Medical glass bottle cracking detection system

Abstract

The invention discloses a medical glass bottle cracking detection system, and relates to the technical field of defect detection, the medical glass bottle cracking detection system comprises a data analysis module, a depth measurement module and a cracking detection module, the data analysis module is electrically connected with the depth measurement module and the cracking detection module, and the data analysis module is electrically connected with the depth measurement module. The data analysis module is used for storing, counting, analyzing and calculating data of crack depth thickness measurement, the depth measurement module is used for detecting the crack depth by using an attenuation phenomenon after light enters a glass material, and the crack detection module is used for performing crack detection on a cap. The opening cracking detection module comprises a translation mechanism, an industrial camera, a detection platform, an opening surface cutting module, a training module and a comparison reasoning module, the translation mechanism is mechanically connected with the detection platform, the translation mechanism is used for driving the whole tray containing the bottle to move, and the industrial camera is used for shooting the bottle opening. The method has the characteristic of high efficiency.

Description

Technical Field

[0001] The invention relates to the technical field of defect detection, and in particular to a burst detection system for medical glass bottles. Background Art

[0002] The bursting defect refers to a defect in the cap that connects the upper part of the glassware product to the blowpipe. If a bottle with a bursting defect enters a pharmaceutical factory and stores drugs for use in a hospital, the medical hose cannot be accurately connected, resulting in leakage and detachment.

[0003] Workers in the workshop manually inspect thousands of glass bottles every day, and they observe them with their naked eyes. After a long time, they will inevitably suffer from visual fatigue and miss inspections. The bottles produced by each production line are varied, and the locations of the burst holes are also ever-changing, with different shapes and sizes. Using traditional visual inspection methods will cause a lot of misjudgments and miss inspections, making it impossible to conduct inspections.

[0004] When testing the depth of the burst defect, light will be directed from the outer edge of the cap to the center of the bottle mouth. Because the thickness of the cap at the burst becomes smaller, the attenuation of the light directed to the burst will become smaller. The burst depth is calculated by measuring the attenuation of light. However, since the detection follows the principle of full coverage of the burst, the light may irradiate the scanning area of ​​the cap circumference outside the burst when measuring multiple times. Here, the attenuation of the measurement may be too large because the light does not pass through the burst, and the burst depth cannot be accurately reflected. Therefore, it is necessary to design a highly efficient burst detection system for pharmaceutical glass bottles. Summary of the invention

[0005] The object of the present invention is to provide a medicine glass bottle burst detection system to solve the problems raised in the above background technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: a burst detection system for medical glass bottles, comprising a data analysis module, a depth measurement module and a burst detection module, wherein the data analysis module is electrically connected to the depth measurement module and the burst detection module, the data analysis module is used to store and count the data of burst depth thickness measurement, and perform analysis and calculation, the depth measurement module is used to detect the burst depth by utilizing the attenuation phenomenon of light after entering the glass material, and the burst detection module is used to perform burst detection on the cap.

[0007] According to the above technical solution, the burst detection module includes a translation mechanism, an industrial camera, a detection platform, a mouth surface cropping module, a training module, and a comparison and reasoning module. The translation mechanism is mechanically connected to the detection platform, and the translation mechanism is used to drive the tray holding the bottles to move as a whole. The industrial camera is used to photograph the bottle mouth, and the detection platform is used to place the bottles to be detected. The mouth surface cropping module is electrically connected to the comparison and reasoning module. The mouth surface cropping module is used to crop the mouth surface of each bottle in the batch of burst photos. The training module is used to manually mark the burst defects and use deep learning to perform burst recognition training. The comparison and reasoning module is used to use a convolutional neural network model to compare and reason the circular bottle mouth surfaces one by one, and mark the bottles with burst defects.

[0008] The depth measurement module includes a light emitting unit, a light sensing unit, an area coordinate division module, a rotation mechanism, and a width calculation module. The area coordinate division module is electrically connected to the width calculation module, and the rotation mechanism is electrically connected to the light emitting unit. The light emitting unit is used to provide an emission source of a light beam to the cap, and the light sensing unit is used to receive light attenuated by a glass material. The rotation mechanism is used to control the incident position of the light to the cap. The area coordinate division module is used to divide the scanning area and the burst area on the cap according to the design parameters, and the width calculation module is used to calculate the width ratio of the scanning area and the burst area according to the divided areas.

[0009] The data analysis module includes an attenuation analysis module, a light intensity control module, and a data storage module. The attenuation analysis module is electrically connected to the light sensing unit, the data storage module is electrically connected to the attenuation analysis module and the light sensing unit, and the light intensity control module is electrically connected to the light emitting unit. The attenuation analysis module is used to analyze the burst depth according to the attenuation degree of the light, the light intensity control module is used to control the direction and intensity of the incident light, and the data storage module is used to store the historical thickness data of the cap and the attenuation degree data of the light.

[0010] According to the above technical solution, the working method of the system is:

[0011] S1. Place multiple bottles with cracked mouth defects collected from the workshop on a tray, take photos of the caps with an industrial camera to obtain batches of cracked mouth photos, then crop the mouth surface of each bottle and save them to a computer. After marking the cracked mouth defects, enter the training module for training to generate a training result model.

[0012] S2. The testing platform moves with the tray as a whole, takes pictures of all the bottles to be tested with a camera to obtain batches of burst mouth pictures, loads the pictures taken by the camera into the program, and cuts out all the circular bottle mouth surfaces to be tested;

[0013] S3, build a convolutional neural network model, call the training result model generated in the first step to compare and infer the circular bottle mouth surfaces one by one, find out the bottles with burst defects in the bottles to be tested, and record the test result information into the memory;

[0014] S4. Performing a light intensity test on the bottle with a cracked hole defect, so that the light is incident on the central light sensing unit from a direction perpendicular to the arc on the surface of the cap, and the arc on the surface of the cap is evenly expanded into a flat area, and the scanning area and the cracked hole area are divided according to the cracked hole photo;

[0015] S5, utilizing the attenuation phenomenon of light after entering the glass material, and analyzing the attenuation degree to obtain the uncalibrated burst depth;

[0016] S6. Correct the obtained burst hole depth according to the proportion of the burst hole area in the scanning area to obtain an accurate burst hole depth.

[0017] According to the above technical solution, in S4, the division of the scanning area and the burst area is as follows: according to the design parameters of the cap, the surface arc of the cap is evenly expanded into a plane area, the outer contour circumference of the cap is L, and the angle of the scanning area emitted by the light at the cap is b, then the width of the scanning area is The angle of the burst area is a, and the width of the burst area is

[0018] According to the above technical solution, in S5, the uncalibrated burst depth measurement method is: since the glass material will affect the attenuation of the light emitted to the surface of the cap, the thicker the glass material, the greater the attenuation, so the light intensity emitted to the burst area will become smaller, when the initial intensity of a beam of light emitted to the cap is α, the measured light intensity is β, the attenuation degree when the cap has no burst is a constant value β0, assuming the burst depth is k, and the thickness of the cap glass material is known to be K, then the light intensity attenuation degree is so The uncalibrated burst depth k is calculated based on this formula.

[0019] According to the above technical solution, in S6, the correction according to the proportion of the burst area in the scanning area specifically includes: since the light is incident randomly multiple times, the probability of incident on the burst area is the ratio of the width of the burst area to the scanning area, and the probability of incident on the burst area is The probability of incident on the scanning area outside the blast hole is The incident light outside the blast hole will change the measured light intensity from β to When the width of the burst area accounts for a larger proportion, light is more likely to be incident on the burst area, resulting in a smaller measurement deviation of the light intensity β. When the width of the burst area accounts for a smaller proportion, light is more likely to be incident on other positions in the scanning area, resulting in a larger measurement deviation of the light intensity β.

[0020] According to the above technical solution, in S6, after multiple measurements, a set of measured light intensities β is obtained, and all measured β are averaged to obtain an average value. Where n is the number of measurements, the correction degree of light intensity β

[0021] According to the above technical solution, the method for comparing with the set burst depth is: substitute the calculated correction degree γ into the calculation formula of the burst depth k to obtain the correct burst depth. The corrected formula is:

[0022] The burst depth k value is calculated according to the formula.

[0023] Compared with the prior art, the beneficial effects achieved by the present invention are as follows: the present invention adopts the batch detection technology of bottle mouth surface bursting, and the worker only needs to place a tray of bottles on the detection table and press the start button, the detection platform starts to move and the camera takes pictures and calculates, which greatly shortens the time of manually inspecting bottle mouth defects one by one with the naked eye, and effectively solves the problems of visual fatigue and low efficiency of manual detection;

[0024] By building a convolutional neural network model, we conduct machine learning on various types of bottle burst defects, identify common defect features and independently infer them, effectively solving the problem of a large number of misjudgments and missed judgments of bottle edge bursts detected by traditional algorithms. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0026] Figure 1 It is a schematic diagram of the overall module structure of the present invention;

[0027] Figure 2 It is a schematic diagram of the principle of the present invention. DETAILED DESCRIPTION

[0028] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0029] See also Figure 1 and Figure 2 The present invention provides a technical solution: a burst detection system for a medical glass bottle, comprising a data analysis module, a depth measurement module and a burst detection module, wherein the data analysis module is electrically connected to the depth measurement module and the burst detection module, the data analysis module is used to store and count the data of the burst depth thickness measurement, and perform analysis and calculation, the depth measurement module is used to detect the burst depth by utilizing the attenuation phenomenon of light after entering the glass material, and the burst detection module is used to perform burst detection on the cap;

[0030] The burst detection module includes a translation mechanism, an industrial camera, a detection platform, a mouth surface cropping module, a training module, and a comparison and reasoning module. The translation mechanism is mechanically connected to the detection platform. The translation mechanism is used to drive the tray holding the bottles to move as a whole. The industrial camera is used to photograph the bottle mouth. The detection platform is used to place the bottles to be detected. The mouth surface cropping module is electrically connected to the comparison and reasoning module. The mouth surface cropping module is used to crop the mouth surface of each bottle in the batch of burst photos. The training module is used to manually mark the burst defects and use deep learning to perform burst recognition training. The comparison and reasoning module is used to use a convolutional neural network model to compare and reason the circular bottle mouth surfaces one by one, and mark the bottles with burst defects.

[0031] The depth measurement module includes a light emitting unit, a light sensing unit, an area coordinate division module, a rotating mechanism, and a width calculation module. The area coordinate division module is electrically connected to the width calculation module, and the rotating mechanism is electrically connected to the light emitting unit. The light emitting unit is used to provide an emission source of a light beam to the cap, the light sensing unit is used to receive light attenuated by the glass material, the rotating mechanism is used to control the incident position of the light to the cap, the area coordinate division module is used to divide the scanning area and the burst area on the cap according to the design parameters, and the width calculation module is used to calculate the width ratio of the scanning area and the burst area according to the divided areas;

[0032] The data analysis module includes an attenuation analysis module, a light intensity control module, and a data storage module. The attenuation analysis module is electrically connected to the light sensing unit, the data storage module is electrically connected to the attenuation analysis module and the light sensing unit, and the light intensity control module is electrically connected to the light emitting unit. The attenuation analysis module is used to analyze the depth of the burst hole according to the attenuation degree of the light, the light intensity control module is used to control the direction and intensity of the incident light, and the data storage module is used to store the historical thickness data of the cap and the attenuation degree data of the light.

[0033] The system works as follows:

[0034] S1. Place multiple bottles with cracked mouth defects collected from the workshop on a tray, take photos of the caps with an industrial camera to obtain batches of cracked mouth photos, then crop the mouth surface of each bottle and save them to a computer. After marking the cracked mouth defects, enter the training module for training to generate a training result model.

[0035] S2. The testing platform moves with the tray as a whole, takes pictures of all the bottles to be tested with a camera to obtain batches of burst mouth pictures, loads the pictures taken by the camera into the program, and cuts out all the circular bottle mouth surfaces to be tested;

[0036] S3, build a convolutional neural network model, call the training result model generated in the first step to compare and infer the circular bottle mouth surfaces one by one, find out the bottles with burst defects in the bottles to be tested, and record the test result information into the memory;

[0037] S4. Performing a light intensity test on the bottle with a cracked hole defect, so that the light is incident on the central light sensing unit from a direction perpendicular to the arc on the surface of the cap, and the arc on the surface of the cap is evenly expanded into a flat area, and the scanning area and the cracked hole area are divided according to the cracked hole photo;

[0038] S5, utilizing the attenuation phenomenon of light after entering the glass material, and analyzing the attenuation degree to obtain the uncalibrated burst depth;

[0039] S6, correcting the obtained burst hole depth according to the proportion of the burst hole area in the scanning area to obtain an accurate burst hole depth;

[0040] In S4, the scanning area and the burst area are divided as follows: according to the design parameters of the cap, the surface arc of the cap is evenly expanded into a flat area, the outer contour circumference of the cap is L, and the angle of the scanning area emitted by the light at the cap is b, then the width of the scanning area is The angle of the burst area is a, and the width of the burst area is

[0041] In S5, the uncalibrated burst depth measurement method is as follows: Since the glass material will affect the attenuation of the light emitted to the surface of the cap, the thicker the glass material, the greater the attenuation, so the light intensity emitted to the burst area will become smaller. When the initial intensity of a beam of light emitted to the cap is α, the measured light intensity is β, and the attenuation degree when the cap has no burst is a constant value β0. Assuming the burst depth is k, and the thickness of the cap glass material is known to be K, the light intensity attenuation degree is so According to this formula, the uncalibrated burst depth k is calculated;

[0042] In S6, the correction based on the proportion of the burst area in the scanning area specifically includes: since the light is incident randomly multiple times, the probability of incident on the burst area is the ratio of the width of the burst area to the scanning area, and the probability of incident on the burst area is The probability of incident on the scanning area outside the blast hole is The incident light outside the blast hole will change the measured light intensity from β to When the width of the burst area accounts for a larger proportion, the light is more likely to be incident on the burst area, resulting in a smaller measurement deviation of the light intensity β. When the width of the burst area accounts for a smaller proportion, the light is more likely to be incident on other positions in the scanning area, resulting in a larger measurement deviation of the light intensity β.

[0043] In S6, after multiple measurements, a set of measured light intensities β is obtained, and all measured β are averaged to obtain an average value Where n is the number of measurements, the correction degree of light intensity β

[0044] The method for comparing with the set burst depth is: substitute the calculated correction degree γ into the calculation formula of the burst depth k to obtain the correct burst depth. The corrected formula is:

[0045] The burst depth k value is calculated according to the formula.

[0046] The probability of light directly entering the non-burst area of ​​the cap is estimated by the ratio of the burst width to the detection area, and the situation of light entering the non-burst area is calculated. Since the measured light intensity β will be smaller due to the small probability of incident on the non-burst area, the measurement result is corrected by the comprehensive probability and deviation value, so that the measured light intensity can be more in line with reality and a more accurate burst depth can be obtained.

[0047] It should be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0048] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art can still modify the technical solutions described in the aforementioned embodiments or replace some of the technical features therein by equivalents. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A system for detecting bursts of medical glass bottles. Features: It includes a data analysis module, a depth measurement module and a burst detection module. The data analysis module is electrically connected to the depth measurement module and the burst detection module. The data analysis module is used to store and count the data of burst depth thickness measurement, and perform analysis and calculation. The depth measurement module is used to detect the burst depth by utilizing the attenuation phenomenon of light after entering the glass material. The burst detection module is used to perform burst detection on the cap.

2. A medical glass bottle burst detection system according to claim 1, Features: The burst mouth detection module includes a translation mechanism, an industrial camera, a detection platform, a mouth surface cropping module, a training module, and a comparison and reasoning module. The translation mechanism is mechanically connected to the detection platform, the translation mechanism is used to drive the tray holding the bottles to move as a whole, the industrial camera is used to photograph the bottle mouth, the detection platform is used to place the bottles to be detected, the mouth surface cropping module is electrically connected to the comparison and reasoning module, the mouth surface cropping module is used to crop the mouth surface part of each bottle in the batch burst mouth photos, the training module is used to manually mark the burst mouth defects and use deep learning to perform burst mouth recognition training, and the comparison and reasoning module is used to use a convolutional neural network model to compare and reason the circular bottle mouth surfaces one by one, and mark the bottles with burst mouth defects; The depth measurement module includes a light emitting unit, a light sensing unit, an area coordinate division module, a rotation mechanism, and a width calculation module. The area coordinate division module is electrically connected to the width calculation module, and the rotation mechanism is electrically connected to the light emitting unit. The light emitting unit is used to provide an emission source of a light beam to the cap, and the light sensing unit is used to receive light attenuated by a glass material. The rotation mechanism is used to control the incident position of the light to the cap. The area coordinate division module is used to divide the scanning area and the burst area on the cap according to the design parameters, and the width calculation module is used to calculate the width ratio of the scanning area and the burst area according to the divided areas. The data analysis module includes an attenuation analysis module, a light intensity control module, and a data storage module. The attenuation analysis module is electrically connected to the light sensing unit, the data storage module is electrically connected to the attenuation analysis module and the light sensing unit, and the light intensity control module is electrically connected to the light emitting unit. The attenuation analysis module is used to analyze the burst depth according to the attenuation degree of the light, the light intensity control module is used to control the direction and intensity of the incident light, and the data storage module is used to store the historical thickness data of the cap and the attenuation degree data of the light.

3. A medical glass bottle burst detection system according to claim 2, Features: The system works as follows: S1. Place multiple bottles with cracked mouth defects collected from the workshop on a tray, take photos of the caps with an industrial camera to obtain batches of cracked mouth photos, then crop the mouth surface of each bottle and save them to a computer. After marking the cracked mouth defects, enter the training module for training to generate a training result model. S2. The testing platform moves with the tray as a whole, takes pictures of all the bottles to be tested with a camera to obtain batches of burst mouth pictures, loads the pictures taken by the camera into the program, and cuts out all the circular bottle mouth surfaces to be tested; S3, build a convolutional neural network model, call the training result model generated in the first step to compare and infer the circular bottle mouth surfaces one by one, find out the bottles with burst defects in the bottles to be tested, and record the test result information into the memory; S4. Performing a light intensity test on the bottle with a cracked hole defect, so that the light is incident on the central light sensing unit from a direction perpendicular to the arc on the surface of the cap, and the arc on the surface of the cap is evenly expanded into a flat area, and the scanning area and the cracked hole area are divided according to the cracked hole photo; S5, utilizing the attenuation phenomenon of light after entering the glass material, and analyzing the attenuation degree to obtain the uncalibrated burst depth; S6. Correct the obtained burst hole depth according to the proportion of the burst hole area in the scanning area to obtain an accurate burst hole depth.

4. A medical glass bottle burst detection system according to claim 3, Features: In S4, the scanning area and the burst area are divided as follows: according to the design parameters of the cap, the surface arc of the cap is evenly expanded into a plane area, the outer contour circumference of the cap is L, and the angle of the scanning area emitted by the light at the cap is b, then the width of the scanning area is The angle of the burst area is a, and the width of the burst area is 5. A medical glass bottle burst detection system according to claim 4, Features: In S5, the uncalibrated burst depth measurement method is as follows: since the glass material will affect the attenuation of the light emitted to the surface of the cap, the thicker the glass material, the greater the attenuation, so the intensity of the light emitted to the burst area will become smaller, when the initial intensity of a beam of light emitted to the cap is α, the measured light intensity is β, and the attenuation when the cap has no burst is a constant value β 0 , assuming the depth of the burst hole is k, and the thickness of the cap glass material is known to be K, then the degree of light intensity attenuation is so The uncalibrated burst depth k is calculated according to this formula.

6. A medical glass bottle burst detection system according to claim 5, Features: In S6, the correction according to the proportion of the burst area in the scanning area specifically includes: since the light is incident randomly multiple times, the probability of incident on the burst area is the ratio of the width of the burst area to the scanning area, and the probability of incident on the burst area is The probability of incident on the scanning area outside the blast hole is The incident light outside the blast hole will change the measured light intensity from β to When the width of the burst area accounts for a larger proportion, light is more likely to be incident on the burst area, resulting in a smaller measurement deviation of the light intensity β. When the width of the burst area accounts for a smaller proportion, light is more likely to be incident on other positions in the scanning area, resulting in a larger measurement deviation of the light intensity β.

7. A medical glass bottle burst detection system according to claim 6, Features: In S6, after multiple measurements, a set of measured light intensities β is obtained, and all measured β are averaged to obtain an average value Where n is the number of measurements, the correction degree of light intensity β 8. A medical glass bottle burst detection system according to claim 7, Features: The method for comparing with the set burst depth is: substitute the calculated correction degree γ into the calculation formula of the burst depth k to obtain the correct burst depth. The corrected formula is: The burst depth k value is calculated according to the formula.

Images