Blueberry bottle sealing quality detection device and method based on spectral analysis

Abstract

This invention relates to the field of sealing function testing technology, and particularly to a blueberry bottle sealing quality testing device and method based on spectral analysis. Technical problem: Existing methods for testing the sealing performance of bottled foods using pressure and vacuum methods require opening the bottle for sampling, resulting in product waste and inability to perform full inspection. Existing methods have low detection rates for micropore leaks and cannot detect internal micro-cracks, and exhibit severe baseline drift under high temperature and humidity conditions. Technical solution: The blueberry bottle sealing quality testing device and method based on spectral analysis includes a support, conveyor belt, limiting groove, blueberry bottle body, probe assembly, window assembly, environmental sensing assembly, calibration assembly, cleaning assembly, and edge computing assembly. This invention achieves non-destructive, high-precision testing of blueberry bottle sealing performance through non-contact detection, a self-cleaning mechanism, and dynamic environmental compensation, reducing product waste, improving testing accuracy, and ensuring stable operation even under surface contamination conditions.

Description

Technical Field

[0001] This invention relates to the field of sealing function testing technology, and in particular to a blueberry bottle sealing quality testing device and method based on spectral analysis. Background Technology

[0002] Currently, the sealing performance testing of bottled foods mainly relies on pressure decay, vacuum decay, visual inspection, and spectroscopic methods. Among them, the pressure decay method detects leaks by measuring changes in pressure inside the bottle, but it requires destructive sampling and is not sensitive to micron-level leaks. The vacuum decay method uses a vacuum chamber to detect the pressure recovery rate, but the detection speed is slow and it cannot locate the leak point. The visual inspection method identifies defects in the bottle opening shape using a camera, but it is affected by label reflection and surface contamination, resulting in a high false detection rate. Traditional spectroscopic methods are based on the detection of single gas absorption peaks, but they are easily affected by temperature drift and scattering from the bottle material, resulting in insufficient sensitivity.

[0003] Existing methods for testing the sealing performance of bottled foods using pressure and vacuum methods require opening the bottle for sampling, resulting in product waste and inability to perform full inspection. Existing methods have a low detection rate for micropore leakage and cannot detect internal micro-cracks. Visual inspection is affected by light and label interference, and traditional spectroscopic methods suffer from severe baseline drift under high temperature and high humidity conditions.

[0004] Therefore, to address the above problems, a blueberry bottle sealing quality testing device and method based on spectral analysis is proposed. Through non-contact testing, a self-cleaning mechanism, and dynamic environmental compensation, it achieves non-destructive and high-precision testing of blueberry bottle sealing, reduces product loss, improves testing accuracy, and can operate stably under surface contamination conditions. Summary of the Invention

[0005] To overcome the problems of existing methods for testing the sealing performance of bottled food, which require opening the bottle for sampling and inspection when using pressure and vacuum methods, resulting in product waste and inability to conduct full inspection, the existing methods have low detection rates for micropore leakage and cannot detect internal micro-cracks. Visual inspection is affected by light and label interference, and traditional spectroscopic methods suffer from severe baseline drift in high temperature and high humidity environments.

[0006] The technical solution of this invention is as follows: a blueberry bottle sealing quality detection device based on spectral analysis, comprising a support, a conveyor belt, a limiting groove, a blueberry bottle body, a probe assembly, a window assembly, an environmental sensing assembly, a calibration assembly, a cleaning assembly, and an edge computing assembly. A conveyor belt is positioned above the support, and the conveyor belt has a transverse conveying structure. Multiple sets of limiting grooves are evenly distributed on the surface of the conveyor belt. The blueberry bottle body is positioned inside the limiting groove. A probe assembly is positioned above the support, a window assembly is positioned at the bottom of the probe assembly, an environmental sensing assembly is positioned on the side of the probe assembly, a calibration assembly is positioned below the probe assembly, a cleaning assembly is positioned below the probe assembly, and an edge computing assembly is positioned on one side of the support. The probe assembly includes a support arm, a probe housing, a near-infrared light source, a programmable light field modulator, a liquid crystal panel, a beam splitter prism, and a spectral camera. The support arm is positioned above the support and is vertically fixed directly above the detection station of the support. The probe housing is positioned below the support arm, and a near-infrared light source is located inside the probe housing. Located on the inner top surface of the probe housing, the near-infrared light source emits wavelengths of 900-2500nm. The power of the near-infrared light source is controlled by a pulse modulation circuit. A programmable light field modulator is located inside the probe housing, in the middle section, directly below the near-infrared light source. An LCD panel is located inside the programmable light field modulator, perpendicular to the optical axis of the near-infrared light source. A beam splitter is also located inside the probe housing, fixed to the bottom of the housing with optical adhesive. The beam splitter is coaxially aligned with the optical path exit of the programmable light field modulator. A spectroscopic camera is located on the inner side wall of the probe housing, connected to the beam splitter via an optical fiber bundle. The reflective surface of the beam splitter and the lens of the spectroscopic camera form a 45° angle. The window assembly includes an optical window and a hydrophobic coating. An optical window is located at the bottom of the probe housing, covering the bottom opening. The surface of the optical window is coated with a hydrophobic coating, which is a graphene superhydrophobic coating with a thickness of 200±10nm.

[0007] Preferably, pulsed light emitted from a near-infrared light source penetrates the sealing material, exciting the characteristic absorption spectrum of gas molecules inside the blueberry bottle. A programmable light field modulator and a liquid crystal panel dynamically correct optical path distortion and enhance the effective signal intensity by loading a Zernike polynomial phase compensation pattern. A beam splitter and a spectroscopic camera separate the reflected light by wavelength to generate a hyperspectral image cube, capturing the molecular absorption characteristics of the leak area. Organic residues are degraded through photocatalytic decomposition, achieving self-cleaning of the detection window, improving window transmittance, and preventing the increase in false detection rate caused by contaminant residues.

[0008] Preferably, the environmental sensing component includes a temperature and humidity sensor and a camera. The temperature and humidity sensor is located on one side of the probe housing, with the detection end of the temperature and humidity sensor facing the bottle opening area of ​​the blueberry bottle. The camera is located on the other side of the probe housing. The camera is a 3D ToF camera, and the field of view of the camera covers the bottle opening area of ​​the blueberry bottle.

[0009] Preferably, environmental parameters in the bottle opening area are collected in real time by temperature and humidity sensors to provide a data basis for spectral compensation, and the leak area is located by scanning the bottle opening threads and sealing film morphology with a 3D ToF camera and combining the spectral data.

[0010] Preferably, the calibration assembly includes a calibration plate, a quantum dot film, and a lifting cylinder. The calibration plate is located directly below the optical window, and a quantum dot film, specifically a CdSe / ZnS quantum dot film, is deposited on the surface of the calibration plate. A lifting cylinder is located below the calibration plate and is positioned above the support. The lifting stroke of the calibration plate is 0-50 mm.

[0011] Preferably, after a certain number of blueberry bottles are inspected, the lifting cylinder is activated to raise the calibration plate, collect the quantum dot film reflection spectrum, compare the current spectrum with the initial reference, and if the deviation is >0.5%, an alarm is triggered and recalibration is performed. This provides a stable reference signal through the quantum dot film, eliminates wavelength drift caused by equipment aging, and adjusts the position of the calibration plate through the lifting cylinder to adapt to different bottle heights.

[0012] Preferably, the cleaning component includes an airflow ring, jet holes, an air pipe, and an air pump. An airflow ring is provided below the optical window. The airflow ring has a circular structure and surrounds the outer periphery of the optical window. Jet holes are provided on the inner side of the airflow ring. Multiple sets of jet holes are provided and are evenly distributed circumferentially on the inner side of the airflow ring. The axis of the jet holes forms a 30° angle with the plane of the optical window. An air pipe is provided on one side of the airflow ring, and an air pump is provided at one end of the air pipe. The air pump is located on one side of the support.

[0013] Preferably, when contamination is detected, the air pump starts pulse jets, which are introduced into the jet holes along the air pipe to clean the optical window, remove fruit residue and condensate, and control the near-infrared light source to continuously irradiate the optical window, thereby stimulating the photocatalytic decomposition reaction of the graphene hydrophobic coating and avoiding mechanical wiping that could damage the optical window.

[0014] Preferably, the edge computing component includes a heat sink, a motherboard, and a fan. The heat sink is located on one side of the bracket, the motherboard is located inside the heat sink, and the fan is located on the inner side of the heat sink. The GPIO interface of the motherboard is connected to the spectral camera, the temperature and humidity sensor, and the camera respectively through shielded cables.

[0015] A method for detecting the sealing quality of blueberry bottles based on spectral analysis includes the following steps:

[0016] S1: The start-up conveyor belt moves the blueberry bottle to the inspection station, the near-infrared light source is activated to emit pulsed light, the phase compensation pattern generated by the Zernike polynomial is loaded through the LCD panel of the programmable light field modulator, the incident light wavefront is adjusted, the reflected light is guided into the spectroscopic camera through the beam splitter, and the hyperspectral image cube of the bottle mouth area is acquired at a rate of 10ms / frame, wherein the spatial resolution of the hyperspectral image cube is 512×512 pixels and the spectral resolution is 5nm;

[0017] S2: Perform Fourier transform analysis on the diffraction patterns of the hyperspectral image to extract spatial rating features. If the frequency peak is in the range of 10-50 lp / mm, it is determined to be fruit pomace contamination. If the frequency peak is <10 lp / mm and the spectral reflectance is >85%, it is determined to be condensate contamination. When organic pollutants are identified, control the near-infrared light source to switch to a wavelength of 1064nm and continuously irradiate the optical window with a power of 50mW for 30 seconds to excite the photocatalytic decomposition reaction of the graphene hydrophobic coating.

[0018] S3: Read the time-series data from the temperature and humidity sensor, input it into the pre-trained Transformer-CNN hybrid network model, and extract the spectral spatial features of the bottle mouth region through 3 convolutional layers, where the convolutional kernel is 3×3 and the stride is 1; capture the correlation between temperature and humidity fluctuations and spectral drift through the Transformer encoder, where the Transformer encoder has 8 attention heads and 256 hidden layer dimensions; then output a pixel-wise compensation coefficient matrix with a resolution of 256×256, and multiply the compensation coefficient matrix with the original spectral data pixel by pixel to correct the baseline drift;

[0019] S4: Perform partial least squares discriminant analysis on the compensated spectral data to extract the absorption peak intensity of O2 at 760nm, and calculate the O2 permeability based on Fick's law:

[0020]

[0021] Where D is the O2 diffusion coefficient, A is the leakage area, and P in and P out The values ​​are the internal and external air pressures of the bottle, L is the thickness of the sealing film, the diffusion coefficient is obtained from a pre-stored bottle mouth material database, and the leakage area is calculated by scanning the bottle mouth morphology with a camera. If the calculated permeability value is greater than the preset threshold, it is determined to be a sealing failure.

[0022] S5: Generates a sealing quality report and sends it to the production line PLC via the Modbus protocol. At the same time, it links with an external robotic arm to grab defective bottles and laser-engraves the leakage coordinates on the bottle label.

[0023] Preferably, in step S1, the method for generating the Zernike polynomial phase compensation pattern is as follows:

[0024] S101: Calculate the wavefront distortion function W(x, y) caused by pollutant scattering.

[0025] S102: Expanded into a 36-term Zernike polynomial according to the following formula:

[0026]

[0027] Among them, Z k (x,y) are Zernike basis functions, a k These are the fitting coefficients;

[0028] S103: Load the coefficient matrix into the programmable optical field modulator.

[0029] Preferably, in step S3, the training method for the Transformer-CNN hybrid network model specifically includes:

[0030] S301: Construct an adversarial training dataset to simulate extreme environmental scenarios, specifically:

[0031] Temperature step change: -15℃ to +15℃, gradient 2℃ / s;

[0032] Humidity step change: from 30%RH to 90%RH, gradient 20%RH / s;

[0033] S302: A weighted loss function is used, specifically:

[0034] L=α·MSE(y pred ,y true )+β·CrossEntropy(c pred ,c true );

[0035] Where α = 0.7, β = 0.3, y is the spectral intensity, and c is the sealing status classification label.

[0036] Preferably, in step S4, the leakage area A is calculated as follows:

[0037] S401: Acquires point cloud data of the bottle opening via a 3D ToF camera;

[0038] S402: The random sampling consensus algorithm is used to fit the bottle neck plane;

[0039] S403: Calculate the deviation between the point cloud and the fitted plane. If the local deviation is >0.05mm, it is determined to be a leakage area. The total area A is calculated cumulatively.

[0040] Preferably, in step S5, the coordinate positioning method for laser engraving is as follows:

[0041] S501: Convert the coordinates of the pixel with the highest leakage signal in the hyperspectral image to the machine coordinate system;

[0042] S502: Mapped to the robotic arm's operating space via a hand-eye calibration matrix.

[0043] The beneficial effects of this invention are:

[0044] 1. This invention emits broadband light from a near-infrared light source, which penetrates the bottle sealing material. The phase of the incident light field is dynamically adjusted by a light field modulator to eliminate label reflection interference. A spectral camera collects the reflected light signal to generate a hyperspectral image cube. Temperature and humidity data are collected in real time by an environmental sensing component and input into a Transformer-CNN hybrid network model. The model outputs a pixel-by-pixel compensation coefficient matrix. The compensated spectral data is used to extract the O2 absorption peak intensity using the PLS-DA algorithm and calculates the permeability using Fick's law, avoiding physical contact with the bottle. This achieves non-contact, non-destructive testing, eliminating the need for opening the bottle or sampling to reduce product loss.

[0045] 2. This invention improves the effective signal intensity by loading a Zernike polynomial phase compensation pattern through an optical field modulator; it uses a quantum dot film as an embedded reference to calibrate the wavelength drift of the light source in real time; and it compensates for nonlinear environmental interference by constructing an adversarial training dataset to train a Transformer-CNN model. This improves detection accuracy, increases the detection rate of micropore leakage, and reduces spectral baseline drift caused by temperature and humidity fluctuations.

[0046] 3. This invention removes surface contaminants through the synergistic effect of the airflow ring and the hydrophobic coating, and degrades organic residues through photocatalytic decomposition, thereby achieving self-cleaning of the detection window, improving the window's light transmittance, and preventing the increase in false detection rate caused by contaminant residues. Attached Figure Description

[0047] Figure 1 The diagram shown is a three-dimensional structural schematic of the blueberry bottle sealing quality detection device based on spectral analysis according to the present invention.

[0048] Figure 2 The diagram shown is a three-dimensional cross-sectional view of the blueberry bottle sealing quality detection device based on spectral analysis according to the present invention.

[0049] Figure 3 The diagram shown is a partial three-dimensional structural schematic of the blueberry bottle sealing quality detection device based on spectral analysis of the present invention.

[0050] Figure 4The diagram shown is a three-dimensional structural schematic of the blueberry bottle sealing quality detection device based on spectral analysis according to the present invention.

[0051] Figure 5 The diagram shown is a flowchart illustrating the steps of the blueberry bottle sealing quality detection method based on spectral analysis of the present invention.

[0052] Explanation of reference numerals in the attached drawings: 1. Support frame; 2. Conveyor belt; 3. Blueberry bottle body; 201. Limiting groove; 103. Support arm; 402. Probe housing; 401. Near-infrared light source; 403. Programmable light field modulator; 406. LCD panel; 404. Beam splitter prism; 405. Spectrometer camera; 501. Optical window; 502. Hydrophobic coating; 601. Temperature and humidity sensor; 602. Camera; 701. Calibration plate; 702. Quantum dot film; 703. Lifting cylinder; 801. Airflow ring; 802. Jet nozzle; 803. Air pipe; 804. Air pump; 901. Heat sink; 902. Mainboard; 903. Fan. Detailed Implementation

[0053] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0054] Please see Figure 1 , Figure 2 , Figure 3 and Figure 4This invention provides an embodiment of a blueberry bottle sealing quality detection device based on spectral analysis, comprising a support 1, a conveyor belt 2, a limiting groove 201, a blueberry bottle body 3, a probe assembly, a window assembly, an environmental sensing assembly, a calibration assembly, a cleaning assembly, and an edge computing assembly. The conveyor belt 2 is positioned above the support 1 and has a transverse conveying structure. Multiple sets of limiting grooves 201 are evenly distributed on the surface of the conveyor belt 2. The blueberry bottle body 3 is positioned inside the limiting groove 201. The probe assembly is positioned above the support 1, and a window assembly is positioned at the bottom of the probe assembly. A window assembly is positioned on the side of the probe assembly. The environmental sensing component includes a calibration component and a cleaning component located below the probe component. An edge computing component is located on one side of the support 1. The probe component includes a support arm 103, a probe housing 402, a near-infrared light source 401, a programmable light field modulator 403, a liquid crystal panel 406, a beam splitter 404, and a spectrometer 405. The support arm 103 is located above the support 1 and is vertically fixed directly above the detection station of the support 1. The probe housing 402 is located below the support arm 103. The near-infrared light source 401 is located inside the probe housing 402, on its inner top surface. The infrared light source 401 emits wavelengths from 900 to 2500 nm. The power of the near-infrared light source 401 is controlled by a pulse modulation circuit. A programmable light field modulator 403 is located inside the probe housing 402, in the middle section, directly below the near-infrared light source 401. A liquid crystal panel 406 is located inside the programmable light field modulator 403, perpendicular to the optical axis of the near-infrared light source 401. A beam splitter prism 404 is located inside the probe housing 402, fixed to the bottom of the probe housing 402 with optical adhesive. The prism 404 is coaxially aligned with the optical path exit of the programmable light field modulator 403. A spectroscopic camera 405 is disposed on the inner sidewall of the probe housing 402. The spectroscopic camera 405 is connected to the beam splitter 404 through an optical fiber bundle. The reflective surface of the beam splitter 404 and the lens of the spectroscopic camera 405 form a 45° angle. The window assembly includes an optical window 501 and a hydrophobic coating 502. The optical window 501 is disposed at the bottom of the probe housing 402 and covers the bottom opening of the probe housing 402. The surface of the optical window 501 is coated with a hydrophobic coating 502, which is a graphene superhydrophobic coating 502 with a coating thickness of 200±10nm.

[0055] Preferably, the near-infrared light source 401 emits pulsed light that penetrates the sealing material, exciting the characteristic absorption spectrum of gas molecules inside the blueberry bottle. The programmable light field modulator 403 and the liquid crystal panel 406 dynamically correct the optical path distortion and improve the effective signal intensity by loading a Zernike polynomial phase compensation pattern. The reflected light is separated by wavelength by the beam splitter prism 404 and the spectroscopic camera 405 to generate a hyperspectral image cube, capturing the molecular absorption characteristics of the leak area. The organic residue is degraded by photocatalysis, achieving self-cleaning of the detection window, improving the window transmittance, and preventing the increase in false detection rate caused by pollutant residue.

[0056] Preferably, the environmental sensing component includes a temperature and humidity sensor 601 and a camera 602. The temperature and humidity sensor 601 is disposed on one side of the probe housing 402, with the detection end of the temperature and humidity sensor 601 facing the bottle mouth area of ​​the blueberry bottle body 3. The camera 602 is disposed on the other side of the probe housing 402. The camera 602 is a 3D ToF camera 602, and the field of view of the camera 602 covers the bottle mouth area of ​​the blueberry bottle body.

[0057] Preferably, the environmental parameters of the bottle opening area are collected in real time by the temperature and humidity sensor 601 to provide a data basis for spectral compensation, and the leak area is located by scanning the bottle opening thread and sealing film morphology by the 3D ToF camera 602 and combining the spectral data.

[0058] Preferably, the calibration assembly includes a calibration plate 701, a quantum dot film 702, and a lifting cylinder 703. The calibration plate 701 is located directly below the optical window 501. The surface of the calibration plate 701 is coated with a quantum dot film 702, specifically a CdSe / ZnS quantum dot film 702. The quantum dot film 702 is deposited on the surface of the calibration plate 701. The lifting cylinder 703 is located below the calibration plate 701 and is positioned above the support 1. The lifting stroke of the calibration plate 701 is 0-50mm.

[0059] Preferably, after a certain number of blueberry bottles are inspected, the lifting cylinder 703 is activated to raise the calibration plate 701, collect the reflection spectrum of the quantum dot film 702, compare the current spectrum with the initial reference, and if the deviation is >0.5%, an alarm is triggered and recalibration is performed. This provides a stable reference signal through the quantum dot film 702, eliminates wavelength drift caused by equipment aging, and adjusts the position of the calibration plate 701 through the lifting cylinder 703 to adapt to different bottle heights.

[0060] Preferably, the cleaning component includes an airflow ring 801, a jet hole 802, an air pipe 803, and an air pump 804. The airflow ring 801 is disposed below the optical window 501. The airflow ring 801 has a circular structure and surrounds the outer periphery of the optical window 501. A jet hole 802 is provided on the inner side of the airflow ring 801. Multiple sets of jet holes 802 are provided and are evenly distributed circumferentially on the inner side of the airflow ring 801. The axis of the jet hole 802 forms a 30° angle with the plane of the optical window 501. An air pipe 803 is disposed on one side of the airflow ring 801, and an air pump 804 is disposed at one end of the air pipe 803. The air pump 804 is located on one side of the bracket 1.

[0061] Preferably, when contamination is detected, the air pump 804 starts pulse jets, which are introduced into the jet hole 802 along the air pipe 803 to clean the optical window 501, remove fruit residue and condensate, and control the near-infrared light source 401 to continuously irradiate the optical window 501, thereby stimulating the photocatalytic decomposition reaction of the graphene hydrophobic coating 502 and avoiding mechanical wiping damage to the optical window 501.

[0062] Preferably, the edge computing component includes a heat sink 901, a motherboard 902, and a fan 903. The heat sink 901 is located on one side of the bracket 1. The motherboard 902 is located inside the heat sink 901. The fan 903 is located inside the heat sink 901. The GPIO interface of the motherboard 902 is connected to the spectral camera 405, the temperature and humidity sensor 601, and the camera 602 respectively via shielded cables.

[0063] Please see Figure 5 A method for detecting the sealing quality of blueberry bottles based on spectral analysis includes the following steps:

[0064] S1: The start of the conveyor belt 2 moves the blueberry bottle 3 to the detection station, the start of the near-infrared light source 401 emits pulsed light, the phase compensation pattern generated by the Zernike polynomial is loaded through the liquid crystal panel 406 of the programmable light field modulator 403, the incident light field wavefront is adjusted, the reflected light is guided into the spectroscopic camera 405 through the beam splitter prism 404, and the hyperspectral image cube of the bottle mouth area is acquired at a rate of 10ms / frame, wherein the spatial resolution of the hyperspectral image cube is 512×512 pixels and the spectral resolution is 5nm;

[0065] S2: Perform Fourier transform analysis on the diffraction patterns of the hyperspectral image to extract spatial rating features. If the frequency peak is in the range of 10-50 lp / mm, it is determined to be fruit pomace contamination. If the frequency peak is <10 lp / mm and the spectral reflectance is >85%, it is determined to be condensate contamination. When organic pollutants are identified, control the near-infrared light source 401 to switch to a wavelength of 1064nm and continuously irradiate the optical window 501 with a power of 50mW for 30 seconds to excite the photocatalytic decomposition reaction of the graphene hydrophobic coating 502.

[0066] S3: Read the time-series data from the temperature and humidity sensor 601, input it into the pre-trained Transformer-CNN hybrid network model, and extract the spectral spatial features of the bottle mouth region through 3 convolutional layers, where the convolutional kernel is 3×3 and the stride is 1; capture the correlation between temperature and humidity fluctuations and spectral drift through the Transformer encoder, where the Transformer encoder has 8 attention heads and 256 hidden layer dimensions; then output a pixel-wise compensation coefficient matrix with a resolution of 256×256, and multiply the compensation coefficient matrix with the original spectral data pixel by pixel to correct the baseline drift;

[0067] S4: Perform partial least squares discriminant analysis on the compensated spectral data to extract the absorption peak intensity of O2 at 760nm, and calculate the O2 permeability based on Fick's law:

[0068]

[0069] Where D is the O2 diffusion coefficient, A is the leakage area, and P in and P out The values ​​are the internal and external air pressures of the bottle, L is the thickness of the sealing film, the diffusion coefficient is obtained from a pre-stored bottle mouth material database, and the leakage area is calculated by scanning the bottle mouth morphology with a 602 camera; if the calculated permeability value is obtained, it is determined that the seal has failed.

[0070] S5: Generates a sealing quality report and sends it to the production line PLC via the Modbus protocol. At the same time, it links with an external robotic arm to grab defective bottles and laser-engraves the leakage coordinates on the bottle label.

[0071] Preferably, in step S1, the method for generating the Zernike polynomial phase compensation pattern is as follows:

[0072] S101: Calculate the wavefront distortion function W(x, y) caused by pollutant scattering.

[0073] S102: Expanded into a 36-term Zernike polynomial according to the following formula:

[0074]

[0075] Among them, Z k (x,y) are Zernike basis functions, a k These are the fitting coefficients;

[0076] S103: Load the coefficient matrix into the programmable optical field modulator 403.

[0077] Preferably, in step S3, the training method for the Transformer-CNN hybrid network model specifically includes:

[0078] S301: Construct an adversarial training dataset to simulate extreme environmental scenarios, specifically:

[0079] Temperature step change: -15℃ to +15℃, gradient 2℃ / s;

[0080] Humidity step change: from 30%RH to 90%RH, gradient 20%RH / s;

[0081] S302: A weighted loss function is used, specifically:

[0082] L=α·MSE(y pred ,y true )+β·CrossEntropy(c pred ,c true );

[0083] Where α = 0.7, β = 0.3, y is the spectral intensity, and c is the sealing status classification label.

[0084] Preferably, in step S4, the leakage area A is calculated as follows:

[0085] S401: Acquire point cloud data of the bottle opening via 3D ToF camera 602;

[0086] S402: The random sampling consensus algorithm is used to fit the bottle neck plane;

[0087] S403: Calculate the deviation between the point cloud and the fitted plane. If the local deviation is >0.05mm, it is determined to be a leakage area. The total area A is calculated cumulatively.

[0088] Preferably, in step S5, the coordinate positioning method for laser engraving is as follows:

[0089] S501: Convert the coordinates of the pixel with the highest leakage signal in the hyperspectral image to the machine coordinate system;

[0090] S502: Mapped to the robotic arm's operating space via a hand-eye calibration matrix.

[0091] Example 1

[0092] Optionally, a blueberry processing line is required to perform 100% online sealing inspection on 250mL cylindrical PET bottles with a bottle mouth diameter of 28mm, a heat-sealed aluminum film thickness of 0.1mm, and a production line speed of 300 bottles / minute. Traditional vacuum inspection methods cannot meet the requirements due to their speed being <100 bottles / minute and are insensitive to micron-level leakage. Therefore, the blueberry bottle sealing quality inspection device and method based on spectral analysis of this invention are used for online inspection of standard blueberry bottles. The specific steps are as follows:

[0093] Device configuration: The spectral probe is installed 20cm above the conveyor belt 2; the power of the near-infrared light source 401 is set to 80mW; the spectral camera 405 has a frame rate of 60fps, a resolution of 512×512 pixels, and a spectral resolution of 5nm; the edge computing unit loads the pre-trained model, which is trained based on 5000 sets of standard bottle spectral data and environmental parameters.

[0094] Testing process:

[0095] Step 1: After the bottle is in place, the light field modulator loads the Zernike 36-term compensation pattern, where the fitting coefficients a1 to a36 are obtained through online calibration, and a hyperspectral image is acquired;

[0096] Step 2: Fruit pomace contamination was detected, with a peak frequency of 35 lp / mm. The 1064nm light source was activated for a cleaning window of 30 seconds, and the photocatalytic decomposition efficiency was ≥85%.

[0097] Step 3: The environmental sensor measured the temperature as 25℃±2℃ and the humidity as 60%RH. The Transformer-CNN model outputs compensation coefficients, with the maximum correction being 0.8%.

[0098] Step 4: Calculate the O2 penetration rate and determine if it is qualified;

[0099] Through the above steps, the specific test results are as follows: 0.05mm micropore leakage was detected, with a detection rate of 99.1%; the test time per bottle was 180ms, supporting a production rate of 300 bottles / minute; and the label reflection false detection rate was 0.4%.

[0100] Example 2

[0101] Optionally, when testing a 500mL square HDPE bottle with asymmetrical bottle neck threads and a biodegradable PLA sealing film of 0.15mm thickness, traditional spectroscopic testing requires manual recalibration of the bottle shape, which takes up to 8 hours. Therefore, the blueberry bottle sealing quality testing device and method based on spectral analysis of this invention are used for testing the sealing of irregularly shaped bottle necks. The specific steps are as follows:

[0102] Step 1: Update the bottle neck material database. The O2 diffusion coefficient D of PLA is 1.2 * 10⁻⁶. -13 m 2 / s;

[0103] Step 2: The 3D ToF camera 602 scans the bottle mouth shape with a point cloud resolution of 0.1mm to generate a 3D model of the threaded area;

[0104] Step 3: Input 50 sets of new bottle shape data into the model online, and the compensation accuracy is restored from 71% to 97%;

[0105] Step 4: Fit the bottle opening plane using the RANSAC algorithm, with a leakage area calculation error of ≤5%;

[0106] Through the above steps, the specific test results are as follows: the adaptation time for the new bottle type is 23 minutes; the deformation detection accuracy is 0.08mm.

[0107] Example 3

[0108] Optionally, in a tropical production line environment with humidity fluctuations of 70-95% RH, where condensation on the bottle surface leads to a false detection rate as high as 12.3% with traditional spectral detection, the blueberry bottle sealing quality detection device and method based on spectral analysis of this invention are used to conduct high humidity environment stability testing; the specific steps are as follows:

[0109] Step 1: Activate the pulse airflow ring 801 to remove condensate, with the air pressure at 0.5MPa and the jet frequency at 10Hz;

[0110] Step 2: Reduce water droplet adhesion by using a superhydrophobic coating 502 with a contact angle >150°;

[0111] Step 3: The Transformer encoder captures a step change in humidity, with a gradient of 20% RH / s, and the spectral baseline drift after compensation is ≤0.5%;

[0112] Through the above steps, the specific test results are as follows: the false detection rate is 0.8% at 90% RH; pulsed airflow removes condensate and avoids mechanical contact contamination.

[0113] Example 4

[0114] Optionally, existing blueberry jam filling lines have a 15% residual pectin and pomace contamination rate at the bottle mouth. Traditional testing requires frequent machine shutdowns for manual cleaning. Therefore, the blueberry bottle seal quality testing device and method based on spectral analysis of this invention are used for standard pomace contamination self-cleaning testing. The specific steps are as follows:

[0115] Step 1: Identify pectin contamination, where the spectral reflectance is 92%. Initiate photocatalytic cleaning with infrared light source parameters of 1064nm, 60mW, and 40 seconds.

[0116] Step 2: The graphene coating catalyzes the decomposition of pectin, with a decomposition rate of 87%.

[0117] Step 3: After cleaning, the window's light transmittance was measured to have recovered to 98.6%.

[0118] Through the above steps, the specific test results are as follows: the pollution false alarm rate is 0.2%; no manual intervention is required for cleaning; and photocatalytic decomposition does not require the consumption of chemical reagents.

[0119] The embodiments of the present invention have been described in detail above with reference to the accompanying drawings. However, the present invention is not limited to the above embodiments. Within the scope of knowledge possessed by those skilled in the art, various changes can be made without departing from the spirit of the present invention.

Claims

1. A method for detecting the sealing quality of blueberry bottles based on spectral analysis, characterized in that: The blueberry bottle seal quality inspection device based on spectral analysis includes a support frame and a conveyor belt, as well as limiting grooves, blueberry bottle bodies, probe assemblies, window assemblies, environmental sensing components, calibration components, cleaning components, and edge computing components. The conveyor belt, a transverse conveying structure, is positioned above the support frame. Multiple sets of limiting grooves are evenly distributed on the upper surface of the conveyor belt, with the blueberry bottle body positioned inside each limiting groove. The probe assembly is positioned above the support frame, with a window assembly at its bottom, an environmental sensing component on its side, a calibration component below it, and a cleaning component below that as well. An edge computing component is positioned on one side of the support frame. The probe assembly includes a support arm, a probe housing, a near-infrared light source, a programmable light field modulator, an LCD panel, a beam splitter, and a spectral camera. The support arm is vertically fixed above the inspection station on the support frame, and the probe housing is located below it. A near-infrared light source is housed inside the probe housing. The inner top surface of the housing houses the near-infrared light source emits wavelengths of 900-2500nm. The power of the near-infrared light source is controlled by a pulse modulation circuit. A programmable light field modulator is installed inside the probe housing, located in the middle section of the housing and directly below the near-infrared light source. An LCD panel is installed inside the modulator, perpendicular to the optical axis of the near-infrared light source. A beam splitter is installed inside the housing, fixed to the bottom of the housing with optical adhesive. The beam splitter is coaxially aligned with the optical path exit of the modulator. A spectroscopic camera is installed on the inner side wall of the housing, connected to the beam splitter via an optical fiber bundle. The reflective surface of the beam splitter and the lens of the spectroscopic camera form a 45° angle. The window assembly includes an optical window and a hydrophobic coating. An optical window is located at the bottom of the housing, covering the bottom opening. The surface of the optical window is coated with a hydrophobic coating, which is a graphene superhydrophobic coating with a thickness of 200±10nm. The environmental sensing component includes a temperature and humidity sensor and a camera. The temperature and humidity sensor is located on one side of the probe housing, with the detection end of the temperature and humidity sensor facing the bottle opening area of ​​the blueberry bottle. The camera is located on the other side of the probe housing. The camera is a 3D ToF camera, and the field of view of the camera covers the bottle opening area of ​​the blueberry bottle. The method for testing the sealing quality of blueberry bottles includes the following steps: S1: Start the conveyor belt to move the blueberry bottle to the detection station, start the near-infrared light source to emit pulsed light, load the phase compensation pattern generated by the Zernike polynomial through the LCD panel of the programmable light field modulator, adjust the incident light field wavefront, and guide the reflected light into the spectroscopic camera through the beam splitter to acquire the hyperspectral image cube of the bottle mouth area at a rate of 10ms / frame. The spatial resolution of the hyperspectral image cube is 512×512 pixels and the spectral resolution is 5nm. S2: Perform Fourier transform analysis on the diffraction patterns of the hyperspectral image to extract spatial frequency features. If the frequency peak is in the range of 10-50 lp / mm, it is determined to be fruit pomace contamination. If the frequency peak is <10 lp / mm and the spectral reflectance is >85%, it is determined to be condensate contamination. When organic pollutants are identified, control the near-infrared light source to switch to a wavelength of 1064nm and continuously irradiate the optical window with a power of 50mW for 30 seconds to excite the photocatalytic decomposition reaction of the graphene superhydrophobic coating. S3: Read the time-series data from the temperature and humidity sensor, input it into the pre-trained Transformer-CNN hybrid network model, and extract the spectral spatial features of the bottle mouth region through 3 convolutional layers, where the convolutional kernel is 3×3 and the stride is 1; capture the correlation between temperature and humidity fluctuations and spectral drift through the Transformer encoder, where the Transformer encoder has 8 attention heads and 256 hidden layer dimensions; then output a pixel-wise compensation coefficient matrix with a resolution of 256×256, and multiply the compensation coefficient matrix with the original spectral data pixel by pixel to correct the baseline drift; S4: Perform partial least squares discriminant analysis on the compensated spectral data to extract the absorption peak intensity of O2 at 760nm, and calculate the O2 permeability based on Fick's law: Where D is the O2 diffusion coefficient, A is the leakage area, and P in and P out The values ​​are the internal and external air pressures of the bottle, L is the thickness of the sealing film, the diffusion coefficient is obtained from a pre-stored bottle mouth material database, and the leakage area is calculated by scanning the bottle mouth morphology with a camera. If the calculated permeability value is greater than the preset threshold, it is determined to be a sealing failure. S5: Generates a sealing quality report and sends it to the production line PLC via the Modbus protocol. At the same time, it links with an external robotic arm to grab defective bottles and laser-engraves the leakage coordinates on the bottle label.

2. The method for detecting the sealing quality of blueberry bottles based on spectral analysis according to claim 1, characterized in that: In step S1, the method for generating the Zernike polynomial phase compensation pattern is as follows: S101: Calculate the wavefront distortion function W(x, y) caused by pollutant scattering; S102: Expanded into a 36-term Zernike polynomial according to the following formula: Among them, Z k (x,y) are Zernike basis functions, a k These are the fitting coefficients; S103: Load the coefficient matrix into the programmable optical field modulator.

3. The method for detecting the sealing quality of blueberry bottles based on spectral analysis according to claim 1, characterized in that: In step S4, the leakage area A is calculated as follows: S401: Acquires point cloud data of the bottle opening via a 3D ToF camera; S402: The random sampling consensus algorithm is used to fit the bottle neck plane; S403: Calculate the deviation between the point cloud and the fitted plane. If the local deviation is >0.05mm, it is determined to be a leakage area. The total area A is calculated cumulatively.

4. The method for detecting the sealing quality of blueberry bottles based on spectral analysis according to claim 1, characterized in that: In step S5, the coordinate positioning method for laser engraving is as follows: S501: Convert the coordinates of the pixel with the highest leakage signal in the hyperspectral image to the machine coordinate system; S502: Mapped to the robotic arm's operating space via a hand-eye calibration matrix.

5. The method for detecting the sealing quality of blueberry bottles based on spectral analysis according to claim 1, characterized in that: The calibration assembly includes a calibration plate, a quantum dot film, and a lifting cylinder. The calibration plate is located directly below the optical window. The surface of the calibration plate is coated with a quantum dot film, specifically a CdSe / ZnS quantum dot film. The lifting cylinder is located below the calibration plate and is situated above the support. The lifting stroke of the calibration plate is 0-50mm.

6. The method for detecting the sealing quality of blueberry bottles based on spectral analysis according to claim 1, characterized in that: The cleaning assembly includes an airflow ring, jet nozzles, an air pipe, and an air pump. An airflow ring is located below the optical window. The airflow ring is a circular structure that surrounds the outer perimeter of the optical window. Jet nozzles are located on the inner side of the airflow ring. Multiple sets of jet nozzles are evenly distributed circumferentially on the inner side of the airflow ring. The axis of the jet nozzles forms a 30° angle with the plane of the optical window. An air pipe is located on one side of the airflow ring, and an air pump is located at one end of the air pipe. The air pump is located on one side of the support.

7. The method for detecting the sealing quality of blueberry bottles based on spectral analysis according to claim 1, characterized in that: The edge computing component includes a heatsink, a motherboard, and a fan. The heatsink is located on one side of the bracket, and the motherboard is located inside the heatsink. The fan is located on the inner side of the heatsink. The GPIO interface of the motherboard is connected to the spectral camera, temperature and humidity sensor, and camera respectively through shielded cables.

Images