A product packaging production line-oriented digital twin simulation and optimization system

Abstract

The application discloses a product packaging production line-oriented digital twin simulation and optimization system, and particularly relates to the field of digital twins. The system divides the production line into a plurality of functionally clear sections through a sub-region function end division module, acquires and stores stable and real-time variable parameters through a collection module, constructs a reference model of each sub-region function end based on static parameters, and modifies the reference model by incorporating multi-dimensional variable interference into a real-time production state model construction module. An intelligent algorithm is used to compare parameters in real time. Secondary data acquisition is performed based on repaired attention state data according to the sub-region function end yield and parameter comparison results, a real-time production state model is established, and secondary comparison is performed with a digital twin ideal model. When the difference interval of the attention state data meets the theoretical difference, the repair is completed. If the difference interval does not meet the theoretical difference, a digital output is performed to a user end for manual repair.

Description

[0001] The application relates to the technical field of digital twinning, and more particularly to a digital twinning simulation and optimization system. BACKGROUND

[0002] Digital twinning is a virtual mirror of a physical entity, which is a dynamic digital model corresponding to a real object or system through data driving. The core lies in the real-time bidirectional mapping between the physical world and the digital space. The core working mechanism lies in (1) dynamic mapping, which synchronizes the sensor data of the physical entity, such as temperature, pressure and position, to the virtual model continuously, so as to ensure that the states of the two are consistent. The mapping relationship is refined layer by layer from the geometric level, the physical level, the behavior level to the rule level, and the model is established. (2) Simulation and deduction, based on the current state input, the future behavior is predicted through a physical engine or an AI algorithm, and the feasibility of the scheme is verified quickly by modifying the virtual model parameters, such as increasing the rotating speed or replacing the material, and the physical trial and error cost is avoided. (3) Closed-loop optimization, the deviation between the ideal model and the actual operation is compared, the fault root cause is located, the optimization strategy is generated at the virtual end, such as adjusting the feeding speed or scheduling the maintenance work order. The instruction is issued to the physical end control system for automatic execution, forming a 'perception-analysis-decision-execution' closed loop.

[0003] However, when it is actually used, there are still some disadvantages: (1) the modeling level has an idealization tendency, because there is a lack of reliable parameter basis verified by practice, the parameter range needs to be explored from zero, the trial and error direction is chaotic, and the monitoring equipment is often not calibrated, so that the measurement data has systematic deviation or distortion. With the inaccurate data modeling, the ideal operation interval calculated will deviate from the true range of parameters, so that the model cannot accurately identify the abnormality, causing false alarm or missed alarm, and finally causing the process monitoring and decision based on the model to fail. (2) From the modeling process, a large amount of tedious work is involved in the comprehensive collection and accurate transformation of multi-dimensional information. The information missing or error in each link may cause deviation between the model and the actual production scene, which must be checked repeatedly and adjusted many times. (3) When the production sequence is adjusted based on the actual production result, a unified adjustment mode of the whole production line is often adopted, without differentiated optimization according to the capacity difference of different processes, which is easy to cause the production redundancy of part of the processes, that is, the output exceeds the actual demand. SUMMARY

[0004] A digital twinning simulation and optimization system for a product packaging production line comprises a sub-regional functional end division module: used for dividing the whole production line into a plurality of sub-regional functional ends according to the process flow of the product packaging production line.

[0005] Parameter acquisition module: for collecting the static parameters required for establishing the digital twin ideal model of the function end of any sub-region, marking the parameters corresponding to the static parameters in the real-time production process as dynamic parameters, and transmitting and storing the static parameter information and dynamic parameters to the system data storage unit;

[0006] Model construction module: based on the static parameter information, the digital twin ideal model of the function end of the sub-region is constructed, and based on the dynamic parameter information, the real-time production state model of the function end of the sub-region is constructed;

[0007] Real-time intelligent comparison module: based on the digital twin ideal model, the ideal running interval is set, and the parameters of the real-time production state model that do not fall within the ideal running interval are recorded as imbalance data;

[0008] Comparison result output module: matching imbalance keywords for imbalance data, outputting corresponding repair instructions according to imbalance keywords, and completing repair, the data of which repair instructions are given are recorded as attention state data, and the attention state data is continuously monitored;

[0009] Comparison result detection module: based on the attention state data after repair, secondary data acquisition is carried out, a new real-time production state model is established, and secondary comparison is carried out with the digital twin ideal model, when the difference interval of the attention state data meets the theoretical difference, the repair is completed, if it does not meet the theoretical difference, the number is output to the user end for manual repair.

[0010] Technical effects and advantages of the present application:

[0011] 1. The modular design of the function end of the sub-region divides the whole line into independent sub-region function ends with clear boundaries according to process logic, independent parameter acquisition and optimization target, retains the information flow logic between segments, avoids global model overload, realizes local optimization of local problems, and reduces the scope of fault troubleshooting.

[0012] 2. Dynamic and static parameter separation acquisition: static parameters as reference anchor points remain stable for a long time and are not affected by short-term working condition fluctuations, ensuring the stability of the reference model. Dynamic parameters update their values quickly with the adjustment of working conditions, timely reflect the current state of the system, and provide accurate real-time basis for production decision-making, realizing flexible regulation and control of the production process.

[0013] 3. Every interval without material running to calculate the drift rate of the monitoring equipment, combined with the drift rate and the actual production model to evaluate the degree of equipment wear and generate the current equipment dynamic parameters, when calculating the actual parameter deviation, combined with this table can specifically exclude the interference caused by the change of equipment performance, make the deviation result more close to the real situation of production, provide more reliable basis for production adjustment and quality control.

[0014] 4. By adjusting parameters based on historical high pass rate models, reliable parameter bases that have been verified in practice can be directly reused, greatly reducing the waste of resources and the risk of non-compliance caused by blind trial and error. It can quickly focus on key fine-tuning directions, significantly shortening the parameter optimization cycle while ensuring the stability of results. It is especially suitable for mass production or experimental scenarios with high requirements for both efficiency and quality. Attached Figure Description

[0015] Figure 1 This is a system structure framework diagram of the present invention.

[0016] Figure 2 This is a schematic diagram of the overall process of the present invention. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] like Figure 1 The system shown is a digital twin simulation and optimization system for product packaging production lines, including a sub-region functional division module, a parameter acquisition module, a model building module, a real-time intelligent comparison module, a comparison result output module, and a comparison result detection module.

[0019] like Figure 2 The digital twin simulation and optimization system for product packaging production lines shown divides the functional areas into sub-regions, collects static parameters to construct an ideal digital twin model, and calculates the drift rate and relative deviation of dynamic parameters of the monitoring equipment by comparing the actual production status with the ideal digital twin model. Based on the actual production status and production plan, the speed of the functional areas in the sub-regions is adjusted and optimized.

[0020] The sub-region function end division module is used to divide the whole production line into a plurality of sub-region function ends with clear function boundaries and operation targets according to the process flow of the product packaging production line, and each sub-region function end is used to implement a specific function. The sub-region function end not only maintains the continuity of the process, but also has the pertinence of parameter collection. The sub-region function end division information is stored in the sub-region function end division unit. These sub-region function ends maintain the continuity of the process while taking into account the pertinence of parameter collection. The continuity of the process ensures that each sub-region function end is connected in order according to the production process, avoiding process interruption or repetition. The pertinence of parameter collection means that each sub-region function end can focus on the collection of its core process parameters, ensuring that the entire digital twin system can achieve fine management and optimization of the production line based on clear function division. The stored data is saved in a field format to ensure that the division results are traceable and callable.

[0021] The parameter collection module is used to collect static parameters required for any sub-region function end to establish a digital twin ideal model. The parameters corresponding to the static parameters in the real-time production process are marked as dynamic parameters, and the static parameter information and dynamic parameters are transmitted and stored in the system data storage unit. The static parameter is a reference value that is preset as a constant value in the model establishment process of the digital twin model and does not change with any variable in the model. The dynamic parameter is a deviation value obtained by real-time monitoring after the static parameter is included in the variable of the influencing parameter of the model, and it corresponds to the static parameter one by one.

[0022] The model construction module is used for reading the static parameters in the system data storage unit. The digital twin ideal model is based on a three-dimensional modeling engine and establishes a reference model for product packaging production line. The core is to map the geometry topology, ideal production line operation mode and conflict-free logic rules in a 1:1 ratio. The digital twin ideal model is established based on the mold wear degree, sub-region function end temperature and sub-region function end product flow coordination value which affect the product qualification rate. The mold wear causes the mold size, shape and surface precision to deviate from the design standard, thereby causing product size out-of-tolerance, appearance defects, functional failure and other problems, ultimately directly reducing the qualification rate. The sub-region function end temperature directly affects the key attributes of the material during production. Insufficient melting temperature will cause poor melt flowability, and excessive heat treatment temperature will cause material brittleness, ultimately causing product performance or forming defects. If the speed of the previous sub-region function end is faster than that of the subsequent sub-region function end, the material will accumulate in the middle link, which may cause physical deformation or damage of the material, and is directly judged as unqualified. If the speed of the previous sub-region function end is slower than that of the subsequent sub-region function end, the subsequent sub-region function end will be in intermittent idle due to lack of material, and incomplete processing is easy to occur when restarting processing.

[0023] The method for establishing a digital twin ideal model is as follows:

[0024] S1: Digital twin model mold wear degree calibration: select a mold that meets the production standard, use a contact roughness meter, select a probe that matches the hardness of the mold material, read the surface roughness, maximum surface height, and minimum surface height of the new production mold according to the measurement results, read the mold surface hardness from the equipment manual, calculate the standard wear degree of the mold in the digital twin ideal model according to the formula, and take the wear degree of the production mold at this time as the static parameter. According to the core wear degree formula, the wear degree of the mold is quantitatively calculated, wherein Ra is the surface roughness, the unit is μm or nm, which needs to be coordinated with the hardness unit; rp is the maximum surface height, the unit is μm or nm, the positive value from the average reference line; rv is the minimum surface height, the unit is μm or nm, the positive value from the average reference line, indicating the depth; H is the surface hardness, the unit is MPa or HV, which needs to be coordinated with the length unit.

[0025] S2: Digital twin ideal model sub-region function end temperature: according to the process flow of the product packaging production line, set the most suitable temperature for each sub-region function end based on the production equipment running temperature; based on historical data, establish a temperature-qualification rate table for the sub-region function end, which shows that under the condition that other production conditions are completely the same, the product qualification rate corresponding to different production temperatures;

[0026] S3: Digital twin model sub-region function end product flow coordination: establish a stable and controllable product flow to ensure that the output of the previous sub-region function end is completely equivalent to the processing capacity of the next sub-region function end, based on the running speed at this time, record the maximum output, actual output, maximum processing capacity, actual processing capacity and buffer zone maximum capacity of the previous sub-region function end, and calculate the sub-region function end product flow coordination value, and take the production efficiency at this time as the static parameter; the formula is wherein C act , C max are the actual output and maximum output of the previous sub-region function end, Q act , Q max are the actual output and maximum output of the next sub-region function end, and β is the ratio of buffer zone inventory to buffer zone maximum capacity.

[0027] The digital twin ideal model constructs the fixed skeleton of the production line, and the real-time production state model will not change this skeleton, but will only modify the geometric model after each maintenance, so as to get rid of the pressure brought by a large number of algorithms.

[0028] The real-time production state model is constructed based on the digital twin ideal model as a benchmark and relying on its core framework. Based on the real-time dynamic parameter collection and processing of the physical entity of the production line, the static parameter values in the digital twin ideal model are calibrated to dynamic parameter values reflecting the actual running state, accurately and dynamically reproducing the digital model of the real-time running state of the physical packaging production line, and realizing the real-time mapping and linkage of the physical production line and the digital model. In this example, dynamic parameters are collected every 20 seconds to establish a real-time production state model.

[0029] The real-time production state model is constructed using the same method as the digital twin ideal model. The mold wear degree, the sub-region functional end temperature, and the sub-region functional end product flow synergy value are calculated using real-time data transmitted by the monitoring equipment. The method is as follows:

[0030] S1: Actual production mold wear degree calibration: use a contact type roughness meter, select a probe suitable for the hardness of the mold material, read the real-time surface roughness, real-time surface maximum height, and real-time surface minimum height of the production mold according to the measurement results, measure the surface hardness of the production mold using the scratch method, use a set of pencils with known hardness from soft to hard, draw at a 45-degree angle on the mold surface, find the hardest pencil that cannot scratch the coating, and use its hardness as the hardness of the mold surface.

[0031] S2: Real-time production state model sub-region functional end temperature: set a contact sensor to read the sub-region functional end production condition temperature in real time and update it in real time.

[0032] S3: Real-time production state model sub-region functional end product flow synergy value: read the maximum output, actual output of the previous sub-region functional end, maximum processing capacity, actual processing capacity of the next sub-region functional end, and buffer capacity to calculate the sub-region functional end product flow synergy value,

[0033] Real-time intelligent comparison module: based on the digital twin ideal model, set the ideal running interval for all static parameters, and mark the parameters of the real-time production state model that do not fall within the ideal running interval as unbalanced data. The real-time intelligent comparison module periodically calibrates the drift value of the monitoring equipment, effectively eliminating the interference of the equipment's own drift on dynamic parameter measurement, and ensuring the authenticity of the dynamic parameters.

[0034] The monitoring equipment drift value calculation formula is: , D: represents the natural percentage of daily performance decay; P t0 : equipment running data after maintenance; P t1 : empty running equipment running data after interval Δt; Δt: time interval between two empty runs; after obtaining the drift value D of the monitoring equipment, the dynamic data is corrected using the drift value D of the monitoring equipment, and the correction formula is: , δ正 : the correction value after the return; t is the number of days since the calibration date; δ is the measured value of the device on the tth day.

[0035] The ideal operation interval is set based on the static parameter, and the calculation method is as follows: for any dynamic parameter X, the ideal operation interval is defined based on the static parameter as the reference value, and the ideal operation interval is calculated as follows: all product pass rates are sorted from good to bad, and the static parameters X1, X2, X3...Xn corresponding to the first n product pass rates are taken n , the sample mean is calculated , and the sample standard deviation is calculated , then the lower limit of the ideal operation interval of the dynamic parameter X is , and the upper limit is ; wherein n: the first n historical data; : the average value of the static parameters corresponding to the first n pass rates; Z: the Z value corresponding to the confidence level.

[0036] The lower limit of the prediction interval is denoted as L i , and the upper limit of the prediction interval is denoted as U i， When the dynamic parameter X of the monitoring device after the return is i ⊂[L i , U i ], the parameter is qualified. If the dynamic parameter value does not fall within the ideal operation interval, mark it as unbalanced data, calculate the unbalance degree h of the ideal operation interval extreme value for the unbalanced data S, , mark the unbalanced data with h less than 8% as mild unbalance, and mark the unbalanced data with h greater than 8% as severe unbalance; write unbalanced keywords according to the data type, sub-region function end and unbalanced data value of the unbalanced data.

[0037] The comparison result output module: matches the unbalanced keywords for the unbalanced data, outputs the corresponding repair instructions according to the unbalanced keywords, and completes the repair. For unbalanced data X i that is not within the ideal operation interval, different processing methods are set, when X i is greater than U i , it may be that the monitoring device deviates, rather than the actual process really reaching the expected pass level, do not adjust the production parameters immediately, keep the current production device settings, and confirm the monitoring device offset value again, if it is not affected by the monitoring device, record all dynamic parameters during this operation and analyze them; if it is affected by the monitoring device, handle it according to the unbalance degree. When the unbalance degree h remains within 8%, it is mostly short-term parameter fluctuation or occasional measurement error, rather than device failure or process failure, trigger an early warning but do not interrupt production, mark X iAbnormal, but not interrupt production, shorten the dynamic parameter sampling period, shorten the sampling period 20s to 10s, calculate the qualified rate, if the continuous ten times sampling are all unbalanced data, read the unbalanced keyword, match the solution in the solution library and execute; when the unbalance degree h remains above 8%, immediately read the unbalanced keyword, match the solution in the solution library and execute, when there is no complete matching solution, give an approximate solution, and manually determine the solution, mark the unbalanced data after giving the repair instruction as attention state data for continuous monitoring attention. Calculate the product qualified rate of all sub-regional function ends after giving the repair instruction, if the product qualified rate does not rise, mark the repair instruction, and manually judge.

[0038] According to the unbalance keyword, the unbalance reason is judged, if it is judged that the unbalance is caused by the running environment, the system generates environment control instruction to eliminate the interference; if it is judged that the unbalance is caused by the production machine itself factor, the system automatically generates running parameter adjustment scheme or outputs part replacement instruction; after the repair instruction is executed, the system marks the unbalanced data of the corresponding module as attention state data, and includes it in the special monitoring list and continuously tracks its change trend, ensures that the unbalance problem is completely solved and avoids recurrence.

[0039] Comparison result detection module: based on the repaired attention state data, secondary data acquisition is carried out, a new real-time production state model is established and compared with the digital twin ideal model, when the difference interval of the attention state data meets the theoretical difference, the repair is completed, and this time instruction is recorded and filed to the solution library; if it does not meet the theoretical difference, output the number to the user end for manual repair, the data after manual repair is automatically converted into new attention state data, repeat the above process until the difference meets the theoretical range, and record the manual repair instruction to the solution library.

[0040] Finally: the above only for the preferred embodiment of the present application, and does not limit the present application, any modification, equivalent replacement, improvement, etc. within the spirit and principles of the present application, should be included in the protection scope of the present application.

Claims

1. A digital twin simulation and optimization system for a product packaging production line, characterized in that, The sub-region function end division module is used for dividing the whole production line into a plurality of sub-region function ends according to the process flow of the product packaging production line; The parameter acquisition module is used for acquiring static parameters required for establishing a digital twin ideal model of any sub-region function end, and the static parameters include mold wear degree, sub-region function end temperature and sub-region function end product flow coordination value; The parameters corresponding to the static parameters in the real-time production process are marked as dynamic parameters, and the static parameter information and the dynamic parameters are transmitted and stored to the system data storage unit; The model construction module is used for constructing a digital twin ideal model of the sub-region function end based on the static parameter information, and constructing a real-time production state model of the sub-region function end based on the dynamic parameter information; the digital twin ideal model is established as follows: S1: Digital twin model mold wear degree calibration: select a mold that meets the production standard, use a contact roughness meter, select a probe that matches the hardness of the mold material, read the surface roughness, surface maximum height, and surface minimum height of the new production mold according to the measurement results, read the mold surface hardness from the equipment manual, calculate the mold wear degree according to the formula and take the production mold wear degree at this time as the static parameter, the calculation method is wherein Ra is the surface roughness, the unit is μm or nm, which needs to be coordinated with the hardness unit; rp is the surface maximum height, the unit is μm or nm, a positive value from the average reference line; rv is the surface minimum height, the unit is μm or nm, a positive value from the average reference line, indicating the depth; H is the surface hardness, the unit is MPa or HV, which needs to be coordinated with the length unit; S2: The digital twin ideal model sub-region function end temperature: according to the process flow of the product packaging production line, the most suitable temperature is set for each sub-region function end based on the production equipment running temperature, and the most suitable temperature at this time is taken as the static parameter; S3: The digital twin model sub-region function end product flow coordination value: a stable and controllable product flow is established to ensure that the output of the previous sub-region function end is completely equivalent to the processing capacity of the next sub-region function end, the production efficiency is calculated based on the maximum output, actual output, maximum processing capacity, actual processing capacity and maximum buffer capacity of the next sub-region function end at this time, and the production efficiency at this time is taken as the static parameter, and the calculation method is to quantitatively calculate according to the sub-region function end product flow coordination value formula, The calculation formula is C act C max These represent the actual output and maximum output of the previous sub-region's functional end, respectively, Q. act Q max These are the actual output and maximum output of the next sub-region's functional end, respectively, and β is the ratio of the buffer inventory to the maximum capacity of the buffer. The real-time intelligent comparison module sets an ideal running interval based on all parameter settings of the digital twin ideal model, and records the parameters of the real-time production state model that do not fall within the ideal running interval as imbalance data; The comparison result output module matches imbalance keywords for the imbalance data, outputs corresponding repair instructions according to the imbalance keywords, completes repair, and gives data of the repair instructions as attention state data, and continuously monitors the attention state data; The comparison result detection module performs secondary data acquisition based on the repaired attention state data, establishes a new real-time production state model and performs secondary comparison with the digital twin ideal model, and completes repair when the difference interval of the attention state data meets the theoretical difference, and outputs a number to the user end for manual repair if the theoretical difference is not met.

2. The product packaging line oriented digital twin simulation and optimization system of claim 1, wherein, The dynamic parameters are dynamic parameters obtained by deviating from the static parameter values of the mold wear degree, sub-region function end temperature and sub-region function end product flow coordination value in the real-time production process after being affected by external factors.

3. The product packaging line oriented digital twin simulation and optimization system of claim 1, wherein, The real-time production state model is established in the same way as the digital twin ideal model, and the digital twin ideal model is established using static parameters, and the real-time production state model is established using dynamic parameters.

4. The product packaging line oriented digital twin simulation and optimization system of claim 1, wherein, The offset value of the monitoring device is periodically calibrated, the dynamic data is corrected using the offset value of the monitoring device, The correction formula is , D represents the natural attenuation percentage of daily performance; P t0 : is the equipment operation data after the equipment is just maintained; P t1 : is the empty running equipment operation data after the interval Δt; Δt is the time interval of two empty running; after obtaining the offset value D of the monitoring equipment, the dynamic data is corrected by using the offset value D of the monitoring equipment, and the correction formula is: , δ 正 : is the correction value after correction; t is the number of days calculated from the calibration day; δ is the measurement value of the equipment on the tth day.

5. The product packaging line oriented digital twin simulation and optimization system of claim 1, wherein, The imbalance data is a dynamic parameter not falling into an ideal operation interval; the ideal operation interval is set with a static parameter as a core, and the ideal operation interval is calculated as follows: all product qualified rates are sorted from good to bad, and the static parameters X1, X2, X3...Xn corresponding to the first n product qualified rates are taken n , the sample mean is calculated , the sample standard deviation is calculated , and the ideal operation interval of the dynamic parameter X is [ ]; wherein, n: the first n historical data; : the average value of the static parameters corresponding to the first n qualified rates; Z: the Z value corresponding to the confidence level; the dynamic parameter after the return is compared with the ideal operation interval one by one, if the dynamic parameter value falls within the ideal operation interval, it is marked as a normal parameter and is not included in the imbalance data; if the dynamic parameter value does not fall within the ideal operation interval, it is marked as imbalance data.

6. The product packaging line oriented digital twin simulation and optimization system of claim 1, wherein, Lower limit of the prediction interval Let it be L i Predict the upper limit of the interval Let it be U i， When the dynamic parameter X of the monitoring equipment is calibrated i ⊂[L i U i When the parameter is within the ideal operating range, it is considered acceptable. If the value of this dynamic parameter does not fall within the ideal operating range, it is marked as unbalanced data. The degree of imbalance h of the extreme value of the ideal operating range is calculated for the unbalanced data S. Data with an imbalance of h less than 8% is marked as slightly imbalanced, and data with an imbalance of h greater than 8% is marked as severely imbalanced. Imbalance keywords are compiled based on the data attributes, sub-region functional ends, and imbalance data values.

7. The product packaging line oriented digital twin simulation and optimization system of claim 6, wherein, For the sub-region function end with slight imbalance data, the dynamic parameter sampling period is shortened, and if the data falls within the ideal running interval after the sampling period is shortened, the data is no longer processed. If the mild imbalance data appears for ten times continuously or the severe imbalance data appears, the imbalance keyword is read, the repair instruction is given according to the processing scheme in the scheme library, and the imbalance data after the repair instruction is marked as the attention state data for continuous monitoring.

Images