An online testing system and method for heat-sealing performance of large flexible packaging

Abstract

This invention relates to the field of heat-sealing performance quality inspection technology for flexible packaging, and discloses an online detection system and method for the heat-sealing performance of large-scale flexible packaging. Specifically, it combines infrared thermal imaging detection technology, millimeter-wave imaging detection technology, and AI visual inspection technology. After the heat-sealing process, infrared thermal imaging detection technology is used to detect the temperature distribution during heat sealing, achieving visual quality inspection of the heat-sealing process. Then, millimeter-wave imaging detection technology is used to detect the physical structure inside the seal after heat sealing, achieving visual quality inspection of the heat-sealed product. This invention, through the sequential detection order of "infrared thermal imaging first, then millimeter-wave imaging," ensures that a product is only deemed qualified if it simultaneously passes two detection checkpoints based on different physical principles, and the results of both checkpoints are verified by the host computer. This greatly improves the overall reliability of the detection system.

Description

Technical Field

[0001] This invention relates to the field of quality inspection technology for heat sealing processes of flexible packaging, and in particular to an online testing system and method for the heat sealing performance of large flexible packaging. Background Technology

[0002] Flexible packaging is gaining increasing popularity in the packaging industry due to its advantages such as low material consumption, lightweight materials, low cost, and ease of carrying. Heat sealing is a widely used and relatively simple sealing method, especially for bagged packaging. However, during the sealing process, factors such as heat-pressing time, temperature, pressure, product contamination, and environmental pollution can easily lead to heat-sealing defects such as bubbles, inclusions, leakage channels, and punctures. These defects, especially leakage channels, pose a serious threat to product quality and safety, resulting in frequent food and drug safety incidents. Currently, methods such as vacuum pressure difference, color impregnation, and tearing are commonly used for product sampling inspection in production. However, these methods are time-consuming, involve complex testing procedures, waste products, and cannot achieve 100% online detection.

[0003] Against this backdrop, rapid, effective, and reliable non-destructive testing (NDT) technologies have been developed. Currently, commonly used NDT methods include optical methods, X-ray methods, infrared detection methods, and ultrasonic methods, but reports on the use of millimeter-wave NDT technology for detecting defects in packaging heat seals are rare. Summary of the Invention

[0004] This invention combines infrared thermal imaging detection technology, millimeter-wave imaging detection technology, and AI visual inspection technology to achieve an intelligent quality inspection closed-loop system for the heat-sealing performance of large flexible packaging components.

[0005] A method for online testing of the heat-sealing performance of large flexible packaging includes the following steps:

[0006] Step 1: The host computer sets the operating parameters for data interaction with the AI ​​visual inspection component. These operating parameters are:

[0007] Let p be a large prime number, and choose an additive cyclic group G of order p, whose generator is G0;

[0008] Define the following hash function:

[0009] H a {0,1} * ×G×G→Z p * Z p * Denotes the set of non-zero elements of the remainder class field modulo p;

[0010] H b{0,1} * ×G×G→{0,1} * {0,1} * Represents a string consisting of any combination of 0s and 1s;

[0011] The host computer is configured with metadata to assist the AI ​​vision inspection component in delivering inspection data. This metadata is: the production code (Pc) of the large flexible package. i ∈{0,1} * Detection serial number Csn i ∈Z p * and the detection task execution number Dtec i =Csn i +[P Ⅰ ×H a (Pc i Csn i ×G0, P Ⅱ )](mod p); where P Ⅰ ∈Z p * and P Ⅱ =P Ⅰ ×G0 is a fixed system parameter;

[0012] Set the auxiliary interaction parameter set {G0, H} b Pc i Csn i Dtec i Send it to the AI ​​visual inspection component;

[0013] Step 2: After the heat sealing process is completed and the inspection serial number Csn is printed... i When large flexible packages arrive at the infrared thermal imaging inspection station, the PLC controller sends a signal to the AI ​​vision inspection component C. IR The AI ​​visual inspection component C issues a trigger signal to begin detection. IR Perform the following testing procedures:

[0014] The entire heat-sealing area of ​​a large flexible package is scanned, and an infrared thermal image and temperature field distribution data of that area are generated simultaneously. This temperature field distribution data is denoted as Tfdd. i ∈{0,1} * ;

[0015] The built-in AI vision model analyzes the infrared thermal image and outputs the infrared detection result R regarding the heat sealing quality. IR ;

[0016] Based on temperature field distribution data Tfdd i and auxiliary interaction parameter set {G0, H b Pci Csn i Dtec i}, Calculate the raw temperature field distribution data delivery format Deld i (Tfdd i )=Tfdd i ⊕H b (Pc i Csn i ×G0, Dtec i ×G0);

[0017] The infrared detection result R IR Deld, the delivery format of the original temperature field distribution data i (Tfdd i (To be delivered to the host computer together)

[0018] Step 3: Once the infrared thermal imaging inspection process is completed and the inspection serial number Csn is printed on... i When the large flexible package arrives at the millimeter-wave imaging inspection station, the PLC controller sends a signal to the AI ​​vision inspection component C. mmWave Send a trigger signal to start the detection;

[0019] AI visual inspection component C mmWave Perform the following testing procedures:

[0020] A broadband millimeter-wave pulse is emitted into the entire heat-sealing area of ​​a large flexible package, and the returned emitted signal is captured in real time. Simultaneously, a high-resolution image and dielectric constant distribution data of the area are generated, denoted as Dcdd. i ∈{0,1} * ;

[0021] The built-in AI vision model analyzes high-resolution images and outputs millimeter-wave detection results (R) regarding heat-sealing quality. mmWave ;

[0022] Based on dielectric constant distribution data Dcdd i and auxiliary interaction parameter set {G0, H b Pc i Csn i Dtec i}, Calculate the original dielectric constant distribution data delivery format Deld i (Dcdd i )=Dcdd i ⊕H b (Pc i Csn i ×G0, Dtec i ×G0);

[0023] The millimeter wave detection result R mmWave Deld, the original dielectric constant distribution data delivery format i (Dcdd i (To be delivered to the host computer together)

[0024] Step four: After receiving the detection data from the AI ​​vision detection component, the host computer performs the following operations:

[0025] Based on the auxiliary interaction parameter set {G0, H} a H b Pc i Csn i P Ⅱ Deld, the raw temperature field distribution data delivery format i (Tfdd i ), calculate temperature field distribution data ETfdd i :

[0026] ETfdd i =Deld i (Tfdd i )⊕H b [Pc i Csn i ×G0,Csn i ×G0+P Ⅱ ×H a (Pc i Csn i ×G0, P Ⅱ )];

[0027] Temperature field distribution data ETfdd i Converted to infrared thermal imaging image I (Tfdd) i The AI ​​vision model running on the host computer analyzes the infrared thermal image I(Tfdd). i The analysis is performed, and the infrared detection results R' regarding the heat sealing quality are output. IR ;

[0028] Based on the auxiliary interaction parameter set {G0, H} a H b Pc i Csn i P Ⅱ} and the raw dielectric constant distribution data delivery format Deld i (Dcdd i ), calculate the dielectric constant distribution data EDcdd i :

[0029] EDcdd i =Deld i (Dcddi )⊕H b [Pc i Csn i ×G0,Csn i ×G0+P Ⅱ ×H a (Pc i Csn i ×G0, P Ⅱ )];

[0030] Dielectric constant distribution data EDcdd i Convert to high-resolution image I(Dcdd) i The AI ​​vision model running on the host computer analyzes the infrared thermal image I(Dcdd). i The analysis is performed, and the millimeter-wave detection results R' regarding the heat seal quality are output. mmWave ;

[0031] Run the intelligent fusion judgment algorithm to perform data fusion judgment and generate the final quality inspection result regarding the heat sealing quality: If R' IR R IR 、R' mmWave and R mmWave If all results are satisfactory, the quality inspection of the heat-sealing performance of the large flexible packaging is deemed satisfactory; otherwise, it is deemed unsatisfactory, and the unsatisfactory inspection result is sent to the PLC controller.

[0032] A large-scale flexible packaging heat-sealing performance online testing system, the system being used to execute the aforementioned large-scale flexible packaging heat-sealing performance online testing method, includes a host computer and an AI vision inspection component C. IR AI visual inspection component C mmWave and PLC controller.

[0033] Preferably, AI visual inspection component C IR It consists of an infrared thermal imager, an AI analysis system, and a data interaction module;

[0034] Infrared thermal imagers are used to generate infrared thermal images and temperature field distribution data of the entire heat-sealing area of ​​large flexible packaging components. On the one hand, the infrared thermal images are transmitted to an AI analysis system with a built-in AI vision model for heat-sealing process quality inspection. On the other hand, the temperature field distribution data is transmitted to a data interaction module for raw data delivery processing.

[0035] Preferably, AI visual inspection component C mmWave It consists of a millimeter-wave detector, an AI analysis system, and a data interaction module;

[0036] The millimeter-wave detector is used to generate high-resolution images and dielectric constant distribution data of the entire heat-sealing area of large flexible packages. On the one hand, the high-resolution images are transmitted to an AI analysis system with an AI vision model for product quality inspection, and on the other hand, the dielectric constant distribution data is transmitted to a data interaction module for raw data delivery processing.

[0037] Preferably, the system further includes a photoelectric sensor, a serial number printing module, a defective product inkjet marking module, and a defective product rejection module;

[0038] The photoelectric sensor is used to continuously emit and detect light beams. The physical level signal output by the photoelectric sensor is continuously monitored by the PLC controller. If a large flexible package is detected, the physical level signal output by the photoelectric sensor will change immediately. Accordingly, the PLC controller first triggers the serial number printing module to print a unique detection serial number at a designated position on the large flexible package, then triggers the quality inspection program, then triggers the defective product inkjet marking module to print an obvious "REJECT" mark at a designated position on the defective product, and finally triggers the defective product rejection module to remove the defective product from the production line.

[0039] The beneficial effects of the present invention are as follows:

[0040] The heat-sealing process of large flexible packages is as follows: It is hot-pressed at a temperature of 120 - 180 °C. After hot-pressing, within 30 - 60 s, the temperature of the hot-pressed part can maintain a residual temperature of 60 - 80 °C. After this heat-sealing process, infrared thermal imaging detection technology is used to detect the temperature distribution during the heat-sealing process to achieve visual quality inspection of the heat-sealing process. Then, millimeter-wave imaging detection technology is used to detect the physical structure inside the seal after heat-sealing to achieve visual quality inspection of the quality of heat-sealed products;

[0041] The above serial detection sequence of "first infrared thermal imaging, then millimeter-wave imaging" enables a product to be judged as qualified only when it has passed two detection levels based on different physical principles simultaneously and the detection results of both detection levels have passed the review of the host computer. This technical means greatly improves the overall reliability of the detection system and effectively overcomes the situation where some defects may not be obvious under one type of detection but will be clearly revealed under another type of detection. For example, a slight contamination with normal temperature performance may be clearly shown in the millimeter-wave image due to dielectric constant differences. Specific embodiments

[0042] The online detection scheme for the heat-sealing performance of large flexible packages has the following specific implementation process:

[0043] Process 1: Carry out serial identification on the large flexible package to be detected. The specific process is as follows:

[0044] The photoelectric sensor I installed at the entrance station of the heat sealing inspection production line continuously emits and detects the light beam. The physical level signal output by the photoelectric sensor I is continuously monitored by the PLC controller. When the large flexible package to be inspected is delivered to the entrance station of the heat sealing inspection production line by the conveyor belt, the physical level signal output by the photoelectric sensor I changes immediately. Based on this, the PLC controller knows that the large flexible package has arrived at the entrance station of the heat sealing inspection production line.

[0045] The PLC controller generates a unique inspection serial number Csn for the large flexible package to be inspected. i ∈Z p * (p is a large prime number, Z) p * (representing the set of non-zero elements of the remainder class domain modulo p), and assigning the detection index Csn i The command is sent to the serial number printing module, which then prints a unique inspection serial number (Csn) at a designated location on the large flexible package to be inspected. i The detection serial number Csn i The barcode is read and uploaded to the host computer by a barcode reader installed after the printing process.

[0046] Furthermore, the barcode reader is used to inspect the production codes printed on the large flexible packaging components. i (Pc i ∈{0,1} * {0,1} * The string (representing any combination of 0s and 1s) is read and uploaded to the host computer.

[0047] At this point, the host computer performs the following operations:

[0048] Configure the operating parameters for data interaction between the host computer and the AI ​​vision inspection component. These operating parameters specifically include:

[0049] Choose an additive cyclic group G of order p, where one generator of the cyclic group G is G0;

[0050] Configure the following secure hash function:

[0051] H a {0,1} * ×G×G→Z p * ;

[0052] H b {0,1} * ×G×G→{0,1} * ;

[0053] Configure metadata to assist the AI ​​vision inspection component in delivering inspection data to the host computer. This metadata specifically includes:

[0054] Production code Pc for large flexible packaging i ∈{0,1} * ;

[0055] Inspection serial number Csn for large flexible packaging i ∈Z p * ;

[0056] Inspection task execution number for large flexible packaging components: Dtec i :

[0057] Dtec i =Csn i +[P Ⅰ ×H a (Pc i Csn i ×G0, P Ⅱ )](modp);

[0058] Among them, P Ⅰ (P Ⅰ ∈Z p * ) and P Ⅱ (P Ⅱ =P Ⅰ ×G0) are fixed parameters of the system;

[0059] Set the auxiliary interaction parameter set {G0, H} b Pc i Csn i Dtec i Send to AI visual inspection component C IR and AI visual inspection component C mmWave Data interaction module;

[0060] Step two involves performing dual non-destructive quality inspection on large flexible packaging components using both infrared thermal imaging and millimeter-wave imaging. The specific process is as follows:

[0061] When the test serial number Csn is printed i When large flexible packages are delivered to the infrared thermal imaging detection station by the conveyor belt, the physical level signal output by the photoelectric sensor II installed at the station changes immediately.

[0062] Based on this, the PLC controller sends a signal to the AI ​​vision inspection component C. IR (Composed of an infrared thermal imager, an AI analysis system, and a data interaction module) sends out a trigger signal to begin detection;

[0063] AI visual inspection component C IR Perform the following testing procedures:

[0064] An infrared thermal imager scans the entire heat-sealed area of ​​a large flexible package, simultaneously generating an infrared thermal image of that area and temperature field distribution data, denoted as Tfdd. i ∈{0,1} * ;

[0065] The AI ​​analysis system's built-in AI vision model analyzes the infrared thermal image and outputs infrared detection results (R) regarding the heat-sealing quality. IR (The quality inspection results include pass / fail and specific defect classifications) and are sent to the data interaction module;

[0066] The data interaction module uses the auxiliary interaction parameter set {G0, H} b Pc i Csn i Dtec i}, Calculate the raw temperature field distribution data delivery format Deld i (Tfdd i ):

[0067] Deld i (Tfdd i )=Tfdd i ⊕H b (Pc i Csn i ×G0, Dtec i ×G0);

[0068] At this point, the data interaction module will transmit the infrared detection result R... IR Deld, the delivery format of the original temperature field distribution data i (Tfdd i (To be delivered to the host computer together)

[0069] When the test serial number Csn is printed i When a large flexible package is delivered to the millimeter-wave imaging inspection station by a conveyor belt, the physical level signal output by the photoelectric sensor III installed at the station immediately changes.

[0070] Based on this, the PLC controller sends a signal to the AI ​​vision inspection component C. mmWave (Composed of a millimeter-wave detector, an AI analysis system, and a data interaction module) sends out a trigger signal to begin detection;

[0071] AI visual inspection component C mmWave Perform the following testing procedures:

[0072] A millimeter-wave detector transmits broadband millimeter-wave pulses to the entire heat-sealed area of ​​a large flexible package and captures the returned transmission signals in real time. Simultaneously, it generates high-resolution images and dielectric constant distribution data for the area, denoted as Dcdd. i ∈{0,1} * ;

[0073] The AI ​​analysis system's built-in AI vision model analyzes high-resolution images and outputs millimeter-wave detection results (R) regarding heat-sealing quality. mmWave (The quality inspection results include pass / fail and specific defect classifications) and are sent to the data interaction module;

[0074] The data interaction module uses the auxiliary interaction parameter set {G0, H} b Pc i Csn i Dtec i}, Calculate the original dielectric constant distribution data delivery format Deld i (Dcdd i ):

[0075] Deld i (Dcdd i )=Dcdd i ⊕H b (Pc i Csn i ×G0, Dtec i ×G0);

[0076] At this point, the data interaction module will transmit the millimeter-wave detection results R... mmWave Deld, the original dielectric constant distribution data delivery format i (Dcdd i (To be delivered to the host computer together)

[0077] Step three involves intelligently determining the heat-sealing quality of large flexible packaging items. The specific process is as follows:

[0078] After receiving the detection data from the AI ​​vision detection component, the host computer performs the following operations:

[0079] Based on the auxiliary interaction parameter set {G0, H} a H b Pc i Csn i P Ⅱ Deld, the raw temperature field distribution data delivery format i (Tfdd i ), calculate temperature field distribution data ETfdd i :

[0080] ETfdd i =Deld i (Tfdd i )⊕H b [Pc i Csn i ×G0,Csn i ×G0+P Ⅱ ×H a (Pc i Csn i ×G0, P Ⅱ )];

[0081] Temperature field distribution data ETfdd i Converted to infrared thermal imaging image I (Tfdd) i The AI ​​vision model running on the host computer analyzes the infrared thermal image I(Tfdd). i The analysis is performed, and the infrared detection results R' regarding the heat sealing quality are output. IR (The quality inspection results include pass / fail and specific defect classifications);

[0082] Based on the auxiliary interaction parameter set {G0, H} a H b Pc i Csn i P Ⅱ} and the raw dielectric constant distribution data delivery format Deld i (Dcdd i ), calculate the dielectric constant distribution data EDcdd i :

[0083] EDcdd i =Deld i (Dcdd i )⊕H b [Pc i Csn i ×G0,Csn i ×G0+P Ⅱ ×H a (Pc i Csn i ×G0, P Ⅱ )];

[0084] Dielectric constant distribution data EDcdd i Convert to high-resolution image I(Dcdd) i The AI ​​vision model running on the host computer analyzes the infrared thermal image I(Dcdd). i The analysis is performed, and the millimeter-wave detection results R' regarding the heat seal quality are output. mmWave(The quality inspection results include pass / fail and specific defect classifications);

[0085] The intelligent fusion judgment algorithm is run to perform data fusion judgment and generate the final quality inspection result regarding the heat sealing quality:

[0086] If R' IR It is qualified;

[0087] R IR It is qualified;

[0088] R' mmWave It is qualified;

[0089] R mmWave It is qualified;

[0090] The final quality inspection result for the heat-sealing quality of the large flexible packaging is then determined to be qualified, and the result is recorded as R. Q (Csn i );

[0091] If R' IR R IR 、R' mmWave R mmWave If any one of the following is unqualified, the final quality inspection result for the heat sealing quality of the large flexible packaging is deemed unqualified, and the result is recorded as R. UnQ (Csn i And simultaneously transmit the final quality inspection results R UnQ (Csn i Send to the PLC controller;

[0092] The PLC controller receives the final quality inspection result R. UnQ (Csn i After that, perform the following operations:

[0093] The tracking inkjet prints the detection serial number Csn. i The specific locations on the conveyor belt of the large flexible packaging items that failed the final quality inspection are as follows:

[0094] When the test serial number Csn is printed i When a large flexible package is delivered to the defective product marking station by the conveyor belt, the physical level signal output by the photoelectric sensor IV installed at the station changes immediately; the PLC controller then sends a marking trigger signal to the defective product inkjet marking module; the defective product inkjet marking module prints a clear REJECT mark at the designated position on the large flexible package.

[0095] Subsequently, when a large flexible package bearing the REJECT mark is conveyed to the defective product rejection station by the conveyor belt, the physical level signal output by the photoelectric sensor V installed at the station changes immediately; the PLC controller then sends a rejection trigger signal to the defective product rejection module; the defective product rejection module then removes the defective product from the production line.

[0096] The qualified products are then conveyed to the next process step by the conveyor belt;

[0097] Furthermore, all data generated throughout the entire testing process is saved to a central database;

[0098] This concludes the testing process for the heat-sealing performance of large flexible packaging components.

Claims

1. A method for online testing of the heat-sealing performance of large flexible packaging, characterized in that, Includes the following steps: Step 1: The host computer sets the operating parameters for data interaction with the AI ​​visual inspection component. These operating parameters are: Let p be a large prime number, and choose an additive cyclic group G of order p, whose generator is G0; Define the following hash function: H a {0,1} * ×G×G→Z p * Z p * Denotes the set of non-zero elements of the remainder class field modulo p; H b {0,1} * ×G×G→{0,1} * {0,1} * Represents a string consisting of any combination of 0s and 1s; The host computer is configured with metadata to assist the AI ​​vision inspection component in delivering inspection data. This metadata is: the production code (Pc) of the large flexible package. i ∈{0,1} * Detection serial number Csn i ∈Z p * and the detection task execution number Dtec i =Csn i +[P Ⅰ ×H a (Pc i Csn i ×G0, P Ⅱ )](mod p); where P Ⅰ ∈Z p * and P Ⅱ =P Ⅰ ×G0 is a fixed system parameter; Set the auxiliary interaction parameter set {G0, H} b Pc i Csn i Dtec i Send it to the AI ​​visual inspection component; Step 2: After the heat sealing process is completed and the inspection serial number Csn is printed... i When large flexible packages arrive at the infrared thermal imaging inspection station, the PLC controller sends a signal to the AI ​​vision inspection component C. IR The AI ​​visual inspection component C issues a trigger signal to begin detection. IR Perform the following testing procedures: The entire heat-sealing area of ​​a large flexible package is scanned, and an infrared thermal image and temperature field distribution data of that area are generated simultaneously. This temperature field distribution data is denoted as Tfdd. i ∈{0,1} * ; The built-in AI vision model analyzes the infrared thermal image and outputs the infrared detection result R regarding the heat sealing quality. IR ; Based on temperature field distribution data Tfdd i and auxiliary interaction parameter set {G0, H b Pc i Csn i Dtec i }, Calculate the raw temperature field distribution data delivery format Deld i (Tfdd i )=Tfdd i ⊕H b (Pc i Csn i ×G0, Dtec i ×G0); The infrared detection result R IR Deld, the delivery format of the original temperature field distribution data i (Tfdd i (To be delivered to the host computer together) Step 3: Once the infrared thermal imaging inspection process is completed and the inspection serial number Csn is printed on... i When the large flexible package arrives at the millimeter-wave imaging inspection station, the PLC controller sends a signal to the AI ​​vision inspection component C. mmWave Send a trigger signal to start the detection; AI visual inspection component C mmWave Perform the following testing procedures: A broadband millimeter-wave pulse is emitted into the entire heat-sealing area of ​​a large flexible package, and the returned emission signal is captured in real time. Simultaneously, a high-resolution image and dielectric constant distribution data of the area are generated, denoted as Dcdd. i ∈{0,1} * ; The built-in AI vision model analyzes high-resolution images and outputs millimeter-wave detection results (R) regarding heat-sealing quality. mmWave ; Based on dielectric constant distribution data Dcdd i and auxiliary interaction parameter set {G0, H b Pc i Csn i Dtec i }, Calculate the original dielectric constant distribution data delivery format Deld i (Dcdd i )=Dcdd i ⊕H b (Pc i Csn i ×G0, Dtec i ×G0); The millimeter wave detection result R mmWave Deld, the original dielectric constant distribution data delivery format i (Dcdd i (To be delivered to the host computer together) Step four: After receiving the detection data from the AI ​​vision detection component, the host computer performs the following operations: Based on the auxiliary interaction parameter set {G0, H} a H b Pc i Csn i P Ⅱ Deld, the raw temperature field distribution data delivery format i (Tfdd i ), calculate temperature field distribution data ETfdd i : ETfdd i =Deld i (Tfdd i )⊕H b [Pc i ,Csn i ×G0,Csn i ×G0+P Ⅱ ×H a (Pc i ,Csn i ×G0,P Ⅱ )]; Temperature field distribution data ETfdd i Converted to infrared thermal imaging image I (Tfdd) i The AI ​​vision model running on the host computer analyzes the infrared thermal image I(Tfdd). i The analysis is performed, and the infrared detection results R' regarding the heat sealing quality are output. IR ; Based on the auxiliary interaction parameter set {G0, H} a H b Pc i Csn i P Ⅱ } and the raw dielectric constant distribution data delivery format Deld i (Dcdd i ), calculate the dielectric constant distribution data EDcdd i : EDcdd i =Deld i (Dcdd i )⊕H b [Pc i ,Csn i ×G0,Csn i ×G0+P Ⅱ ×H a (Pc i ,Csn i ×G0,P Ⅱ )]; Dielectric constant distribution data EDcdd i Convert to high resolution image I(Dcdd) i The AI ​​vision model running on the host computer analyzes the infrared thermal image I(Dcdd). i The analysis is performed, and the millimeter-wave detection results R' regarding the heat seal quality are output. mmWave ; Run the intelligent fusion judgment algorithm to perform data fusion judgment and generate the final quality inspection result regarding the heat sealing quality: If R' IR R IR 、R' mmWave and R mmWave If all results are satisfactory, the quality inspection of the heat-sealing performance of the large flexible packaging is deemed satisfactory; otherwise, it is deemed unsatisfactory and sent to the PLC controller.

2. An online testing system for the heat-sealing performance of large flexible packaging, the system being used to execute the online testing method for the heat-sealing performance of large flexible packaging as described in claim 1, characterized in that, Including host computer, AI visual inspection component C IR AI visual inspection component C mmWave and PLC controller.

3. The online testing system for the heat-sealing performance of large flexible packaging according to claim 2, characterized in that, AI visual inspection component C IR It consists of an infrared thermal imager, an AI analysis system, and a data interaction module; Infrared thermal imagers are used to generate infrared thermal images and temperature field distribution data of the entire heat-sealing area of ​​large flexible packaging components. On the one hand, the infrared thermal images are transmitted to an AI analysis system with a built-in AI vision model for heat-sealing process quality inspection. On the other hand, the temperature field distribution data is transmitted to a data interaction module for raw data delivery processing.

4. The online testing system for the heat-sealing performance of large flexible packaging according to claim 2, characterized in that, AI visual inspection component C mmWave It consists of a millimeter-wave detector, an AI analysis system, and a data interaction module; The millimeter-wave detector is used to generate high-resolution images and dielectric constant distribution data of the entire heat-sealed area of ​​large flexible packaging components. On the one hand, the high-resolution images are transmitted to an AI analysis system with a built-in AI vision model for product quality inspection. On the other hand, the dielectric constant distribution data is transmitted to a data interaction module for raw data delivery processing.

5. The online testing system for the heat-sealing performance of large flexible packaging according to any one of claims 2-4, characterized in that, The system also includes a photoelectric sensor, a serial number printing module, a defective product inkjet marking module, and a defective product rejection module; The photoelectric sensor is used to continuously emit and detect the light beam. The physical level signal output by the photoelectric sensor is continuously monitored by the PLC controller. If a large flexible package is detected, the physical level signal output by the photoelectric sensor will change. Based on this, the PLC controller first triggers the serial number printing module to print a unique detection serial number at a designated position on the large flexible package, then triggers the quality inspection program, then triggers the defective product inkjet marking module to print a clear REJECT mark at a designated position on the defective product, and finally triggers the defective product rejection module to remove the defective product from the production line.