Flexible package side edge one-time folding forming method and system using reinforcement learning

Abstract

The invention discloses a flexible package side edge one-time folding forming method and system using reinforcement learning, and relates to the technical field of package forming. The method comprises the steps that multi-layer composite material information and interlayer bonding structure information of a target flexible package material are collected; performing transition folding rail angle learning by utilizing reinforcement learning and a preset interlayer alignment index to generate a transition folding rail angle parameter; after a target flexible packaging material is put into the forming unit, one-time folding forming control is carried out, a state sensor is activated for real-time monitoring, and real-time edge folding state data is generated; adjustable parameter learning in the edge folding process is conducted, and updated transition folding rail angle parameters are generated. The technical problem that in the prior art, due to the fact that the interlayer alignment precision is low in flexible package side edge folding forming, the flexible package quality is low is solved, and the technical effects that high-precision one-time folding forming of the flexible package side edges is achieved, and the flexible package quality is improved are achieved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of packaging forming, in particular to a soft packaging side one-time folding forming method and system using reinforcement learning. BACKGROUND

[0002] In the field of soft packaging production, the folding forming quality of the soft packaging side edge is crucial. In the prior art, the folding forming of the soft packaging side edge has the problems of low interlayer alignment precision and inability to form once, which leads to large residual rebound angle of creases, high interlayer misalignment and peeling probability, affecting the quality and use effect of the soft packaging.

[0003] The prior art has the technical problem of low interlayer alignment precision in the folding forming of the soft packaging side edge, which leads to low soft packaging quality. SUMMARY

[0004] The present application provides a soft packaging side one-time folding forming method and system using reinforcement learning, which is used to solve the technical problem of low interlayer alignment precision in the folding forming of the soft packaging side edge in the prior art, which leads to low soft packaging quality.

[0005] In view of the above problems, the present application provides a soft packaging side one-time folding forming method and system using reinforcement learning.

[0006] In a first aspect of the present application, a soft packaging side one-time folding forming method using reinforcement learning is provided, which comprises:

[0007] Collecting multi-layer composite material information and interlayer adhesion structure information of a target soft packaging material; based on the multi-layer composite material information and the interlayer adhesion structure information, using reinforcement learning to learn a transition folding rail angle with a preset interlayer alignment index, generating a transition folding rail angle parameter; after the target soft packaging material is put into a forming unit, controlling a folding edge device to perform one-time folding forming control with the transition folding rail angle parameter, and activating a state sensor to perform real-time monitoring, generating real-time folding edge state data; learning adjustable parameters of the folding edge process with the real-time folding edge state data, adjusting the transition folding rail angle parameter, and generating an updated transition folding rail angle parameter.

[0008] In a second aspect of the present application, a soft packaging side one-time folding forming system using reinforcement learning is provided, which comprises:

[0009] A material information acquisition module is used to collect multi-layer composite material information and interlayer bonding structure information of the target flexible packaging material; a transition folding track angle learning module is used to use reinforcement learning to perform transition folding track angle learning with preset interlayer alignment indicators based on the multi-layer composite material information and the interlayer bonding structure information, and generate transition folding track angle parameters; a real-time monitoring module is used to control the folding device to perform a folding forming control once after the target flexible packaging material is placed into the forming unit, and activate the status sensor for real-time monitoring to generate real-time folding status data; a parameter adjustment module is used to learn the adjustable parameters of the folding process with the real-time folding status data, adjust the transition folding track angle parameters, and generate updated transition folding track angle parameters.

[0010] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0011] The system collects information about the target flexible packaging material's multilayer composite material and interlayer bonding structure. It then uses reinforcement learning to learn the transition fold angle using preset interlayer alignment indicators, generating transition fold angle parameters. After placing the target flexible packaging material into a forming unit, it performs a single folding control, activating a status sensor for real-time monitoring and generating real-time folding status data. Using this real-time folding status data, it learns the adjustable parameters of the folding process, adjusts the transition fold angle parameters, and generates updated transition fold angle parameters. This system achieves high-precision, single-shot folding of the flexible packaging side, improving the quality of the flexible packaging. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0013] Figure 1 A schematic diagram of a process for a one-time folding method for a flexible packaging side using reinforcement learning provided in an embodiment of the present application;

[0014] Figure 2 Schematic diagram of the structure of a flexible packaging side one-time folding forming system using reinforcement learning provided in an embodiment of the present application.

[0015] Description of the accompanying drawings: material information collection module 10, transition folding angle learning module 20, real-time monitoring module 30, parameter adjustment module 40. DETAILED DESCRIPTION

[0016] This application provides a method and system for folding the side of a flexible package in one step using reinforcement learning, which is used to solve the technical problem in the prior art of low inter-layer alignment accuracy in folding the side of a flexible package, resulting in poor quality of the flexible package.

[0017] The following will be combined with the accompanying drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making any creative work are within the scope of protection of this application.

[0018] Example 1, as Figure 1 As shown, the present application provides a method for forming a flexible package side edge by folding it in one step using reinforcement learning, the method comprising:

[0019] Step S100: Collecting multi-layer composite material information and inter-layer bonding structure information of the target flexible packaging material.

[0020] Specifically, sensor equipment is used to comprehensively collect information about the multi-layer composite material of the target flexible packaging material, covering the material type, thickness specifications, tensile strength, elastic modulus and other physical and mechanical performance parameters of each layer, as well as interlayer bonding structure information, including the material composition of the bonding layer, bonding process (such as hot pressing and adhesive bonding), bonding strength index, and microstructure distribution of the bonding interface. This information will serve as the basic input for subsequent reinforcement learning to establish a mapping relationship between material properties and folding track angles, ensuring that the subsequent generation of transition folding track angle parameters can accurately adapt to the actual characteristics of the material.

[0021] Step S200: Based on the multi-layer composite material information and the inter-layer bonding structure information, reinforcement learning is used to perform transition folding angle learning with a preset inter-layer alignment index to generate transition folding angle parameters.

[0022] Specifically, the target folding angle is first determined and a transition space for the folding track angle is constructed. Information about the multilayer composite material and the interlayer bonding structure is used as retrieval factors. Folding track adaptation evaluation samples are collected according to preset interlayer alignment indicators (including at least the crease residual springback angle, interlayer misalignment, and interlayer debonding probability). Reinforcement learning is then used to train an interlayer folding track adaptation analysis model. When the model is called, the empirical value of the transition folding track angle (the most frequently used angle in the historical records) is used as the starting point. Through transition smoothing optimization, the folding track residual springback angle, interlayer misalignment, and interlayer debonding probability are iteratively decreased. The stopping condition is when the preset minimum transition unit or the iteratively obtained fitness indicator set meets the interlayer alignment indicator threshold. The resulting termination fitness indicator set, termination transition folding track angle parameters, and termination condition identifier are generated. If the termination condition is marked as the fitness index set meeting the threshold, the transition folding track angle parameter is generated using the termination transition folding track angle parameter; if it is a preset minimum transition unit, the termination fitness index set is compared with the threshold, and the satisfied and unsatisfied index items are extracted. According to the material information, the preheating influence relationship corresponding to each indicator under the termination transition folding track angle parameter is constructed, and the preheating parameter is generated through collaborative analysis, and then the transition folding track angle parameter is generated in combination with the termination transition folding track angle parameter.

[0023] Step S300: After the target flexible packaging material is placed into the forming unit, the folding device is controlled to perform a folding forming control using the transition folding track angle parameter, and the status sensor is activated for real-time monitoring to generate real-time folding status data.

[0024] Specifically, after the target flexible packaging material is placed in the forming unit, the folding mechanism is precisely controlled based on the generated transition folding track angle parameters to perform a single folding operation on the side of the flexible packaging. During this process, status sensors such as thickness sensors, infrared thermometers, displacement encoders, and stress sensors are simultaneously activated to collect real-time data on local thickness, temperature, displacement, and tension fluctuations in the folding area. This data, containing multi-dimensional physical quantities, generates real-time folding status data. This provides real-time dynamic monitoring information for subsequent parameter adjustments during the folding process, ensuring precise control and status feedback of the folding process.

[0025] Step S400: using the real-time folding state data to learn the adjustable parameters of the folding process, adjusting the transition folding track angle parameters, and generating updated transition folding track angle parameters.

[0026] Specifically, the transition folding rail angle parameter is first used as a benchmark, and an offset comparison is performed with the real-time folding state data (including local thickness, temperature, displacement and tension fluctuation data of the folding area) to accurately locate the real-time state offset vector. The folding forming knowledge base is then connected, which stores folding forming samples based on the folding state-correction action-correction reward triple structure, where the corrective action is the parameter correction behavior for the folding state, and the correction reward is the corresponding correction effect. With the help of sample data in the knowledge base, the real-time state offset vector is corrected. By analyzing the cause of the offset and matching the corresponding corrective action, the transition folding rail angle parameter is dynamically adjusted, and finally the updated transition folding rail angle parameter is generated to achieve adaptive optimization of the folding process.

[0027] In one possible implementation, step S200 further includes:

[0028] The preset interlayer alignment index at least includes a crease residual springback angle, an interlayer misalignment amount, and an interlayer peeling probability.

[0029] Specifically, the preset interlayer alignment indicators include at least the residual springback angle of the crease, the amount of interlayer misalignment, and the probability of interlayer delamination. These three indicators will be independently evaluated in the subsequent fitness analysis. The residual springback angle measures the angular deviation caused by material rebound at the crease after folding, the interlayer misalignment characterizes the relative displacement between layers during the folding process, and the probability of interlayer delamination assesses the likelihood of interlayer adhesive failure and separation due to the folding operation. By quantifying these three indicators separately, the impact of the transition fold angle parameters on the folding quality of flexible packaging materials can be comprehensively and independently evaluated.

[0030] In one possible implementation, step S200 further includes:

[0031] Step S210: Determine the target folding angle and construct a folding track angle transition space.

[0032] Step S220: using the multi-layer composite material information and the inter-layer bonding structure information as retrieval factors, collecting folding track adaptation evaluation samples based on the preset inter-layer alignment index, and using reinforcement learning to train an inter-layer folding track adaptation analysis model.

[0033] Step S230: calling the inter-layer fold rail adaptation analysis model, screening the transition fold rail angle that meets the inter-layer alignment index threshold based on the fold rail angle transition space, and generating the transition fold rail angle parameter.

[0034] Specifically, based on the design standards and actual process requirements for the flexible packaging side folding process, a target folding angle is determined. This angle represents the ideal angle to be achieved after the flexible packaging side folding is completed. With this target folding angle as the center, a reasonable angle fluctuation range is set, combining material properties and production experience. This creates a transition space for the folding track angle. This provides a specific parameter search range for the subsequent search and optimization of the transition folding track angle using a reinforcement learning algorithm, ensuring that the folding track angle can be adjusted within the range that meets the molding requirements.

[0035] The collected multilayer composite material information of the target flexible packaging material (including physical and mechanical performance parameters such as the material, thickness, and tensile strength of each layer) and interlayer bonding structure information (such as the bonding layer material composition, bonding process, and bonding strength) are used as retrieval factors. Based on preset interlayer alignment indicators (crease residual rebound angle, interlayer misalignment, and interlayer delamination probability), folding track adaptation evaluation samples that match the current material properties are retrieved and collected from historical production data. These samples contain actual test data for each indicator at different folding track angles. The samples are iteratively trained using a reinforcement learning algorithm to construct an interlayer folding track adaptation analysis model. The model can learn the mapping relationship between material properties and folding track angle, thereby possessing the ability to predict the optimal folding track angle parameters based on the input material information.

[0036] First, the most frequently used transition fold angles in historical flexible packaging side-folding records are collected as empirical values. Using these empirical values ​​as a starting point, the interlayer fold angle adaptation analysis model is invoked. Within the fold angle transition space, iterative optimization is performed to reduce the residual springback angle of the crease, the amount of interlayer misalignment, and the probability of interlayer delamination. During the iterative process, the angle parameters are adjusted using a transition smoothing optimization algorithm. The iteration is terminated when the resulting fitness index set meets the preset interlayer alignment threshold or reaches the preset minimum transition unit. The resulting termination fitness index set, termination transition fold angle parameters, and termination condition identifier are generated. If the termination condition is identified as the fitness index set meeting the threshold, the target transition folding track angle parameter is directly generated using the termination transition folding track angle parameter; if the termination condition is identified as the preset minimum transition unit, the termination fitness index set is compared with the interlayer alignment index threshold, and the satisfied and unsatisfied index items are extracted. According to the multi-layer composite material information and the interlayer bonding structure information, the preheating influence relationship corresponding to each index under the termination transition folding track angle parameter is constructed. The first preheating temperature range of the unsatisfied index item is determined through collaborative analysis, and the temperature value that makes the index influence meet the preset threshold is selected as the preheating parameter in combination with the preheating influence relationship of the satisfied index item. Finally, the transition folding track angle parameter is generated using the preheating parameter and the termination transition folding track angle parameter.

[0037] In one possible implementation, step S230 further includes:

[0038] Step S231: collecting empirical values ​​of transition folding track angles, wherein the empirical values ​​of transition folding track angles are the most frequently used transition folding track angles in historical records of flexible package side one-time folding.

[0039] Step S232: calling the inter-layer fold rail adaptation analysis model, taking the empirical value of the transition fold rail angle as the starting point, performing descending iterations of the crease residual springback angle, inter-layer misalignment amount and inter-layer peeling probability through transition smoothing optimization, and using the preset minimum transition unit or the fitness index set obtained by iteration to meet the inter-layer alignment index threshold as the iteration stop condition, generating the termination fitness index set and termination transition fold rail angle parameters and the termination condition identifier when the iteration stops.

[0040] Step S233: If the termination condition is identified as the fitness index set obtained by iteration meets the inter-layer alignment index threshold, the transition bend angle parameter is generated using the termination transition bend angle parameter.

[0041] Step S234: If the termination condition is identified as a preset minimum transition unit, a preheating parameter analysis is performed during folding based on the termination fitness index set and the termination transition folding angle parameter, and the transition folding angle parameter is generated in combination with the preheating parameter.

[0042] Specifically, by statistically analyzing historical production data from single-folding of the side of flexible packaging, the most frequently used transition fold angles in past production practices were extracted and determined as empirical values ​​for the transition fold angles. This empirical value, a representative angle parameter selected from a large number of historical molding cases, reflects the commonly used angle settings for folding similar flexible packaging materials. This provides an initial reference for subsequent optimization of the transition fold angle, reducing the search space for algorithm iterations and improving optimization efficiency.

[0043] The trained inter-layer folding track adaptation analysis model is called, and the empirical value of the transition folding track angle is used as the starting point for iterative optimization. The iteration is performed in the folding track angle transition space through the transition smoothing optimization algorithm. During the iteration process, the residual rebound angle of the crease, the amount of inter-layer misalignment and the probability of inter-layer peeling are continuously optimized to gradually reduce the values ​​of various indicators. When the preset minimum transition unit (i.e., the minimum step size of the angle adjustment) is reached, or the fitness indicator set obtained by the iteration meets the preset inter-layer alignment indicator threshold, the iteration is stopped, and the termination fitness indicator set, the termination transition folding track angle parameter, and the termination condition identifier used to identify the reason for the iteration stop are generated.

[0044] When the termination condition indicates that the fitness index set obtained by iteration has met the preset interlayer alignment index threshold (that is, the residual rebound angle of the crease, the amount of interlayer misalignment and the probability of interlayer peeling all meet the standards required by the process), it means that the termination transition folding track angle parameter obtained through iterative optimization can ensure the quality of the folding molding of the side of the flexible packaging. At this time, the termination transition folding track angle parameter is directly output as the final transition folding track angle parameter to control the subsequent folding device to perform the folding molding operation.

[0045] If the termination condition is identified as the preset minimum transition unit, it indicates that the iterative optimization has reached the minimum step size for angle adjustment but still has not yet made the fitness index set meet the interlayer alignment index threshold. At this point, a fold preheat parameter analysis is carried out based on the termination fitness index set and the termination transition fold angle parameter: first, the termination fitness index set is compared with the index threshold to distinguish between satisfied and unsatisfied index items. Then, based on the multilayer composite material information and the interlayer bonding structure information, a preheat influence relationship model corresponding to the crease residual springback angle, interlayer misalignment, and interlayer delamination probability under the termination angle parameter is constructed. This model uses a collaborative analysis of satisfied and unsatisfied index items, determines the first preheating temperature range corresponding to the unsatisfied index items, and combines the preheat influence relationship of the satisfied index items to select the temperature value within this temperature range that makes the index influence change meet the preset threshold, thus generating the preheating parameters. Finally, the preheating parameters are combined with the termination transition fold angle parameter to generate the transition fold angle parameter that comprehensively considers the material preheating effect, thereby compensating for the index deviation caused by the limited angle iteration and ensuring the folding forming quality.

[0046] In one possible implementation, step S234 further includes:

[0047] Step S2341: Compare the termination fitness indicator set with the inter-layer alignment indicator threshold to extract indicator-satisfied items and indicator-unsatisfied items.

[0048] Step S2342: constructing corresponding preheating influence relationships of the crease residual springback angle, interlayer misalignment and interlayer peeling probability under the termination transition folding track angle parameter according to the multilayer composite material information and the interlayer bonding structure information.

[0049] Step S2343: Based on the preheating influence relationship, perform collaborative analysis of the index items that meet the index items and the index items that do not meet the index items to generate the preheating parameters.

[0050] Step S2344: generating the transition folding track angle parameter using the preheating parameter and the termination transition folding track angle parameter.

[0051] Specifically, the termination fitness index set obtained when the iteration stops (including the specific numerical values ​​of the residual rebound angle of the crease, the amount of interlayer misalignment and the probability of interlayer peeling) is compared one by one with the preset interlayer alignment index threshold. Through numerical comparison, the satisfied index items that meet or exceed the threshold requirements and the unsatisfied index items that do not meet the threshold requirements are accurately identified, so as to clarify the indicator direction that needs to be optimized in a targeted manner in the future, and provide an accurate indicator screening basis for the subsequent construction of preheating influence relationships and collaborative analysis based on material properties.

[0052] Based on the collected information of multilayer composite materials (covering thermal physical properties such as thermal expansion coefficient, elastic modulus, thickness of each layer material) and interlayer bonding structure information (including the heat-resistant temperature range of the bonding layer material, the bonding strength-temperature variation curve, etc.), for the termination transition folding track angle parameter, the corresponding influence relationship between the crease residual rebound angle, interlayer dislocation and interlayer peeling probability and the preheating temperature is constructed. By analyzing the changes in the mechanical properties of the material at different preheating temperatures, a mathematical model is established to quantitatively characterize the influence of preheating temperature on various indicators. For example, the change in the crease residual rebound angle, the fluctuation range of the interlayer dislocation and the decrease in the interlayer peeling probability when the preheating temperature increases by 1°C are determined, providing a quantitative basis for the subsequent collaborative analysis of preheating parameters.

[0053] Based on the constructed preheating influence relationship, a collaborative analysis is carried out on the indicators that meet the requirements and the indicators that do not meet the requirements. First, based on the preheating influence relationship corresponding to the indicator that does not meet the requirements (such as the probability of interlayer peeling exceeds the threshold), the first preheating temperature range that can optimize its indicator value to within the threshold is determined. For example, by analyzing the negative correlation between the preheating temperature and the probability of interlayer peeling, it is concluded that the temperature needs to be increased to 60-70°C to reduce the risk of peeling. At the same time, combined with the preheating influence relationship of the indicator that meets the requirements (such as the residual rebound angle of the crease has reached the standard), the indicator changes of the indicator that meets the requirements within the temperature range are evaluated. For example, it is found that the crease rebound angle will increase by 0.5° due to material softening at 60-70°C. It is necessary to ensure that the increase does not exceed the preset threshold of 0.8°. Finally, within the temperature range, the temperature value (such as 65°C) that can optimize the unsatisfied indicator to a qualified level and make the satisfied indicator change within the allowable range is screened out, and the preheating parameters containing specific temperature parameters are generated to achieve a collaborative balance of indicator optimization.

[0054] By establishing a mapping model between preheating parameters and changes in material mechanical properties, parameters such as preheating temperature and preheating time are converted into changes in material elastic modulus and yield strength. Based on the relationship between changes in material mechanical properties and the folding angle, a compensation value for the final transition folding angle parameter is calculated. For example, when the preheating temperature is 65°C, the material elastic modulus decreases by 10%. According to calculations, the final transition folding angle parameter needs to be adjusted from 30° to 30.5°, resulting in a final transition folding angle parameter of 30.5°. This achieves the fusion of preheating parameters and final transition folding angle parameters, generating a transition folding angle parameter that adapts to the material's preheating state.

[0055] In one possible implementation, step S2343 further includes:

[0056] Step S23431: According to the preheating influence relationships corresponding to the residual springback angle of the crease, the interlayer misalignment amount and the interlayer peeling probability, perform index optimization of the unsatisfied index items and determine the first preheating temperature range that does not meet the index items.

[0057] Step S23432: combining the preheating influence relationship of the index items, determining the index influence update of the index items under the first preheating temperature range, screening the temperature value of the index influence that meets the preset threshold, and generating the preheating parameters.

[0058] Specifically, by establishing a mathematical model to quantitatively analyze the relationship between the residual rebound angle of the crease, the amount of interlayer misalignment and the probability of interlayer peeling and the preheating temperature, historical data are used to fit the curve equations of each indicator changing with temperature. For items that do not meet the indicators, such as the probability of interlayer peeling exceeds the threshold, the corresponding preheating influence relationship equation is solved together with the indicator threshold to calculate the lower and upper limits of the temperature range that can make the indicator meet the standard. At the same time, combined with the heat resistance limit of the material and the feasibility of the process, the temperature areas that cause material degradation or excessive energy consumption are excluded, and finally the first preheating temperature range that does not meet the indicator is determined. For example, it is calculated that when the preheating temperature is between 65°C and 75°C, the probability of interlayer peeling can be reduced from 0.25 to below 0.1, and the thermal deformation of the material is within the allowable range.

[0059] After obtaining the first preheating temperature range that does not meet the index item, the preheating influence relationship data corresponding to the index item that meets the index item is retrieved (such as the function curve of the residual springback angle of the crease changing with temperature), and the temperature range is substituted into the relationship model to calculate the index change. For example, if the first preheating temperature range is 60-70°C, the model calculation shows that the residual springback angle of the crease increases with the increase of temperature within this range, and the angle increases by 0.05° for every 1°C increase. The preset threshold is that the angle increment does not exceed 0.5°, then the temperature upper limit can be determined to be 60+(0.5 / 0.05)=70°C. At the same time, the stability of the indicators of the temperature boundary points is verified in combination with the historical molding data, and the temperature values ​​such as 65°C that make the change in the index item (0.25°) far less than the threshold and do not meet the optimization standard of the index item are screened out, and finally a preheating scheme containing precise temperature parameters is generated.

[0060] In one possible implementation, step S300 further includes:

[0061] Step S310: The state sensors include a thickness sensor, an infrared thermometer, a displacement encoder, and a stress sensor; the real-time folding state data includes local thickness, temperature, displacement, and tension fluctuation data of the folding area.

[0062] Specifically, the state sensors include a thickness sensor, an infrared thermometer, a displacement encoder, and a stress sensor. The thickness sensor is used to measure local thickness changes in the folding area in real time, the infrared thermometer monitors the temperature distribution in the folding area, the displacement encoder records the displacement during the folding process, and the stress sensor detects tension fluctuations in the folding area. These sensors work together to generate real-time folding state data covering the local thickness, temperature, displacement, and tension fluctuations in the folding area. This provides comprehensive and accurate data support for learning the adjustable parameters of the folding process and adjusting the transition folding track angle parameters, ensuring real-time monitoring of the folding state and precise control.

[0063] In one possible implementation, step S400 further includes:

[0064] Step S410: performing an offset comparison on the real-time folding state data using the transition folding track angle parameter to locate the real-time state offset vector.

[0065] Step S420: connecting to the hem forming knowledge base, performing correction of the real-time state offset vector, and generating the updated transition folding track angle parameter.

[0066] Specifically, the transition folding track angle parameter is used as the standard reference value, and compared with the folding state data such as local thickness, temperature, displacement and tension fluctuation of the folding area collected in real time by the state sensor, and the deviation between the actual data and the reference value is calculated, so as to locate the real-time state offset vector that can reflect the degree and direction of the current folding state deviation from the expected state, providing an accurate deviation basis for the subsequent parameter adjustment of the folding process.

[0067] The system connects to the hemming knowledge base through an API interface. This knowledge base uses a triple structure (hemming state, corrective action, and corrective reward) to store historical forming samples. The hemming state includes multidimensional feature vectors such as local thickness deviation and temperature anomaly range. After obtaining the real-time state offset vector, the cosine similarity algorithm is used to search the knowledge base for the historical hemming state with the highest matching degree. For example, if the current offset vector is thickness +0.2mm, temperature -5°C, and displacement +1.5mm, it will match the record with similar offset features in the historical sample. The corresponding corrective actions are extracted (such as increasing the folding rail angle by 0.5° and the preheating temperature by 10°C), and the optimal corrective strategy is determined through the Q-value evaluation mechanism of reinforcement learning and combined with the correction reward (such as increasing the indicator achievement rate by 20%). Finally, the real-time offset vector and the corrective action are linearly superimposed and calculated. The located real-time state offset vector (such as the deviation of data such as local thickness, temperature, displacement and tension fluctuation in the folding area) and the corresponding corrective action retrieved from the folding forming knowledge base (such as adjusting the folding rail angle, changing preheating temperature and other parameters) are calculated through the mathematical method of linear superposition. For example, if the real-time state offset vector shows that the temperature in the folding area is 5°C lower than expected, the corresponding corrective action is to increase the preheating temperature by 10°C and increase the folding rail angle by 0.5°. These offsets and corrections are then superimposed according to a linear relationship to generate the updated transition folding rail angle parameters to achieve precise control of the folding process.

[0068] In one possible implementation, step S420 further includes:

[0069] Step S421: The hem-folding knowledge base stores hem-folding samples based on a triple structure, where the triple structure includes a hem-folding state, a correction action, and a correction reward.

[0070] Step S422: The correction action is a parameter correction behavior performed on the folding state, and the correction reward is the corresponding correction effect.

[0071] Specifically, the hemming forming knowledge base stores hemming forming samples organized in a triple structure, where the hemming state is a comprehensive representation of real-time data such as local thickness, temperature, displacement and tension fluctuations in the hemming area during the hemming process, forming a multi-dimensional feature vector; the corrective action refers to the parameter adjustment measures taken for specific hemming state deviations, such as modifying the transition folding track angle, adjusting the preheating temperature, etc.; the corrective reward is an indicator for quantitatively evaluating the effectiveness of the corrective action, such as the reduction ratio of inter-layer misalignment, the reduction value of the residual rebound angle of the crease, etc. Through this structured storage method, data support is provided for the offset correction of the real-time hemming state.

[0072] Corrective action refers to the parameter adjustment measures taken for the folding state when it deviates from the expected state (such as abnormal local thickness in the folding area, temperature fluctuation, displacement deviation or tension imbalance, etc.), such as increasing the transition folding angle by 0.5°, raising the preheating temperature by 10°C or adjusting the tension control parameters. Corrective reward is a quantitative evaluation of the effectiveness of the corrective action. By calculating the reduction ratio of the interlayer misalignment after correction, the reduction of the residual rebound angle of the crease or the decrease in the probability of interlayer peeling, etc., the actual effect of the corrective action on improving the quality of folding forming is measured. For example, a certain correction reduces the interlayer misalignment from 0.3mm to 0.1mm, and this difference is the corresponding corrective reward.

[0073] Example 2, based on the same inventive concept as the flexible packaging side folding forming method using reinforcement learning in the previous embodiment, Figure 2 As shown, the present application provides a flexible packaging side folding forming system using reinforcement learning. The system and method embodiments in the present application are based on the same inventive concept. The system includes:

[0074] The material information acquisition module 10 is used to acquire the multi-layer composite material information and inter-layer bonding structure information of the target flexible packaging material.

[0075] The transition folding track angle learning module 20 is used to perform transition folding track angle learning based on the multi-layer composite material information and the interlayer bonding structure information using reinforcement learning with a preset interlayer alignment index to generate a transition folding track angle parameter.

[0076] The real-time monitoring module 30 is used to control the folding device to perform a folding forming control according to the transition folding track angle parameter after the target flexible packaging material is placed into the forming unit, and activate the status sensor for real-time monitoring to generate real-time folding status data.

[0077] The parameter adjustment module 40 is used to learn the adjustable parameters of the folding process based on the real-time folding state data, adjust the transition folding track angle parameters, and generate updated transition folding track angle parameters.

[0078] Furthermore, the system is also used to implement the following functions:

[0079] The preset interlayer alignment index at least includes a crease residual springback angle, an interlayer misalignment amount, and an interlayer peeling probability.

[0080] Furthermore, the system is also used to implement the following functions:

[0081] The target folding angle is determined and a folding track angle transition space is constructed; the multi-layer composite material information and the interlayer bonding structure information are used as retrieval factors, folding track adaptation evaluation samples are collected based on the preset interlayer alignment index, and an interlayer folding track adaptation analysis model is trained using reinforcement learning; the interlayer folding track adaptation analysis model is called, and based on the folding track angle transition space, a transition folding track angle that meets the interlayer alignment index threshold is screened to generate the transition folding track angle parameters.

[0082] Furthermore, the system is also used to implement the following functions:

[0083] Collect the empirical value of the transition folding track angle, which is the most frequently used transition folding track angle in the historical flexible packaging side one-time folding forming record; call the inter-layer folding track adaptation analysis model, take the empirical value of the transition folding track angle as the starting point, and through transition smoothing optimization, perform descending iterations of the crease residual rebound angle, inter-layer misalignment amount and inter-layer peeling probability, and use the preset minimum transition unit or the fitness index set obtained by iteration to meet the inter-layer alignment index threshold as the iteration stop condition, generate the termination fitness index set and termination transition folding track angle parameter and termination condition identifier when the iteration stops; if the termination condition identifier is that the fitness index set obtained by iteration meets the inter-layer alignment index threshold, generate the transition folding track angle parameter with the termination transition folding track angle parameter; if the termination condition identifier is the preset minimum transition unit, perform preheating parameter analysis during folding based on the termination fitness index set and the termination transition folding track angle parameter, and generate the transition folding track angle parameter in combination with the preheating parameter.

[0084] Furthermore, the system is also used to implement the following functions:

[0085] The termination fitness index set is compared with the interlayer alignment index threshold to extract the index items that meet the index and the index items that do not meet the index; based on the multi-layer composite material information and the interlayer bonding structure information, the corresponding preheating influence relationships of the crease residual rebound angle, interlayer misalignment and interlayer peeling probability under the termination transition fold track angle parameters are constructed; based on the preheating influence relationship, a collaborative analysis of the index items that meet the index and the index items that do not meet the index is performed to generate the preheating parameters; the transition fold track angle parameters are generated based on the preheating parameters and the termination transition fold track angle parameters.

[0086] Furthermore, the system is also used to implement the following functions:

[0087] According to the preheating influence relationships corresponding to the residual rebound angle of the crease, the amount of interlayer misalignment and the probability of interlayer peeling, the indicator optimization of the unsatisfied indicator items is performed to determine the first preheating temperature range that does not meet the indicator items; combined with the preheating influence relationship that meets the indicator items, the indicator influence update that meets the indicator items under the first preheating temperature range is determined, the temperature value of the indicator influence that meets the preset threshold is screened, and the preheating parameters are generated.

[0088] Furthermore, the system is also used to implement the following functions:

[0089] The state sensors include a thickness sensor, an infrared thermometer, a displacement encoder and a stress sensor; the real-time folding state data includes local thickness, temperature, displacement and tension fluctuation data of the folding area.

[0090] Furthermore, the system is also used to implement the following functions:

[0091] The real-time folding state data is offset compared with the transition folding track angle parameter to locate the real-time state offset vector; the folding forming knowledge base is connected to perform correction of the real-time state offset vector to generate the updated transition folding track angle parameter.

[0092] Furthermore, the system is also used to implement the following functions:

[0093] The hemming forming knowledge base stores hemming forming samples based on a triple structure, wherein the triple structure includes a hemming state, a correction action, and a correction reward; the correction action is a parameter correction behavior performed on the hemming state, and the correction reward is the corresponding correction effect.

[0094] It should be noted that the order in which the embodiments of the present application are presented is for illustrative purposes only and does not necessarily represent the superiority or inferiority of the embodiments. Furthermore, the foregoing descriptions of specific embodiments of this specification are provided. Furthermore, the processes depicted in the accompanying drawings do not necessarily require the specific order or sequential sequence shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0095] The above description is only a preferred embodiment of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should be included in the scope of protection of the present application.

[0096] This specification and drawings are merely illustrative of the present application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Obviously, those skilled in the art may make various modifications and variations to this application without departing from the scope of this application. Thus, this application is intended to include such modifications and variations as fall within the scope of this application and its equivalents.

Claims

1. A one-time folding forming method for flexible packaging side using reinforcement learning, characterized in that: include: Collect information about the multi-layer composite material and interlayer bonding structure of the target flexible packaging material; Based on the multi-layer composite material information and the interlayer bonding structure information, using reinforcement learning with a preset interlayer alignment index to perform transition folding track angle learning to generate transition folding track angle parameters; After the target flexible packaging material is placed in the forming unit, the folding device is controlled to perform a folding forming control according to the transition folding track angle parameter, and the state sensor is activated for real-time monitoring to generate real-time folding state data; The adjustable parameters of the folding process are learned using the real-time folding state data, and the transition folding track angle parameters are adjusted to generate updated transition folding track angle parameters.

2. The method for forming a flexible package side edge by one-time folding using reinforcement learning according to claim 1, characterized in that: The preset interlayer alignment index at least includes a crease residual springback angle, an interlayer misalignment amount, and an interlayer peeling probability.

3. The method for forming a flexible package side edge by one-time folding using reinforcement learning according to claim 2, characterized in that: Based on the multi-layer composite material information and the interlayer bonding structure information, reinforcement learning is used to perform transition folding track angle learning with a preset interlayer alignment index to generate transition folding track angle parameters, including: Determine the target folding angle and construct the transition space of the folding track angle; Using the multi-layer composite material information and the interlayer bonding structure information as retrieval factors, collecting folding track adaptation evaluation samples based on the preset interlayer alignment index, and using reinforcement learning to train an interlayer folding track adaptation analysis model; The inter-layer fold rail adaptation analysis model is called, and a transition fold rail angle that meets an inter-layer alignment index threshold is screened based on the fold rail angle transition space to generate the transition fold rail angle parameter.

4. The method for forming a flexible package side edge by one-time folding using reinforcement learning according to claim 3, characterized in that: The inter-layer fold rail adaptation analysis model is called, and based on the fold rail angle transition space, a transition fold rail angle that meets the inter-layer alignment index threshold is screened to generate the transition fold rail angle parameter, including: Collecting empirical values ​​of transition folding track angles, wherein the empirical values ​​of transition folding track angles are the most frequently used transition folding track angles in the history of flexible packaging side one-time folding molding; The interlayer folding track adaptation analysis model is called, and starting from the empirical value of the transition folding track angle, through transition smoothing optimization, descending iterations of the crease residual springback angle, interlayer misalignment amount, and interlayer peeling probability are performed, and the iteration stopping condition is that a preset minimum transition unit or the fitness index set obtained by iteration meets the interlayer alignment index threshold, and a termination fitness index set and a termination transition folding track angle parameter and a termination condition identifier are generated when the iteration stops; If the termination condition is that the fitness index set obtained by iteration satisfies the inter-layer alignment index threshold, the transition bend angle parameter is generated using the termination transition bend angle parameter; If the termination condition is identified as a preset minimum transition unit, a preheating parameter analysis is performed during folding based on the termination fitness index set and the termination transition folding track angle parameter, and the transition folding track angle parameter is generated in combination with the preheating parameter.

5. The method for forming a flexible package side edge by one-time folding using reinforcement learning as claimed in claim 4, characterized in that: Performing a preheating parameter analysis during folding based on the termination fitness index set and the termination transition folding track angle parameter, and generating the transition folding track angle parameter in combination with the preheating parameter, includes: Comparing the termination fitness index set with the inter-layer alignment index threshold, and extracting index-satisfying items and index-unsatisfying items; According to the multi-layer composite material information and the interlayer bonding structure information, the corresponding preheating influence relationships of the crease residual springback angle, the interlayer misalignment amount and the interlayer peeling probability under the termination transition folding track angle parameter are constructed; Based on the preheating influence relationship, a collaborative analysis of the index items that meet the index and the index items that do not meet the index is performed to generate the preheating parameters; The transition folding track angle parameter is generated by using the preheating parameter and the termination transition folding track angle parameter.

6. The method for forming a flexible package side edge by one-time folding using reinforcement learning according to claim 5, characterized in that: Based on the preheating impact relationship, a collaborative analysis of the index items that meet the index items and the index items that do not meet the index items is performed to generate the preheating parameters, including: According to the preheating influence relationships corresponding to the residual springback angle of the crease, the interlayer dislocation amount, and the interlayer peeling probability, respectively, the index optimization of the unsatisfied index items is performed to determine the first preheating temperature range of the unsatisfied index items; In combination with the preheating influence relationship of the index items, the index influence update of the index items under the first preheating temperature range is determined, and the temperature value of the index influence meeting the preset threshold is screened to generate the preheating parameter.

7. The method for forming a flexible package side edge by one-time folding using reinforcement learning according to claim 1, characterized in that: The state sensors include a thickness sensor, an infrared thermometer, a displacement encoder and a stress sensor; the real-time folding state data includes local thickness, temperature, displacement and tension fluctuation data of the folding area.

8. The method for forming a flexible package side edge by one-time folding using reinforcement learning according to claim 7, characterized in that: The adjustable parameters of the folding process are learned using the real-time folding state data, the transition folding rail angle parameters are adjusted, and the updated transition folding rail angle parameters are generated, including: Performing an offset comparison on the real-time folding state data using the transition folding track angle parameter to locate a real-time state offset vector; Connecting to the hem forming knowledge base, performing the correction of the real-time state offset vector, and generating the updated transition folding track angle parameter.

9. The method for forming a flexible package side edge by one-time folding using reinforcement learning according to claim 8, characterized in that: The hemming forming knowledge base stores hemming forming samples based on a triple structure, where the triple structure includes hemming state, correction action, and correction reward; The correction action is a parameter correction behavior performed on the folding state, and the correction reward is the corresponding correction effect.

10. A flexible packaging side one-time folding forming system using reinforcement learning, characterized in that: The system is used to implement the flexible package side one-time folding forming method using reinforcement learning according to any one of claims 1 to 9, and the system comprises: A material information acquisition module, used to collect information about the multi-layer composite material and the interlayer bonding structure of the target flexible packaging material; a transition folding track angle learning module, configured to perform transition folding track angle learning based on the multi-layer composite material information and the interlayer bonding structure information using reinforcement learning with a preset interlayer alignment index to generate a transition folding track angle parameter; A real-time monitoring module is used to control the folding device to perform a single folding and forming control based on the transition folding track angle parameter after the target flexible packaging material is placed in the forming unit, and activate the status sensor for real-time monitoring to generate real-time folding status data; The parameter adjustment module is used to learn the adjustable parameters of the folding process based on the real-time folding state data, adjust the transition folding rail angle parameters, and generate updated transition folding rail angle parameters.

Images