Soft bag package sealing integrity leakage detection instrument and detection method

By using neural networks and large language models in soft bag packaging seal integrity leak detection instruments, combining pressure sensors and vacuum pumps, the problem that the existing technology cannot quantitatively evaluate the airtightness of soft bag packaging is solved, and the seal integrity of soft bag packaging is achieved Accurate quantitative evaluation and improvement of testing efficiency.

CN119043600APending Publication Date: 2024-11-29BOMA SCI INSTR (SHANGHAI) CO LTD

Patent Information

Application Number
CN202411483642.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-23
Publication Date
2024-11-29

AI Technical Summary

Technical Problem

Existing flexible packaging sealing performance testing standards cannot quantitatively give the airtightness of the flexible packaging and cannot accurately evaluate its sealing integrity.

Method used

A soft bag packaging seal integrity leak detection instrument was designed, using a system including a sealing cavity, pressure sensor and processor. It uses a vacuum pump to pump air and detect pressure changes through a neural network to calculate the air tightness value of the soft bag packaging. Combining the large language model of the frozen part of the layer fits the relationship between the vacuum degree of the sealing cavity and the vacuum pump, providing quantitative airtightness evaluation.

Benefits of technology

A quantitative evaluation of the integrity of the soft bag packaging seal is achieved, providing specific airtightness values, saving costs, and improving the accuracy and efficiency of testing.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119043600A_ABST
    Figure CN119043600A_ABST
Patent Text Reader

Abstract

The invention discloses a soft bag packaging sealing integrity leakage detection instrument and a detection method, and belongs to the technical field of artificial intelligence. In the instrument, a first air tightness detection module is formed by training a first neural network, the first neural network comprises a first input layer and a first two-dimensional neuron middle layer, and during training, pressure electricity information of standard and qualified soft bag packages under a set value vacuum degree and measured air tightness are input into the first input layer; a first two-dimensional neuron intermediate layer is learned, during measurement, pressure information of a soft bag package to be detected under a set value vacuum degree is input to a first input layer of a first air tightness detection module, and the pressure electric information of the soft bag package to be detected and the pressure information of a standard qualified soft bag package in the first two-dimensional neuron intermediate layer are clustered, so that the pressure electric information of the soft bag package to be detected and the pressure information of the standard qualified soft bag package in the first two-dimensional neuron intermediate layer are clustered; the air tightness of the standard qualified soft bag package with the minimum Euler distance is corrected by the correction factor to be the first air tightness of the soft bag package to be detected. According to the invention, the quantitative air tightness value of soft bag packaging can be given.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to a soft bag packaging sealing integrity leakage detection instrument and a detection method, belonging to the technical field of artificial intelligence. Background Art

[0002] GB / T 15171 "Test Method for Sealing Performance of Flexible Packages" is a national standard for testing the sealing performance of flexible packages such as food and medicine. This standard specifies the test method, test equipment, test conditions, test steps and result judgment of the sealing performance of flexible packages, and provides a unified standard and basis for testing the sealing performance of food flexible packages. The test process is: place the sample in the vacuum chamber of the flexible package sealing tester to ensure that the sample and the vacuum chamber are well sealed. Start the tester and apply pressure or vacuum to the sample according to the prescribed procedure. Observe and record the changes in the sample, such as leakage, rupture, etc. Based on the test results, determine whether the sealing performance of the sample meets the requirements.

[0003] However, the national standard for this test cannot quantitatively give the airtightness of flexible packaging. Summary of the invention

[0004] The object of the present invention is to provide a soft bag packaging sealing integrity leakage detection instrument, which can provide a quantitative airtightness value of the soft bag packaging.

[0005] To achieve the above-mentioned purpose, the present invention provides a soft bag package sealing integrity leakage detection instrument, which includes a sealing chamber, a pressure sensor and a first processor. The soft bag package to be detected is placed in the sealing chamber, and the sealing chamber is evacuated by a vacuum pump. When the vacuum degree in the sealing chamber is lower than a set value, the evacuation is stopped; the pressure sensor detects the pressure generated by the expansion of the soft bag package to be detected, and converts the pressure value into pressure electrical information; the first processor includes: a first airtightness detection module, the first airtightness detection module outputs the first airtightness of the soft bag package to be detected according to the pressure electrical information, and the first airtightness detection module is trained by a first neural network, and the first neural network includes a first input layer and a first two-dimensional neural network The first two-dimensional neural network intermediate layer is used for the training of the first two-dimensional neural network intermediate layer. During training, the pressure electrical information of the standard qualified soft bag packaging under the set vacuum degree and the measured airtightness are input into the first input layer. The pressure electrical information of the standard qualified soft bag packaging and the measured airtightness are learned into the first two-dimensional neural network intermediate layer. During measurement, the pressure information of the soft bag packaging to be tested under the set vacuum degree is input into the first input layer of the first airtightness detection module. The pressure electrical information of the soft bag packaging to be tested is clustered with the pressure information of the standard qualified soft bag packaging in the first two-dimensional neural network intermediate layer. The airtightness of the standard qualified soft bag packaging in the first two-dimensional neural network intermediate layer with the smallest Euler distance is corrected by the correction factor to be the first airtightness of the soft bag packaging to be tested.

[0006] To achieve the above-mentioned purpose, the present invention also provides a method for detecting the sealing integrity of a soft bag package, which comprises the following steps: Place the soft bag package to be tested in the sealed cavity, and use a vacuum pump to evacuate the sealed cavity. When the vacuum degree in the sealed cavity is lower than the set value, stop evacuating; The pressure sensor is used to detect the pressure generated by the expansion of the soft bag package to be detected in the sealing chamber under the set vacuum degree, and the pressure value is converted into pressure electrical information; The first airtightness of the soft bag package to be tested is outputted according to the pressure electrical information through the first airtightness detection module. The first airtightness detection module is trained by a first neural network. The first neural network includes a first input layer and a first two-dimensional neuron middle layer. During training, the pressure electrical information of the standard qualified soft bag package under the vacuum degree of the set value and the measured airtightness are inputted into the first input layer, and the pressure electrical information of the standard qualified soft bag package and the measured airtightness are learned into the first two-dimensional neuron middle layer. During measurement, the pressure information of the soft bag package to be tested under the vacuum degree of the set value is inputted into the first input layer of the first airtightness detection module, and the pressure electrical information of the soft bag package to be tested is clustered with the pressure information of the standard qualified soft bag package in the first two-dimensional neuron middle layer. The airtightness of the standard qualified soft bag package with the smallest Euler distance is corrected by the correction factor as the first airtightness of the soft bag package to be tested.

[0007] Compared with the prior art, the soft bag packaging seal integrity leakage detection instrument and detection method provided by the present invention can provide a quantitative airtightness value of the soft bag packaging seal. BRIEF DESCRIPTION OF THE DRAWINGS

[0008] Figure 1 It is a block diagram of the composition of a soft bag packaging sealing integrity leakage detection instrument provided by the first embodiment of the present invention.

[0009] Figure 2 It is a block diagram of the composition of a soft bag packaging sealing integrity leakage detection instrument provided by the second embodiment of the present invention.

[0010] Figure 3 It is a block diagram of the composition of a soft bag packaging sealing integrity leakage detection instrument provided by the third embodiment of the present invention.

[0011] Figure 4 It is a block diagram of the composition of a soft bag packaging seal integrity leakage detection instrument provided by the fourth embodiment of the present invention.

[0012] Figure 5 It is a block diagram of the composition of a vacuum degree detector provided in the fourth embodiment of the present invention. DETAILED DESCRIPTION

[0013] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0014] In the present invention, the terms "first" and "second" are used for descriptive purposes only and should not be understood as indicating or implying relative importance.

[0015] First embodiment

[0016] Figure 1 FIG. 1 is a block diagram of the composition of the soft bag packaging seal integrity leakage detection instrument provided by the first embodiment of the present invention. Figure 1 As shown, the soft bag package sealing integrity leakage detection instrument provided by the first embodiment of the present invention includes a sealing chamber, a pressure sensor and a first processor. The soft bag package to be detected is placed in the sealing chamber, and the sealing chamber is evacuated by a vacuum pump. When the vacuum degree in the sealing chamber is lower than the set value, the evacuation is stopped; the pressure sensor detects the pressure generated by the expansion of the soft bag package to be detected, and converts the pressure value into pressure electrical information; the first processor includes: a first airtightness detection module, the first airtightness detection module outputs the first airtightness of the soft bag package to be detected according to the pressure electrical information, and the first airtightness detection module is trained by a first neural network, and the first neural network includes a first input layer and a first two-dimensional The neuron middle layer, during training, the pressure electrical information of the standard qualified soft bag package under the vacuum degree of the set value and the airtightness obtained by measuring it are input into the first input layer, and the pressure electrical information of the standard qualified soft bag package and the airtightness obtained by measuring it are learned into the first two-dimensional neuron middle layer. During measurement, the pressure information of the soft bag package to be tested under the vacuum degree of the set value is input into the first input layer of the first airtightness detection module, and the pressure electrical information of the soft bag package to be tested is clustered with the pressure information of the standard qualified soft bag package in the first two-dimensional neuron middle layer. The airtightness of the standard qualified soft bag package with the smallest Euler distance is corrected by the correction factor to be the first airtightness of the soft bag package to be tested.

[0017] The soft bag packaging sealing integrity leakage detection instrument provided in the first embodiment also includes a flowmeter and a vacuum pump. The vacuum pump is configured to evacuate the sealed cavity under the control of the first processor. A flowmeter is provided between the sealed cavity and the vacuum pump. The flowmeter is used to measure the flow rate Q of the vacuum pump evacuation; the working time T of the vacuum pump is provided by the first processor. The soft bag packaging sealing integrity leakage detection instrument provided in the first embodiment also includes a first input unit, which is used to input instructions and data, and the data includes the volume V of the sealed cavity. The first processor also includes a data acquisition module and a vacuum degree inference model. The data acquisition module is configured to acquire the volume V of the sealed cavity, the flow rate Q of the vacuum pump, and the working time T of the vacuum pump, and input the volume V of the sealed cavity, the flow rate Q of the vacuum pump, and the working time T of the vacuum pump into the vacuum degree inference model. The vacuum degree inference model outputs the vacuum degree of the sealed cavity. The vacuum degree inference model is trained by a large language model with some layers frozen. The large language model with some layers frozen includes an input embedding layer, a position embedding layer, a multi-head attention mechanism layer, a first residual connection layer, a feedforward neural network, and a second residual connection layer, a normalization layer, and a linear output layer.

[0018] In the first embodiment, the method for training a large language model with some layers frozen includes: Step 1: Take N data sets from the database, each of which includes a pre-trained data vector And the vacuum degree of the sealed chamber tested , the pre-training data vector Input to the input embedding layer of the large language model with some layers frozen, and the estimated vacuum degree of the sealed cavity is input from the second residual connection layer : , in, , To freeze the current parameter vector of the large language model with some layers, is the flow rate of the vacuum pump in the nth data set, is the working time of the vacuum pump in the nth data group; Step 2: Let the loss function L be: , The gradient of the neural network parameters is calculated according to the following formula: ; Step 3: Update the current parameter vector of the large language model with some layers frozen according to the following formula: ; In the formula, is the learning rate, represents the gradient of ; Step 4: Determine whether the loss function L is the minimum. If so, end the training. If not, replace the current parameter vector with the new parameter vector and return to step 1.

[0019] The soft bag packaging seal integrity leakage detection instrument provided in the first embodiment of the present invention also includes a first display, which is a human-machine interface for displaying input data and instructions, and also for displaying measurement results, wherein the measurement results include the airtightness of the soft bag packaging seal, the vacuum degree of the sealing cavity, etc.

[0020] The soft bag packaging seal integrity leakage detection instrument provided in the first embodiment of the present invention also includes a first memory, which is used to store an application for measuring the seal integrity of the soft bag packaging, and the application includes a first airtightness detection module, a data acquisition module, a vacuum degree inference module, etc.

[0021] The soft bag packaging seal integrity leakage detection instrument provided in the first embodiment of the present invention further includes a first communication unit, which is used to communicate with a host computer and send the air tightness of the soft bag packaging seal, the vacuum degree of the sealing cavity, etc. to the host computer.

[0022] The first embodiment of the present invention provides a soft bag packaging sealing integrity leakage detection instrument. Since the pressure electrical information of a standard qualified soft bag package under a set vacuum degree and the airtightness obtained by measuring it are learned in advance into the first two-dimensional neuron intermediate layer, during measurement, the pressure information of the soft bag package to be tested under the set vacuum degree is input into the first input layer of the first airtightness detection module, and the pressure electrical information of the soft bag package to be tested is clustered with the pressure information of the standard qualified soft bag package in the first two-dimensional neuron intermediate layer. The airtightness of the standard qualified soft bag package with the smallest Euler distance is corrected by the correction factor to be the first airtightness of the soft bag package to be tested, so the specific airtightness value of the soft bag package to be tested can be given.

[0023] The present invention fits the relationship between the vacuum degree of the sealed cavity and the volume of the sealed cavity, the flow rate of the vacuum pump, and the working time of the vacuum pump by freezing a large language model of some layers. Therefore, when measuring, the vacuum degree of the sealed cavity can be inferred as long as the volume of the sealed cavity, the flow rate of the vacuum pump and the working time are known, and there is no need to use other instruments for measurement, which saves costs.

[0024] Second embodiment

[0025] Figure 2 FIG. 2 is a block diagram of a soft bag packaging seal integrity leakage detection instrument provided by the second embodiment of the present invention. Figure 2As shown, the soft bag packaging sealing integrity leakage detection instrument provided by the second embodiment of the present invention includes a sealing cavity, an image sensor and a first processor. The soft bag packaging to be detected is placed in the sealing cavity, and the sealing cavity is evacuated by a vacuum pump. When the vacuum degree in the sealing cavity is lower than the set value, the evacuation is stopped; the image sensor is used to obtain an image of the inside of the sealing cavity including the soft bag packaging to be detected at the vacuum degree of the set value to obtain image information, and the first processor includes: a feature extraction unit, a convolutional neural network, and a second airtightness detection module, wherein the feature extraction unit extracts feature information according to the image information; the convolutional neural network extracts image information of the soft bag packaging to be detected at the empty degree of the set value according to the feature information; the second airtightness detection module outputs the second airtightness of the soft bag packaging to be detected according to the image information of the soft bag packaging to be detected at the empty degree of the set value.

[0026] In the second embodiment of the present invention, the second air tightness detection module is trained by a second neural network, and the second neural network includes a second input layer and a second two-dimensional neural network intermediate layer. During training, the image information of the standard qualified soft bag packaging at the set value of the empty space is input into the second input layer, and a mapping relationship between the image information of the soft bag packaging at the set value of the empty space and the air tightness obtained by measuring the soft bag packaging is established, and the image information of the standard qualified soft bag packaging at the set value of the empty space is learned into the second two-dimensional neural network intermediate layer.

[0027] During measurement, the image information of the soft bag package to be tested at the set value of emptiness is input into the input layer of the second airtightness detection module, and the similarity between the image information of the soft bag package to be tested at the set value of emptiness and the image information of the standard qualified soft bag package at the set value of emptiness in the second two-dimensional neuron middle layer is calculated, and the airtightness of the seal of the standard qualified soft bag package with the greatest similarity is corrected by the correction factor as the second airtightness of the seal of the soft bag package to be tested.

[0028] The soft bag packaging seal integrity leakage detection instrument provided in the second embodiment of the present invention further includes: a flow meter and a vacuum pump, the vacuum pump is configured to evacuate the sealed cavity under the control of the first processor, a flow meter is provided between the sealed cavity and the vacuum pump, the flow meter is used to measure the flow rate Q of the vacuum pump; the working time T of the vacuum pump is provided by the first processor; the leakage detection instrument further includes a first input unit, which is used to input instructions and data, the data including the volume V of the sealed cavity. The first processor further includes a data acquisition module and a vacuum degree inference model, the data acquisition module is configured to acquire the volume V of the sealed cavity, the flow rate Q of the vacuum pump and the working time T of the vacuum pump, and the volume V of the sealed cavity, the flow rate Q of the vacuum pump and the working time T of the vacuum pump are input into the vacuum degree inference model, the vacuum degree inference model outputs the vacuum degree of the sealed cavity, and the vacuum degree inference model is trained by a large language model with some layers frozen.

[0029] In the second embodiment, the method for training the large language model with some layers frozen is the same as the method for training the large language model with some layers frozen in the first embodiment, and will not be repeated here.

[0030] The soft bag packaging seal integrity leakage detection instrument provided in the second embodiment of the present invention also includes a first display, which is a human-machine interface for displaying input data and instructions, and also for displaying measurement results, wherein the measurement results include the airtightness of the soft bag packaging seal, the vacuum degree of the sealing cavity, etc.

[0031] The soft bag packaging seal integrity leakage detection instrument provided in the second embodiment of the present invention also includes a first memory, which is used to store an application for measuring the seal integrity of the soft bag packaging, and the application includes a second airtightness detection module, a data acquisition module, a vacuum degree inference module, etc.

[0032] The soft bag packaging seal integrity leakage detection instrument provided in the second embodiment of the present invention further includes a first communication unit, which is used to communicate with a host computer and send the air tightness of the soft bag packaging seal, the vacuum degree of the sealing cavity, etc. to the host computer.

[0033] The second embodiment of the present invention provides a soft bag packaging sealing integrity leakage detection instrument. Since the image information of the standard qualified soft bag packaging at the emptyness of the set value is input into the second input layer in advance, and a mapping relationship between the image information of the soft bag packaging at the emptyness of the set value and the airtightness obtained by measuring it is established, the image information of the standard qualified soft bag packaging at the emptyness of the set value is learned into the second two-dimensional neuron middle layer. During measurement, the image information of the soft bag packaging to be detected at the emptyness of the set value is input into the input layer of the second airtightness detection module, and the similarity between the image information of the soft bag packaging to be detected at the emptyness of the set value and the image information of the standard qualified soft bag packaging at the emptyness of the set value in the second two-dimensional neuron middle layer is calculated. The airtightness of the standard qualified soft bag packaging with the largest similarity is corrected by the correction factor to be the second airtightness of the soft bag packaging to be detected;

[0034] The second embodiment of the present invention uses a large language model with a frozen part of the layer to fit the relationship between the vacuum degree of the sealed cavity and the volume of the sealed cavity, the flow rate of the vacuum pump, and the working time of the vacuum pump. Therefore, when measuring, the vacuum degree of the sealed cavity can be inferred as long as the volume of the sealed cavity, the flow rate of the vacuum pump and the working time are known, and there is no need to use other instruments for measurement, which saves costs.

[0035] Third embodiment

[0036] Figure 3 FIG. 1 is a block diagram of a soft bag packaging seal integrity leakage detection instrument provided by the third embodiment of the present invention. Figure 3 As shown, the soft bag package sealing integrity leakage detection instrument provided by the third embodiment of the present invention includes a sealing chamber, a pressure sensor, an image sensor and a first processor. The soft bag package to be detected is placed in the sealing chamber, and the sealing chamber is evacuated by a vacuum pump. When the vacuum degree in the sealing chamber is lower than a set value, the evacuation is stopped; the pressure sensor detects the pressure generated by the expansion of the soft bag package to be detected, and converts the pressure value into pressure electrical information; the image sensor is used to obtain an image of the inside of the sealing chamber including the soft bag package to be detected at the vacuum degree of the set value to obtain image electrical information, and the first processor includes: a feature extraction unit, a convolutional neural network A network, a first airtightness detection module, a second airtightness detection module and a calculation module, wherein the first airtightness detection module outputs the first airtightness of the soft bag package seal to be detected according to the pressure electrical information; the feature extraction unit extracts feature information according to the image; the convolutional neural network extracts the image information of the soft bag package to be detected at the emptyness of the set value according to the feature information; the second airtightness detection module outputs the second airtightness of the soft bag package seal to be detected according to the image information of the soft bag package to be detected at the emptyness of the set value; the calculation module calculates and averages the first airtightness and the second airtightness to obtain the airtightness of the soft bag package seal.

[0037] The first airtightness detection module is trained by a first neural network, which includes a first input layer and a first two-dimensional neural network intermediate layer. During training, the pressure electrical information of a standard qualified soft bag package under the vacuum degree of the set value and the airtightness of the standard qualified soft bag package sealed by measuring it are input into the first input layer, and the pressure electrical information of the standard qualified soft bag package and the airtightness measurement are learned into the first two-dimensional neural network intermediate layer.

[0038] During measurement, the pressure information of the soft bag package to be tested under the set vacuum degree is input into the first input layer of the first airtightness detection module, the pressure electrical information of the soft bag package to be tested is clustered with the pressure information of the standard qualified soft bag package in the first two-dimensional neuron middle layer, and the airtightness of the standard qualified soft bag package in the first two-dimensional neuron middle layer with the smallest Euler distance is corrected by the correction factor to be the first airtightness of the soft bag package to be tested.

[0039] The second air tightness detection module is trained by a second neural network, which includes a second input layer and a second two-dimensional neural network middle layer. During training, the image information of the standard qualified soft bag packaging at the set value of the empty space is input into the second input layer, and a mapping relationship between the image information of the soft bag packaging at the set value of the empty space and the air tightness obtained by measuring the soft bag packaging is established, and the image information of the standard qualified soft bag packaging at the set value of the empty space is learned into the second two-dimensional neural network middle layer.

[0040] During measurement, the image information of the soft bag package to be tested at the set value of the emptiness is input into the input layer of the second airtightness detection module, and the similarity between the image information of the soft bag package to be tested at the set value of the emptiness and the image information of the standard qualified soft bag package in the second two-dimensional neural middle layer at the set value of the emptiness is calculated, and the airtightness of the standard qualified soft bag package in the second two-dimensional neural middle layer with the greatest similarity is corrected by the correction factor to become the second airtightness of the soft bag package to be tested.

[0041] The calculation module adds the first air tightness and the second air tightness and then takes the average to obtain the air tightness of the soft bag packaging seal to be tested.

[0042] The soft bag packaging seal integrity leakage detection instrument provided in the third embodiment of the present invention further includes: a flow meter, a vacuum pump, the vacuum pump is configured to evacuate the sealed cavity under the control of the first processor, a flow meter is provided between the sealed cavity and the vacuum pump, the flow meter is used to measure the flow rate Q of the vacuum pump; the working time T of the vacuum pump is provided by the first processor; the leakage detection instrument further includes a first input unit, which is used to input instructions and data, the data including the volume V of the sealed cavity. The first processor also includes a data acquisition module and a vacuum degree inference model, the data acquisition module is configured to acquire the volume V of the sealed cavity, the flow rate Q of the vacuum pump and the working time T of the vacuum pump, and the volume V of the sealed cavity, the flow rate Q of the vacuum pump and the working time T of the vacuum pump are input into the vacuum degree inference model, the vacuum degree inference model outputs the vacuum degree of the sealed cavity, and the vacuum degree inference model is trained by a large language model with some layers frozen.

[0043] In the third embodiment, the method for training the large language model with some layers frozen is the same as the method for training the large language model with some layers frozen in the first embodiment, and will not be repeated here.

[0044] The soft bag packaging seal integrity leakage detection instrument provided in the third embodiment of the present invention also includes a first display, which is a human-machine interface for displaying input data and instructions, and also for displaying measurement results, wherein the measurement results include the airtightness of the soft bag packaging seal, the vacuum degree of the sealing cavity, etc.

[0045] The soft bag packaging seal integrity leakage detection instrument provided in the third embodiment of the present invention also includes a first memory, which is used to store an application for measuring the seal integrity of the soft bag packaging, and the application includes a feature extraction unit, a convolutional neural network, a first airtightness detection module, a second airtightness detection module, a data acquisition module, and a vacuum degree inference module, etc.

[0046] The soft bag packaging seal integrity leakage detection instrument provided in the third embodiment of the present invention further includes a first communication unit, which is used to communicate with a host computer and send the air tightness of the soft bag packaging seal, the vacuum degree of the sealing cavity, etc. to the host computer.

[0047] The soft bag packaging sealing integrity leakage detection instrument provided by the third embodiment of the present invention includes the advantages of the soft bag packaging sealing integrity leakage detection instruments provided by the first and second embodiments, which will not be described again here.

[0048] Fourth embodiment

[0049] Figure 4 FIG. 4 is a block diagram of a soft bag packaging seal integrity leakage detection instrument provided by the fourth embodiment of the present invention. Figure 4 As shown, the fourth embodiment of the present invention provides a soft bag package sealing integrity leakage detection instrument comprising a sealing chamber, a pressure sensor, an image sensor and a first processor. The soft bag package to be detected is placed in the sealing chamber, and the sealing chamber is evacuated by a vacuum detector. When the vacuum degree in the sealing chamber is lower than a set value, the evacuation is stopped; the pressure sensor detects the pressure generated by the expansion of the soft bag package to be detected, and converts the pressure value into pressure electrical information; the image sensor is used to obtain an image of the inside of the sealing chamber including the soft bag package to be detected at the vacuum degree of the set value to obtain image electrical information, and the first processor comprises: a feature extraction unit, a volume A convolutional neural network, a first airtightness detection module, a second airtightness detection module and a calculation module, wherein the first airtightness detection module outputs the first airtightness of the soft bag package to be detected according to the pressure electrical information; the feature extraction unit extracts feature information according to the image; the convolutional neural network extracts the image information of the soft bag package to be detected at the emptyness of the set value according to the feature information; the second airtightness detection module outputs the second airtightness of the soft bag package to be detected according to the image information of the soft bag package to be detected at the emptyness of the set value; the calculation module calculates the airtightness of the soft bag package to be detected according to the first airtightness and the first airtightness.

[0050] The first airtightness detection module is trained by a first neural network, which includes a first input layer and a first two-dimensional neural network intermediate layer. During training, the pressure electrical information of a standard qualified soft bag package under the vacuum degree of the set value and the airtightness obtained by measuring it are input into the first input layer, and the pressure electrical information of a standard qualified soft bag package and the airtightness obtained by measuring it are learned into the first two-dimensional neural network intermediate layer.

[0051] During measurement, the pressure information of the soft bag package to be tested under the set vacuum degree is input into the first input layer of the first airtightness detection module, the pressure electrical information of the soft bag package to be tested is clustered with the pressure information of the standard qualified soft bag package in the first two-dimensional neuron middle layer, and the airtightness of the standard qualified soft bag package in the first two-dimensional neuron middle layer with the smallest Euler distance is corrected by the correction factor to be the first airtightness of the soft bag package to be tested.

[0052] The second air tightness detection module is trained by a second neural network, which includes a second input layer and a second two-dimensional neural network intermediate layer. During training, the image information of the standard qualified soft bag packaging at the set value of the empty space is input into the second input layer, and a mapping relationship between the image information of the soft bag packaging at the set value of the empty space and the air tightness obtained by measuring the soft bag packaging is established, and the image information of the standard qualified soft bag packaging at the set value of the empty space is learned into the second two-dimensional neural network intermediate layer.

[0053] During measurement, the image information of the soft bag package to be tested at the set value of the emptiness is input into the input layer of the second airtightness detection module, and the similarity between the image information of the soft bag package to be tested at the set value of the emptiness and the image information of the standard qualified soft bag package in the second two-dimensional neural middle layer at the set value of the emptiness is calculated, and the airtightness of the standard qualified soft bag package in the second two-dimensional neural middle layer with the greatest similarity is corrected by the correction factor to become the second airtightness of the soft bag package to be tested.

[0054] The calculation module adds the first air tightness and the second air tightness and then takes the average to obtain the air tightness of the soft bag packaging seal to be tested.

[0055] The soft bag packaging seal integrity leakage detection instrument provided in the fourth embodiment of the present invention also includes a first display, which is a human-machine interface for displaying input data and instructions, and also for displaying measurement results, wherein the measurement results include the airtightness of the soft bag packaging seal.

[0056] The soft bag packaging seal integrity leakage detection instrument provided in the fourth embodiment of the present invention also includes a first memory, which is used to store an application for measuring the seal integrity of the soft bag packaging, and the application includes a judgment neural network, a first airtightness detection module, a second airtightness detection module and a calculation module.

[0057] The soft bag packaging seal integrity leakage detection instrument provided by the first embodiment of the present invention further includes a first communication unit, which is used to communicate with the host computer and the vacuum degree detector, and send the airtightness of the soft bag packaging seal to the host computer.

[0058] Figure 5 : is a block diagram of a vacuum degree detector provided by a fourth embodiment of the present invention, such as Figure 5 As shown, the vacuum degree detector provided by the second embodiment of the present invention includes: a flow meter, a vacuum pump and a second processor, the vacuum pump is configured to evacuate the sealed cavity under the control of the processor, a flow meter is provided between the sealed cavity and the vacuum pump, and the flow meter is used to measure the flow rate Q of the vacuum pump; the working time T of the vacuum pump is provided by the second processor; the vacuum degree detector also includes an input unit, which is used to input instructions and data, and the data includes the volume V of the sealed cavity. The processor includes a data acquisition module and a vacuum degree inference model, the data acquisition module is configured to acquire the volume V of the sealed cavity, the flow rate Q of the vacuum pump and the working time T of the vacuum pump, and the volume V of the sealed cavity, the flow rate Q of the vacuum pump and the working time T of the vacuum pump are input into the vacuum degree inference model, and the vacuum degree inference model outputs the vacuum degree of the sealed cavity, and the vacuum degree inference model is trained by a large language model with some layers frozen.

[0059] In the fourth embodiment, the method for training the large language model with some layers frozen is the same as the method for training the large language model with some layers frozen in the first embodiment, and will not be repeated here.

[0060] The vacuum degree detector provided in the fourth embodiment of the present invention further includes a second display, which is a human-machine interface for displaying input data and instructions, and also for displaying measurement results, wherein the measurement results include the vacuum degree of the sealed cavity.

[0061] The vacuum degree detector provided in the fourth embodiment of the present invention further includes a second memory for storing an application for vacuum degree measurement, wherein the application includes a vacuum degree inference model.

[0062] The vacuum degree detector provided in the fourth embodiment of the present invention also includes a second communication unit, which is used to communicate with the host computer and the soft bag packaging sealing integrity leakage detection instrument to send the vacuum degree of the sealing cavity to the host computer and the soft bag packaging sealing integrity leakage detection instrument.

[0063] The flexible bag packaging sealing integrity leakage detection instrument provided by the fourth embodiment of the present invention includes the advantages of the flexible bag packaging sealing integrity leakage detection instruments provided by the first and second embodiments, which will not be described again here.

[0064] Fifth embodiment

[0065] The fifth embodiment of the present invention only describes the contents that are different from the first to fourth embodiments, and the same contents will not be described repeatedly.

[0066] The fifth embodiment of the present invention provides a method for detecting the sealing integrity of a soft bag package, comprising the following steps: Place the soft bag package to be tested in the sealed cavity, and use a vacuum pump to evacuate the sealed cavity. When the vacuum degree in the sealed cavity is lower than the set value, stop evacuating; Using a pressure sensor to detect the pressure generated by the expansion of the soft bag package to be detected, and converting the pressure value into pressure electrical information; The first air tightness of the soft bag package to be detected is output according to the pressure electrical information through the first air tightness detection module. The first air tightness detection module is trained by a first neural network. The first neural network includes a first input layer and a first two-dimensional neural middle layer. During training, the pressure electrical information of the standard qualified package under the vacuum degree of the set value and the air tightness obtained by measuring it are input into the first input layer, and the pressure electrical information of the standard qualified soft bag package and the air tightness obtained by measuring it are learned into the first two-dimensional neural middle layer. During measurement, the pressure information of the soft bag package to be detected under the vacuum degree of the set value is input into the first input layer of the first air tightness detection module, so that the pressure electrical information of the soft bag package to be detected is clustered with the pressure information of the standard qualified soft bag package in the first two-dimensional neural middle layer, and the air tightness of the standard qualified soft bag package with the smallest Euler distance is corrected by the correction factor to be the first air tightness of the soft bag package to be detected.

[0067] The method for detecting leakage of sealing integrity of a soft bag package provided in the fifth embodiment of the present invention further includes the following steps: obtaining an image of the interior of a sealing cavity of a soft bag package to be detected at a vacuum degree of a set value by an image sensor to obtain image electrical information, and extracting feature information based on the image electrical information by a feature extraction unit; extracting image information of the soft bag package to be detected at the empty degree of the set value based on the feature information by a convolutional neural network; and outputting a second airtightness of the seal of the soft bag package to be detected based on the image information of the soft bag package to be detected at the empty degree of the set value by a second airtightness detection module.

[0068] The soft bag packaging seal integrity leakage detection method provided in the fifth embodiment of the present invention further comprises the following steps: adding the first air tightness and the second air tightness by a calculation module and then averaging them to obtain the air tightness of the soft bag packaging seal to be detected.

[0069] The fifth embodiment of the present invention provides a method for detecting the sealing integrity of a soft bag package, and further includes the following steps: The volume V of the sealed chamber, the flow rate Q of the vacuum pump and the working time T of the vacuum pump are obtained through the data acquisition module; The volume V of the sealed chamber, the flow rate Q of the vacuum pump and the working time T of the vacuum pump are input into the vacuum degree inference model, and the vacuum degree inference model outputs the vacuum degree of the sealed chamber. The vacuum degree inference model is trained by a large language model with some layers frozen.

[0070] The soft bag packaging seal integrity leakage detection method provided by the fifth embodiment of the present invention includes the advantages of the soft bag packaging seal integrity leakage detection instruments provided by the first and second embodiments, which will not be described again here.

[0071] The above shows and describes the basic principles and main features of the present invention and the advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments. The above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A soft bag package sealing integrity leakage detection instrument, comprising a sealing chamber, a pressure sensor and a first processor, wherein the soft bag package to be detected is placed in the sealing chamber, and the sealing chamber is evacuated by a vacuum pump. When the vacuum degree in the sealing chamber is lower than a set value, the evacuation is stopped; the pressure sensor detects the pressure generated by the expansion of the soft bag package to be detected, and converts the pressure value into pressure electrical information; It is characterized in that The first processor includes: a first airtightness detection module, the first airtightness detection module outputs the first airtightness of the soft bag package to be detected according to the pressure electrical information, the first airtightness detection module is trained by a first neural network, the first neural network includes a first input layer and a first two-dimensional neuron middle layer, during training, the pressure electrical information of the standard-qualified soft bag package under the vacuum degree of the set value and the airtightness obtained by measuring the standard-qualified soft bag package are input into the first input layer, and the pressure electrical information of the standard-qualified soft bag package and the airtightness obtained by measuring the standard-qualified soft bag package are learned into the first two-dimensional neuron middle layer, during measurement, the pressure information of the soft bag package to be detected under the vacuum degree of the set value is input into the first input layer of the first airtightness detection module, so that the pressure electrical information of the soft bag package to be detected is clustered with the pressure information of the standard-qualified soft bag package in the first two-dimensional neuron middle layer, and the airtightness of the standard-qualified soft bag package in the first two-dimensional neuron middle layer with the smallest Euler distance is corrected by the correction factor to be the first airtightness of the soft bag package to be detected.

2. The soft bag packaging seal integrity leakage detection instrument according to claim 1, It is characterized in that Also includes: The image sensor is used to obtain an internal image of a sealed cavity including a vacuum degree of a soft bag package to be detected at a set value to obtain image electrical information. The first processor includes: a feature extraction unit, a convolutional neural network, and a second airtightness detection module, wherein the feature extraction unit extracts feature information according to the image electrical information; the convolutional neural network extracts image information of the soft bag package to be detected at the empty degree of the set value according to the feature information; and the second airtightness detection module outputs the second airtightness of the soft bag package to be detected according to the image information of the soft bag package to be detected at the empty degree of the set value.

3. The soft bag packaging seal integrity leakage detection instrument according to claim 2, It is characterized in that The first processor further includes a calculation module, which adds the first air tightness and the second air tightness and then takes an average to obtain the air tightness of the soft bag packaging seal to be tested.

4. The soft bag packaging seal integrity leakage detection instrument according to claim 3, It is characterized in that The second air tightness detection module is trained by a second neural network, which includes a second input layer and a second two-dimensional neural network middle layer. During training, the image information of the standard qualified soft bag package at the set value of the empty space is input into the second input layer, and a mapping relationship between the image information of the soft bag package at the set value of the empty space and the air tightness obtained by measuring the soft bag package is established, and the image information of the standard qualified soft bag package at the set value of the empty space and the measured air tightness are learned into the second two-dimensional neural network middle layer; during measurement, the image information of the soft bag package to be detected at the set value of the empty space is input into the input layer of the second air tightness detection module, and the similarity between the image information of the soft bag package to be detected at the set value of the empty space and the image information of the standard qualified soft bag package at the set value of the empty space in the second two-dimensional neural network middle layer is calculated, and the air tightness of the standard qualified soft bag package in the second two-dimensional neural network middle layer with the largest similarity is corrected by the correction factor to be the second air tightness of the soft bag package to be detected.

5. The soft bag packaging seal integrity leakage detection instrument according to any one of claims 1 to 4, It is characterized in that The first processor includes a data acquisition module and a vacuum degree inference model. The data acquisition module is configured to acquire the volume V of the sealed chamber, the flow rate Q of the vacuum pump and the working time T of the vacuum pump. The first processor includes a vacuum degree inference model. The volume V of the sealed chamber, the flow rate Q of the vacuum pump and the working time T of the vacuum pump are input into the vacuum degree inference model. The vacuum degree inference model outputs the vacuum degree of the sealed chamber. The vacuum degree inference model is trained by a large language model with some layers frozen.

6. The soft bag packaging seal integrity leakage detection instrument according to claim 5, It is characterized in that Methods for training a large language model with some layers frozen include: Step 1: Take N data sets from the database, each of which includes a pre-trained data vector And the vacuum degree of the sealed chamber tested , the pre-training data vector Input to the input embedding layer of the large language model with some layers frozen, and the estimated vacuum degree of the sealed cavity is input from the second residual connection layer ,n=1,…,N; , in, , To freeze the current parameter vector of the large language model with some layers, is the flow rate of the vacuum pump in the nth data set, is the working time of the vacuum pump in the nth data group; Step 2: Let the loss function L be: , The gradient of the neural network parameters is calculated according to the following formula: ; Step 3: Update the current parameter vector of the large language model with some layers frozen according to the following formula: ; In the formula, is the learning rate, represents the gradient of ; Step 4: Determine whether the loss function L is the minimum. If so, end the training. If not, replace the current parameter vector with the new parameter vector and return to step 1.

7. A method for detecting the sealing integrity of a soft bag package. It is characterized in that The steps include: Place the soft bag package to be tested in the sealed cavity, and use a vacuum pump to evacuate the sealed cavity. When the vacuum degree in the sealed cavity is lower than the set value, stop evacuating; The pressure sensor is used to detect the pressure generated by the expansion of the soft bag package to be detected in the sealing chamber under the set vacuum degree, and the pressure value is converted into pressure electrical information; The first airtightness of the soft bag package to be tested is output according to the pressure electrical information through the first airtightness detection module. The first airtightness detection module is trained by a first neural network. The first neural network includes a first input layer and a first two-dimensional neuron middle layer. During training, the pressure electrical information of the standard qualified soft bag package under the vacuum degree of the set value and the measured airtightness are input into the first input layer, and the pressure electrical information and the measured airtightness of the standard qualified soft bag package are learned into the first two-dimensional neuron middle layer; during measurement, the pressure information of the soft bag package to be tested under the vacuum degree of the set value is input into the first input layer of the first airtightness detection module, so that the pressure electrical information of the soft bag package to be tested and the pressure information of the standard qualified soft bag package in the first two-dimensional neuron middle layer are clustered, and the airtightness of the standard qualified soft bag package with the smallest Euler distance is corrected by the correction factor to be the first airtightness of the soft bag package to be tested.

8. The method for detecting the sealing integrity of a soft bag package according to claim 7, It is characterized in that The following steps are also included: An image sensor is used to acquire an image of the interior of a sealed cavity including a vacuum degree of a set value of the soft bag package to be inspected to obtain image electrical information, and a feature extraction unit is used to extract feature information based on the image electrical information; a convolutional neural network is used to extract image information of the soft bag package to be inspected at the empty degree of the set value based on the feature information; and a second airtightness detection module is used to output the second airtightness of the seal of the soft bag package to be inspected based on the image information of the soft bag package to be inspected at the empty degree of the set value.

9. The method for detecting the sealing integrity of a soft bag package according to claim 8, It is characterized in that The following steps are also included: The calculation module adds the first air tightness and the second air tightness and then takes the average to obtain the air tightness of the soft bag packaging seal to be tested.

10. The method for detecting the sealing integrity of a soft bag package according to any one of claims 7 to 9, It is characterized in that The following steps are also included: The volume V of the sealed chamber, the flow rate Q of the vacuum pump and the working time T of the vacuum pump are obtained through the data acquisition module; The volume V of the sealed chamber, the flow rate Q of the vacuum pump and the working time T of the vacuum pump are input into the vacuum degree inference model, and the vacuum degree inference model outputs the vacuum degree of the sealed chamber. The vacuum degree inference model is trained by a large language model with some layers frozen.

Citation Information

Patent Citations

  • Method for detecting leakage of pipeline based on artificial neural network

    CN101008992A

  • Cosmetic package quality detection method and system

    CN113706527A

  • Condenser vacuum degree prediction method based on residual memory network

    CN114819354A

  • Ostomy bag sealing performance detection device

    CN117309278A

  • Visual detection method and system for food packaging air tightness

    CN118172359A

Cited By

  • Intelligent control system and method of soft bag package sealing integrity leakage detector

    CN121275259A

  • Intelligent control system and method for soft bag package seal integrity leak detector

    CN121275259B