[0010] The blueberry bottle mouth sealing optical nondestructive testing method includes the following methods:

[0011] S101: First, the system is initialized, and the bottle body is positioned;

[0012] S102: The dual-color laser line light source is modulated, and the screw micrometer and dynamic light field are dynamically calibrated;

[0013] S103: The environmental micro-prism array is adapted;

[0014] S104: Multi-spectral data is collected;

[0015] S105: Intelligent identification of defects on the blueberry bottle mouth is performed;

[0016] S106: The detection results are output and the quality is traced.

[0017] As preferred, when the system initialization and bottle positioning are performed, the following steps are included:

[0018] S201: Turn on the power of the optical detection device, the system automatically performs hardware self-checking, checks the power stability of the dual-color laser line light source, the error needs to be ≤±2%, the stepping accuracy of the rotating motor, the stepping angle error ≤0.1°, the dark current noise of the CMOS sensor array, the dark current noise <0.05e- / pixel / s;

[0019] S202: Input the blueberry bottle specification parameters through the control terminal, including the bottle mouth diameter, height and material refractive index, the system automatically matches the preset annular micro-prism array tilt angle, calculates the bottle mouth curvature using the formula: θ = arcsin(n1 / n2), where n1 is the air refractive index and n2 is the bottle material refractive index;

[0020] S203: The conveyor belt transports the blueberry bottle to the detection station, the pneumatic clamping jaw clamps the bottle body from both sides, the laser ranging sensor measures the deviation of the bottle mouth axis from the center line of the detection device, and the six-axis correction platform fine-tunes the bottle body position to ensure that the bottle mouth center and the optical axis of the dual-color laser line light source coincide to within ≤0.1mm.

[0021] As preferred, when the light source configuration and dynamic calibration are performed, the following steps are included:

[0022] S301: The dual-color laser line light source emits laser lines with wavelengths λ1=450nm (blue light) and λ2=660nm (red light), which are expanded into an annular light curtain covering 360° of the bottle mouth through a Fresnel lens array, and the light intensity ratio is set to I1 / I2=1.5 to optimize the surface mark contrast;

[0023] S302: Rotate the screw micrometer to drive the mounting ring to move axially along the screw shaft, adjust the working distance of the Fresnel lens to the bottle mouth end face to D = bottle mouth radius x 0.8, ensure that the light incidence angle θ ≤ 30°, avoid total reflection;

[0024] S303: Rotate the motor to drive the mounting bracket, and the CMOS sensor array real-time collects the bottle mouth reflected light intensity distribution. The system automatically adjusts the extension amount of the electric telescopic rod of the light reflection mounting plate through the PID algorithm, the adjustment accuracy is ± 0.02 mm, and the standard deviation of the light intensity in the annular region is ≤ 5%.

[0025] As preferred, when the annular micro-prism array is adapted, the following steps are included:

[0026] S401: The visual recognition system scans the bottle mouth shape, selects the matching model from the replaceable top plate library, and completes the top plate replacement within 15 seconds through the magnetic quick replacement mechanism;

[0027] S402: According to the bottle mouth thread parameters, control the electric telescopic rod to adjust the height H of the light reflection mounting plate, so that the prism vertex of the annular micro-prism array is in the same phase plane with the root of the bottle mouth thread;

[0028] S403: Emit test laser pulses, after passing through the light transmission mirror and the annular micro-prism array, form uniformly distributed diffuse reflection spots on the bottle mouth, verify that there is no blind area in the coverage area, and the spot density is ≥ 200 points / cm 2 .

[0029] As preferred, when the multi-spectral data is collected, the following steps are included:

[0030] S501: The double-color laser line light source emits alternately at a frequency of 1 kHz, and the CMOS sensor array synchronously captures the reflection images of λ1 and λ2 in double exposure mode, and the exposure time is set to t1 = 5 μs (blue light) and t2 = 8 μs (red light);

[0031] S502: Fourier transform the original signal through the light source amplification channel to extract the spatial frequency f = 1 / (2Δx), Δx is the high frequency component of the pixel pitch, and the edge contrast of the surface micro mark is enhanced;

[0032] S503: Combine the position encoding data of the screw micrometer, use the structure light triangulation principle, reconstruct the bottle mouth three-dimensional point cloud, the resolution is ≤ 50 μm, and the geometric deviation of the sealing surface from the bottle mouth end face 0-1 mm area is extracted.

[0033] As preferred, when the defect intelligent identification is performed, the following steps are included:

[0034] S601: Extract 6 types of features from the three-dimensional point cloud, including texture roughness (Ra, μm level), edge sharpness (gradient amplitude ≥ threshold T1), surface continuity (curvature change rate ≤ T2), sealing surface flatness (PV value ≤ λ / 10), laser line fracture length (threshold L ≥ 0.2 mm), and refractive angle abnormal area (Δθ ≥ 0.5°);

[0035] S602: Input the feature vector into the pre-trained convolutional neural network, which contains 5 convolutional layers and 3 fully connected layers, and output the defect type probability distribution, including crack, scratch, deformation and loose sealing;

[0036] S603: Delaunay triangulation is performed on the identified defect area to generate a three-dimensional coordinate marker cloud, and the defect position (polar coordinate accuracy ± 0.1°), size (error ≤ 5%) and depth (resolution ≤ 20 μm) are accurately labeled.

[0037] As preferred, when the result output and quality traceability are performed, the following steps are included:

[0038] S701: Automatic grading according to defect type:

[0039] A: No defect or Ra <0.5 μm

[0040] B: 0.5 μm≤Ra<1.0 μm, no structural damage

[0041] C: Ra≥1.0 μm or crack / scratch depth >50 μm;

[0042] S702: If the sealing surface has a continuous defect area with an area >1mm 2 , it is automatically marked as unqualified;

[0043] S703: Generate a detection report, including a three-dimensional defect distribution heat map, key size measurement data, defect type statistical histogram and quality grade certificate;

[0044] S704: Upload the detection data hash value to the blockchain node to realize the non-tamperable detection record, and support subsequent quality traceability by scanning the code to query the original data.

[0045] The beneficial effects of the present application are:

[0046] 1、Compared with the traditional blueberry bottle mouth sealing optical nondestructive testing device, single light source or simple lens group is mostly used, it is difficult to cover the 360° annular area of the bottle mouth, leading to insufficient illumination of the edge or complex curved surface area of the bottle mouth, and small defects are easy to be missed, at the same time, it relies on monochromatic light or low resolution sensor, cannot obtain the multi-spectral reflection characteristics of the bottle mouth surface, it is difficult to distinguish the texture and defect characteristics of the material itself, and the false detection rate is high, the blueberry bottle mouth sealing optical nondestructive testing device synchronously collects multi-spectral reflection data through double-color laser line light source and Fresnel lens array, enhances the contrast of surface micro marks by using wavelength difference, effectively improves the detection sensitivity of monochromatic light, and combines structure light triangulation and screw micrometer coding data to reconstruct the three-dimensional point cloud of the bottle mouth sealing surface, realizes quantitative judgment of the flatness of multiple parameters, and greatly improves the detection accuracy;

[0047] 2, the detection light source in the optical detection device is refracted by the reflection assembly, so that the light can uniformly cover the 360° annular area of the bottle mouth, the screw rod is installed in the installation groove, the screw rod is driven to rotate by the rotating motor, the installation bracket is driven to move up and down by the screw rod, the installation ring is installed by the installation bracket, the Fresnel lens is installed by the installation ring, the light source in the optical detection device is refracted by the Fresnel lens, the Fresnel lens can effectively collect and converge light, and the small marks on the surface of the blueberry bottle mouth are refracted by light, the position distance between the installation ring and the blueberry bottle is detected by the screw micrometer, the stable and uniform light is provided by the double-color laser line light source, the double-color laser line is uniformly irradiated to the bottle mouth, the refracted light is analyzed by the analysis assembly, and whether the bottle mouth has defects is judged. BRIEF DESCRIPTION OF DRAWINGS

[0048] Figure 1 The first three-dimensional structure schematic diagram of the blueberry bottle mouth sealing optical nondestructive testing device is shown.

[0049] Figure 2 The second three-dimensional structure schematic diagram of the blueberry bottle mouth sealing optical nondestructive testing device is shown.

[0050] Figure 3 The bottom surface three-dimensional structure schematic diagram of the blueberry bottle mouth sealing optical nondestructive testing device is shown.

[0051] Figure 4 The internal three-dimensional structure schematic diagram of the blueberry bottle mouth sealing optical nondestructive testing device is shown.

[0052] Figure 5 The local three-dimensional structure schematic diagram of the blueberry bottle mouth sealing optical nondestructive testing device is shown.

[0053] Figure 6The blueberry bottle mouth sealing optical nondestructive detection method flow frame schematic diagram of the present application is shown.

[0054] Mark explanation: 1, optical detection device; 201, installation card box; 202, replaceable top plate; 203, light transmission mirror; 204, electric telescopic rod; 205, light reflection mounting plate; 206, annular micro-prism array; 301, installation slot; 302, screw rod; 303, installation support; 304, installation ring; 305, Fresnel lens; 306, screw micrometer; 307, double-color laser line light source; 308, rotating motor; 401, light transmission bottom plate; 402, CMOS sensor array; 403, light source amplification channel; 404, light source receiver; 405, nondestructive detection analyzer. DETAILED DESCRIPTION

[0055] The present application will be further described below in combination with the drawings and examples.

[0056] Please refer to Figures 1-5 , the present application provides an embodiment: blueberry bottle mouth sealing optical nondestructive detection device, including optical detection device 1, reflection component, installation slot 301, screw rod 302, installation support 303, installation ring 304, Fresnel lens 305, screw micrometer 306, double-color laser line light source 307, rotating motor 308 and analysis component, the top surface of optical detection device 1 is provided with reflection component, the inside of optical detection device 1 is provided with installation slot 301 on both sides, the inside of installation slot 301 is provided with screw rod 302, the outside of screw rod 302 is provided with installation support 303, the both ends of installation support 303 are provided with installation ring 304, the surface of installation ring 304 is provided with multiple groups of Fresnel lens 305, the array distribution of Fresnel lens 305, the side of installation support 303 is provided with screw micrometer 306, the upper side of screw micrometer 306 is provided with double-color laser line light source 307, the top surface of optical detection device 1 is provided with rotating motor 308, the side of optical detection device 1 is provided with analysis component.

[0057] Preferably, the detection light source inside the optical detection device 1 is refracted by the reflection assembly to ensure that the light uniformly covers the 360° annular area of the bottle mouth. The screw rod 302 is installed through the installation slot 301, the screw rod 302 is rotated by the rotating motor 308, the installation bracket 303 is moved up and down by the screw rod 302, the installation ring 304 is installed by the installation bracket 303, the Fresnel lens 305 is installed by the installation ring 304, the light source inside the optical detection device 1 is refracted by the Fresnel lens 305, the Fresnel lens 305 can effectively collect and converge light, and the small scratches on the surface of the blueberry bottle mouth are illuminated and refracted. The position distance between the installation ring 304 and the blueberry bottle is detected by the screw micrometer 306, the stable and uniform light is provided by the double-color laser line light source 307, the double-color laser line is uniformly irradiated to the bottle mouth, the refracted light is analyzed by the analysis assembly, and whether the bottle mouth has defects is judged.

[0058] As a preferred, the reflection assembly includes a mounting box 201, a replaceable top plate 202, a light transmission mirror 203, an electric telescopic rod 204, a reflective mounting plate 205 and an annular micro-prism array 206. The top surface of the optical detection device 1 is provided with the mounting box 201, the inside of the mounting box 201 is provided with the replaceable top plate 202, the bottom surface of the replaceable top plate 202 is provided with the light transmission mirror 203, the bottom surface of the replaceable top plate 202 is provided with the electric telescopic rod 204, the electric telescopic rod 204 is provided with multiple groups, the bottom end of the electric telescopic rod 204 is provided with the reflective mounting plate 205, and the bottom surface of the reflective mounting plate 205 is provided with the annular micro-prism array 206. In use, the replaceable top plate 202 is installed by the mounting box 201, the light transmission mirror 203 is installed by the replaceable top plate 202, the replaceable top plate 202 can be replaced at any time according to different bottle types to adapt to the needs of different bottle types, the light source inside the optical detection device 1 is refracted by the light transmission mirror 203, the reflective mounting plate 205 is lifted and lowered by the electric telescopic rod 204, the annular micro-prism array 206 is installed by the reflective mounting plate 205, and the annular micro-prism array 206 ensures that the light uniformly covers the 360° annular area of the bottle mouth.

[0059] As preferred, the analysis assembly comprises a light transmission bottom plate 401, a CMOS sensor array 402, a light source amplification channel 403, a light source receiver 404 and a non-destructive testing analyzer 405, the inner side of the optical detection device 1 is provided with the light transmission bottom plate 401, the surface of the light transmission bottom plate 401 is provided with the CMOS sensor array 402, the CMOS sensor array 402 is provided in multiple groups, one side of the optical detection device 1 is provided with the light source amplification channel 403, one end of the light source amplification channel 403 is provided with the light source receiver 404, the other side of the optical detection device 1 is provided with the non-destructive testing analyzer 405, in use, the CMOS sensor array 402 is installed through the light transmission bottom plate 401, the refracted light source is received through the CMOS sensor array 402, the electrical signal of the received light source is amplified through the light source amplification channel 403, the amplified light source is signal-converted through the light source receiver 404, and the tiny defects of the bottle mouth are detected and analyzed through the non-destructive testing analyzer 405.

[0060] Please refer to Figure 6 In the embodiment, the blueberry bottle mouth sealing optical non-destructive detection method comprises the following methods:

[0061] S101: First, initialize the system and position the bottle body;

[0062] S102: Modulate the double-color laser line light source 307, dynamically calibrate the screw micrometer 306 and the dynamic light field;

[0063] S103: Perform environment micro-prism array 206 adaptation;

[0064] S104: Collect multi-spectral data;

[0065] S105: Intelligent identification of defects of the blueberry bottle mouth;

[0066] S106: Output the detection result and trace the quality.

[0067] As preferred, during the system initialization and bottle body positioning, the following steps are included:

[0068] S201: Turn on the power of the optical detection device 1, the system automatically performs hardware self-checking, checks the power stability of the double-color laser line light source 307, the error needs to be ≤±2%, the stepping accuracy of the rotating motor 308, the stepping angle error ≤0.1°, the dark current noise of the CMOS sensor array 402, the dark current noise <0.05e- / pixel / s;

[0069] S202: The terminal inputs the blueberry bottle specification parameters, including the bottle mouth diameter, height, and material refractive index, the system automatically matches the preset annular micro-prism array 206 tilt angle, and calculates the bottle mouth curvature using the formula: θ = arcsin(n1 / n2), where n1 is the air refractive index, and n2 is the bottle material refractive index;

[0070] S203: The conveyor belt transports the blueberry bottle to the detection station, the pneumatic clamps clamp the bottle body from both sides, the laser ranging sensor measures the deviation of the bottle mouth axis from the center line of the detection device, and the six-axis correction platform fine-tunes the bottle position to ensure that the bottle mouth center and the light axis of the dual-color laser line light source 307 are coincident to within 0.1 mm.

[0071] As a preferred, when configuring the light source and dynamic calibration, the following steps are included:

[0072] S301: The dual-color laser line light source 307 emits laser lines with wavelengths λ1 = 450 nm (blue light) and λ2 = 660 nm (red light), which are expanded into an annular light curtain covering 360° of the bottle mouth through the Fresnel lens 305 array, with an optical intensity ratio of I1 / I2 = 1.5 to optimize the surface mark contrast;

[0073] S302: Rotate the screw micrometer 306 to drive the mounting ring 304 to move axially along the screw rod 302, so that the working distance between the Fresnel lens 305 and the bottle mouth end surface is adjusted to D = bottle mouth radius x 0.8, ensuring that the light incidence angle θ ≤ 30° to avoid total reflection;

[0074] S303: Rotate the motor 308 to drive the mounting bracket 303, and the CMOS sensor array 402 real-time collects the bottle mouth reflected light intensity distribution, and the system automatically adjusts the extension amount of the electric telescopic rod 204 of the light reflection mounting plate 205 through the PID algorithm, with an adjustment accuracy of ±0.02 mm, so that the standard deviation of the annular area light intensity is ≤5%.

[0075] As a preferred, when adapting the annular micro-prism array, the following steps are included:

[0076] S401: The visual recognition system scans the bottle mouth shape, selects a matching model from the replaceable top plate library, and completes the replacement of the top plate 202 within 15 seconds through the magnetic quick replacement mechanism;

[0077] S402: According to the bottle mouth thread parameters, control the electric telescopic rod 204 to adjust the height H of the light reflection mounting plate 205, so that the prism vertex of the annular micro-prism array 206 and the bottle mouth thread root are in the same phase plane;

[0078] S403: Emit test laser pulses, after passing through the light-transmitting lens 203 and the annular micro-prism array 206, form a uniformly distributed diffuse reflection spot on the bottle mouth, verify that there is no blind area in the coverage area, and the spot density is ≥200 points / cm 2 .

[0079] As preferred, when performing multi-spectral data acquisition, the following steps are included:

[0080] S501: The dual-color laser line light source 307 emits alternately at a frequency of 1 kHz, and the CMOS sensor array 402 synchronously captures the reflection images of λ1 and λ2 in a double-exposure mode, with exposure times set to t1=5μs (blue light) and t2=8μs (red light), respectively;

[0081] S502: Fourier transform the original signal through the light source amplification channel 403 to extract the spatial frequency f=1 / (2Δx), Δx being the high-frequency component of the pixel pitch, and enhance the edge contrast of the surface micro-mark;

[0082] S503: Combine the position encoding data of the screw micrometer 306, and use the structured light triangulation principle to reconstruct the three-dimensional point cloud of the bottle mouth, with a resolution of ≤50μm, and focus on extracting the geometric deviation of the sealing surface in the 0-1mm area from the bottle mouth end surface.

[0083] As preferred, when performing intelligent identification of defects, the following steps are included:

[0084] S601: Extract 6 types of features from the three-dimensional point cloud, including texture roughness (Ra, μm level), edge sharpness (gradient amplitude ≥ threshold T1), surface continuity (curvature change rate ≤ T2), sealing surface flatness (PV value ≤ λ / 10), laser line break length (threshold L ≥ 0.2mm), and refraction angle abnormal area (Δθ ≥ 0.5°);

[0085] S602: Input the feature vector into the pre-trained convolutional neural network, which contains 5 convolutional layers and 3 fully connected layers, and output the defect type probability distribution, including cracks, scratches, deformations, and loose sealing;

[0086] S603: Perform Delaunay triangulation on the identified defect area to generate a three-dimensional coordinate marker cloud, accurately marking the defect position (polar coordinate accuracy ±0.1°), size (error ≤5%), and depth (resolution ≤20μm).

[0087] As preferred, when performing result output and quality tracing, the following steps are included:

[0088] S701: Automatically grade according to the defect type:

[0089] A: No defects or Ra<0.5μm

[0090] B: 0.5 pm < Ra < 1.0 pm, no structural damage

[0091] C: Ra≥ 1.0 pm or crack / scratch depth > 50 pm;

[0092] S702: If the sealing surface has a continuous defect area > 1 mm 2 , automatically marked as unqualified;

[0093] S703: Generate a detection report, including a three-dimensional defect distribution heat map, key size measurement data, defect type statistical histogram, and quality grade certificate;

[0094] S704: Upload the detection data hash value to the blockchain node to realize tamper-proof detection records and support subsequent quality traceability by scanning the code to query the original data.

[0095] Effectiveness verification of blueberry bottle mouth sealing optical nondestructive detection device and method

[0096] 1. Experimental design

[0097] Objective: To verify the recognition ability of the six-step detection process for blueberry bottle mouth sealing defects, and to quantify the improvement of detection accuracy and efficiency.

[0098] Sample: Bottle type: 3 specifications (diameter 38 mm / 42 mm / 48 mm, height 80 mm / 100 mm / 120 mm)

[0099] Material: Sodium calcium glass (n = 1.52), high borosilicate glass (n = 1.47), PET plastic (n = 1.58)

[0100] Artificial defects: Sealing surface scratch (depth 20 pm / 50 pm / 80 pm)

[0101] Edge crack (length 0.3 mm / 0.5 mm / 1.0 mm)

[0102] Thread deformation (offset 0.1 mm / 0.2 mm / 0.3 mm)

[0103] Simulated poor sealing (gap 50 pm / 100 pm / 150 pm)

[0104] 2. Equipment configuration

[0105]

[0106]

[0107] 3. Experimental steps

[0108] S801: input bottle type parameters: diameter 42 mm, height 100 mm, material refractive index 1.52 (corresponding to high borosilicate glass); six-axis correction platform adjustment time: 8.2 seconds (bottle opening center and optical axis coincidence 0.05 mm);

[0109] S802: ring light curtain coverage angle: 360°±0.3° (verified by laser power meter); Fresnel lens working distance D = 16.8 mm (bottle opening radius 21 mm x 0.8); light intensity standard deviation: 4.1% (after PID adjustment);

[0110] S803: top plate replacement time: 12 seconds; prism vertex and thread root phase difference: 0.02 mm (measured by laser interferometer); diffuse reflection spot density: 220 points / cm 2 (uniformity 98.7%);

[0111] S804: double exposure image synchronization error: 1.5 μs; three-dimensional point cloud reconstruction time: 0.8 seconds / bottle; sealing surface geometric deviation detection accuracy: ±8 μm;

[0112] S805: CNN training data set: 20,000 defect images; test set recognition rate: 99.2% for cracks, 98.5% for scratches, 97.8% for deformation, and 99.5% for tightness; defect positioning error: position ±0.08°, size error 4.2%, depth resolution 15 μm;

[0113] S806: detection report generation time: 3.1 seconds / bottle; quality grade determination accuracy: 100%.

[0114] 4. Results analysis

[0115]

[0116]

[0117] 5. Conclusion

[0118] The annular micro-prism array + dynamic light field calibration technology reduces the detection blind area from 12% in the traditional method to 0.3%, the dual-color laser line fusion detection improves the micro-defect recognition rate by 42%, the blockchain traceability function shortens the quality traceability time by 80%, the detection efficiency is improved by 5-8 times, and the sealing determination accuracy is ≥99.8%.

[0119] The embodiments of the application are described in detail above in combination with the drawings, but the application is not limited to the above-described embodiments, and various changes can be made within the knowledge of those skilled in the art without departing from the purpose of the application.

Claims

1. Optical non-destructive testing device for blueberry bottle sealing; characterized by: The invention comprises an optical detection device (1), a reflection component, a mounting groove (301), a screw (302), a mounting bracket (303), a mounting ring (304), a Fresnel lens (305), a micrometer screw (306), a two-color laser line light source (307), a rotating motor (308) and an analysis component. The top surface of the optical detection device (1) is provided with a reflection component. Both sides of the interior of the optical detection device (1) are provided with mounting grooves (301). The interior of the mounting groove (301) is provided with a screw (302). The outer side of the screw (302) is provided with a mounting bracket (303). Both ends of the mounting bracket (303) are provided with mounting rings (304). The surface of the mounting ring (304) is provided with multiple groups of Fresnel lenses (305). The Fresnel lenses (305) are distributed in an array. A micrometer screw (306) is provided on one side of the mounting bracket (303). A two-color laser line light source (307) is provided above the micrometer screw (306). The optical detection device (1) is provided with a rotating motor (308) on the top surface, and an analysis component is provided on one side of the optical detection device (1); the reflection component includes a mounting card box (201), a replaceable top plate (202), a light-transmitting mirror (203), an electric telescopic rod (204), a reflective mounting plate (205) and an annular microprism array (206); the ... optical detection device (1) is provided with a rotating motor (308) on the top surface, and an analysis component is provided on the side of the optical detection device (1); the optical detection device (1) is provided with a rotating motor (308) on the top surface, and an analysis component is provided on the side of the optical detection device (1); the optical detection device (1) is provided with a rotating motor (308) on the top surface, and an analysis component is provided on the side of the optical detection device (1); the optical detection device (1) is provided with a rotating motor (308) on the top surface, and A mounting card box (201) is provided, a replaceable top plate (202) is provided inside the mounting card box (201), a light-transmitting mirror (203) is provided on the bottom surface of the replaceable top plate (202), an electric telescopic rod (204) is provided on the bottom surface of the replaceable top plate (202), a plurality of electric telescopic rods (204) are provided, a reflective mounting plate (205) is provided at the bottom end of the electric telescopic rod (204), and an annular microprism array (206) is provided on the bottom surface of the reflective mounting plate (205).

2. The optical nondestructive testing device for blueberry bottle sealing according to claim 1, characterized in that: The analysis component comprises a light-transmitting base plate (401), a CMOS sensor array (402), a light source amplifying channel (403), a light source receiver (404) and a non-destructive testing analyzer (405); the light-transmitting base plate (401) is provided on the inner side of the optical detection device (1); the CMOS sensor array (402) is provided on the surface of the light-transmitting base plate (401); the CMOS sensor array (402) is provided in multiple groups; the light source amplifying channel (403) is provided on one side of the optical detection device (1); the light source receiver (404) is provided at one end of the light source amplifying channel (403); and the non-destructive testing analyzer (405) is provided on the other side of the optical detection device (1).

3. A detection method using the optical nondestructive testing device for blueberry bottle sealing according to claim 1 or 2, characterized in that: The following steps are included: S101: First, initialize the system and position the bottle; S102: The two-color laser line light source (307) is modulated, and the micrometer (306) and the dynamic light field are dynamically calibrated; S103: performing annular microprism array (206) adaptation; S104: Collecting multispectral data; S105: Intelligently identify defects on the mouth of the blueberry bottle; S106: Output the test results and trace the quality; The following steps are included when configuring and calibrating the light source: S301: a two-color laser line light source (307) emits a laser line with a wavelength of λ1 = 450 nm blue light and λ2 = 660 nm red light, and the light intensity ratio is set to I1 / I2 = 1.5 to optimize the surface micro-trace contrast; S302: rotating the micrometer screw (306), driving the mounting ring (304) to move axially along the screw (302), so that the working distance between the Fresnel lens (305) and the bottle end face is adjusted to D = bottle radius × 0.8, ensuring that the light incident angle θ ≤ 30° to avoid total reflection; S303: The motor (308) is rotated to drive the mounting bracket (303), and the CMOS sensor array (402) collects the intensity distribution of the light reflected from the bottle mouth in real time. The system automatically adjusts the extension and contraction of the electric telescopic rod (204) of the reflective mounting plate (205) through the PID algorithm, with an adjustment accuracy of ±0.02mm, so that the standard deviation of the light intensity in the annular area is ≤5%.

4. The detection method according to claim 3, wherein: When performing system initialization and bottle positioning, the following steps are included: S201: Turn on the power of the optical detection device (1), and the system automatically performs a hardware self-test to check the power stability of the dual-color laser line light source (307), the error is ≤±2%, the step accuracy of the rotating motor (308), the step angle error is ≤0.1°, and the dark current noise of the CMOS sensor array (402), the dark current noise is <0.05e⁻ / pixel / s; S202: Input the blueberry bottle specification parameters through the control terminal, including the bottle diameter, height and material refractive index, and the system automatically matches the preset annular microprism array (206) tilt angle; S203: The conveyor belt transports the blueberry bottle to the inspection station, the pneumatic grippers clamp the bottle body from both sides, the laser distance sensor measures the deviation between the axis of the bottle mouth and the center line of the inspection device, and the six-axis correction platform fine-tunes the position of the bottle body to ensure that the overlap between the center of the bottle mouth and the optical axis of the two-color laser line light source (307) is ≤0.1mm.

5. The detection method according to claim 3, wherein: When adapting the annular microprism array, the following steps are included: S401: The visual recognition system scans the bottle mouth shape, selects a matching model from the replaceable top plate library, and replaces the top plate (202) within 15 seconds through a magnetic quick-change mechanism; S402: According to the bottle thread parameters, control the electric telescopic rod (204) to adjust the reflective mounting plate (205) height H, so that the prism vertex of the annular microprism array (206) and the root of the bottle thread are in the same phase plane; S403: Emitting a test laser pulse, after passing through the light-transmitting mirror (203) and the annular microprism array (206), forming evenly distributed diffuse reflection spots at the bottle mouth, verifying that there is no blind spot in the coverage area and the spot density is ≥200 points / cm².

6. The detection method according to claim 3, wherein: When collecting multispectral data, the following steps are included: S501: The dual-color laser line light source (307) emits alternately at a frequency of 1 kHz, and the CMOS sensor array (402) uses a double exposure mode to synchronously capture the reflection images of λ1 and λ2, with the exposure time set to t1 = 5 μs for blue light and t2 = 8 μs for red light, respectively; S502: Performing Fourier transform on the original signal through the light source amplification channel (403), extracting the spatial frequency f=1 / (2Δx), where Δx is the high-frequency component of the pixel spacing, and enhancing the edge contrast of the surface micro-scratches; S503: Combining the position coding data of the micrometer screw (306), using the principle of structured light triangulation, reconstructing the three-dimensional point cloud of the bottle mouth with a resolution of ≤50μm, focusing on extracting the geometric deviation of the sealing surface in the 0-1mm area from the bottle mouth end face.

7. The detection method according to claim 3, wherein: When performing intelligent defect identification, the following steps are included: S601: Extract six types of features from the 3D point cloud, including texture roughness, edge sharpness, surface continuity, sealing surface flatness, laser line break length, and refraction angle abnormality area; S602: Input the feature vector into the pre-trained convolutional neural network, which contains 5 convolutional layers and 3 fully connected layers, and outputs the probability distribution of defect types, including four categories: cracks, scratches, deformation, and poor sealing; S603: Perform Delaunay triangulation on the identified defect area to generate a three-dimensional coordinate tag cloud to accurately mark the defect position, size and depth. The polar coordinate accuracy of the defect position is ±0.1°, the size error is ≤5%, and the depth resolution is ≤20μm.

8. The detection method according to claim 3, wherein: When outputting results and tracing quality, the following steps are included: S701: Automatic classification according to defect type: Grade A: no defects or Ra<0.5μm Class B: 0.5μm≤Ra<1.0μm, no structural damage Grade C: Ra ≥ 1.0 μm or crack / scratch depth > 50 μm; S702: If the sealing surface has a continuous defect area greater than 1mm², it will be automatically marked as unqualified; S703: Generate an inspection report, including a three-dimensional defect distribution heat map, key dimension measurement data, a defect type statistical histogram, and a quality grade certificate; S704: Upload the hash value of the test data to the blockchain node to ensure that the test record cannot be tampered with, and support the query of the original data by scanning the code during subsequent quality traceability.

Images