Multi-material pressure self-adaptive closed-loop adjusting system of advertisement making printer

Abstract

The invention discloses a multi-material pressure self-adaptive closed-loop adjustment system of an advertisement making printer, and relates to the technical field of printing pressure adjustment of printers. Comprising a terahertz-environment coupling advanced sensing module, a double-drive pre-judgment adjustment module, a multi-dimensional cross feedback module, a cross-domain collaborative execution module, a real-time lightweight evolution module and a man-machine interaction module. According to the invention, through a terahertz-environment coupling advanced sensing architecture, full-dimension collaborative sensing of deep mechanical characteristics, geometric characteristics, environmental parameters and dynamic deformation trends of materials is realized, data are fused in combination with an improved federated filtering algorithm, environmental interference is effectively avoided, the sensing precision is significantly improved, and the method is suitable for large-scale popularization and application. A digital twinborn and mechanism model double-drive pre-judgment and three-ring dynamic coupling control is adopted to replace a traditional post closed-loop correction mode, high pre-judgment precision and high response speed are both considered, printing defects such as double images and uneven ink layers caused by dynamic deformation response lag are solved fundamentally, and a passive adaptation technical framework existing in the industry for a long time is broken.

Description

Technical Field

[0001] This invention relates to the field of printing pressure adjustment technology for printers, specifically to a multi-material pressure adaptive closed-loop adjustment system for advertising production printers. Background Technology

[0002] To meet the diverse needs of the advertising industry, such as large-scale outdoor advertising, high-precision indoor displays, and personalized customization, professional printing solutions are required to adapt to advertising materials of different materials and sizes. As the core equipment in this field, advertising production printers integrate a variety of printing technologies such as wide-format inkjet, UV curing, and thermal transfer, and can flexibly adapt to various advertising carriers such as lightbox cloth, acrylic sheets, textiles, and photographic paper.

[0003] However, in existing technologies, in multi-material printing scenarios for advertising printing, some printing equipment has a single sensing dimension and has not formed a sensing system that covers material thickness, hardness, and surface condition. It relies solely on a single sensor or manual input to obtain information, which makes it difficult to support accurate adaptation. The control logic is rigid, and it adopts a preset parameter plus open-loop adjustment mode. It lacks a dynamic mapping model between material characteristics and pressure parameters, and there is no closed-loop feedback mechanism for printing quality. It cannot cope with batch differences in materials and dynamic deformation during the printing process. The coordination of the actuators is insufficient. Modules such as nozzle spacing adjustment, pressure roller pressure control, and platform suction force adjustment are independent of each other, and the response speed is difficult to match the needs of high-speed printing. It is impossible to achieve synchronous and continuous optimization of pressure parameters. These limitations together lead to the need for machine stoppage for manual debugging and repeated trial printing when switching between multiple materials. The production efficiency and material utilization rate are greatly reduced, and the waste of consumables is serious. The mismatch between pressure and material can easily cause quality defects such as deformation of thin materials, blurring of thick materials, and ghosting of wide images, increasing the risk of mass production. Summary of the Invention

[0004] The purpose of this invention is to provide a multi-material pressure adaptive closed-loop adjustment system for advertising production printers, so as to solve the problems in the prior art mentioned in the background.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a multi-material pressure adaptive closed-loop adjustment system for an advertising production printer, comprising a terahertz-environment coupling advanced sensing module, a dual-drive predictive adjustment module, a multi-dimensional cross-feedback module, a cross-domain collaborative execution module, a real-time lightweight evolution module, and a human-computer interaction module; Each module constructs a dynamic interactive network with edge computing nodes through a 5G industrial gateway, forming a full-link active control system with coupled perception, dual-drive prediction, cross feedback, and real-time evolution. The terahertz-environment coupling advanced sensing module collects the characteristics of the printing material, environmental parameters and dynamic deformation trends, and outputs a full-dimensional coupled dataset to the dual-drive prediction and adjustment module. The dual-drive predictive adjustment module integrates digital twin and mechanism model to generate the optimal pressure adjustment command to the cross-domain collaborative execution module, while synchronizing data to the real-time lightweight evolution module. The cross-domain collaborative execution module responds to commands to complete cross-domain collaborative adjustment, and the execution status is fed back to the dual-drive prediction and adjustment module. The multi-dimensional cross-feedback module collects quality, execution, and environmental data, cross-verifies them, and feeds them back to the dual-drive prediction and adjustment module and the real-time lightweight evolution module. The real-time lightweight evolution module optimizes the model and then back-empowers the dual-drive prediction and adjustment module. The human-computer interaction module interacts bidirectionally with the core module to enable command issuance and full-link status display.

[0006] Preferably, the terahertz-environment coupled advanced sensing module includes a terahertz spectral sensing unit, a high-speed visual prediction unit, an environmental dynamic sensing unit, and an improved coupling fusion processing unit; The terahertz spectral sensing unit is used to penetrate and detect the internal mechanical parameters of the material, the high-speed visual prediction unit is used to collect the outline of the substrate and predict the pressure requirements, and the environmental dynamic perception unit is used to collect environmental parameters such as temperature and humidity in real time. The three units respectively acquire full-dimensional data of material, morphology and environment and transmit them synchronously to the improved coupling fusion processing unit. This fusion unit uses an improved federated filtering algorithm to couple and fuse three types of data to output a full-dimensional coupled dataset, completing the advanced perception of the printing path and providing accurate and comprehensive data support for the subsequent dual-drive prediction and adjustment module; The terahertz-environment coupled advanced sensing module is based on mechanical parameter inversion calculation and environment-material coupled deformation prediction calculation. The mechanical parameter inversion calculation formula is as follows: ; In the formula: Terahertz spectral attenuation coefficient, unit: dB / mm. The frequency range is 0.1-10 THz for terahertz waves; a, b, c, and d are empirical coefficients obtained through experimental calibration. Different coefficient values ​​correspond to different materials and different terahertz frequency ranges, and are used to fit the relationship between the spectral attenuation coefficient and mechanical parameters. The unit for the elastic modulus of the substrate is GPa; Poisson's ratio of the substrate reflects the ratio of lateral deformation to longitudinal deformation when the material is subjected to pressure, and is used to predict the deformation trend of the material under printing pressure. The formula for predicting deformation due to environment-material coupling is as follows: ; In the formula: To predict dynamic strain; For printing time; The terahertz spectral attenuation coefficient; For printing stress; This is the temperature influence coefficient; For printing temperature; Humidity influence coefficient; This represents the actual humidity. This is the standard humidity.

[0007] Preferably, the dual-drive predictive adjustment module includes a twin modeling submodule, a mechanism predictive submodule, a dual-drive fusion submodule, and a dynamic coupling control submodule; The twin modeling submodule constructs a 1:1 digital twin model based on the coupled sensing dataset, maps the full state of the substrate in real time and outputs high-precision prediction values. The mechanism prediction submodule generates preliminary prediction values ​​based on material mechanics and environmental laws, making up for the time-consuming aspect of twin modeling. The dual-drive fusion submodule receives the outputs of both.

[0008] Preferably, the cross-domain collaborative execution module includes a cross-domain collaborative control unit, a piezoelectric precision adjustment unit, a zoned flexible pressure unit, and a dynamic adsorption adaptation unit; The cross-domain collaborative control unit is the core control unit. It has a built-in collaborative algorithm and is responsible for parsing the optimal pressure adjustment command output by the dual-drive predictive adjustment module, allocating parameters and verifying the status. It also receives feedback from each execution unit in real time and dynamically calibrates the linkage parameters. Piezoelectric precision adjustment units are used for nozzle spacing adjustment; The partitioned flexible pressure unit is equipped with multiple independent partitions to adapt to the pressure distribution on the curved surface of irregularly shaped parts; The dynamic adsorption adapter unit consists of multiple electromagnetic adsorption arrays, used to fix flexible or irregularly shaped substrates. Each electromagnetic adsorption array operates synchronously according to the instructions of the cross-domain collaborative control unit, and at the same time, it provides feedback on the real-time execution status.

[0009] Preferably, the multi-dimensional cross-feedback module includes a multi-dimensional detection unit, a cross-validation unit, and a feedback correction unit; The multi-dimensional detection unit serves as the core of data acquisition, integrating a laser interferometric thickness gauge and a high-speed visual inspection instrument to simultaneously acquire and output three types of raw data: quality, execution, and environment. The cross-validation unit receives raw data and removes abnormal data, improving the reliability of the feedback data. The feedback correction unit performs a quality assessment based on the verified data; The quality assessment is based on a quality assessment formula, as follows: ; In the formula: For print quality level; , , and These are the weighting coefficients; This refers to the actual ink layer thickness. Standard ink layer thickness; This represents the actual overprinting deviation; Standard overprinting deviation; Standard adhesion; Standard adhesion; Actual printing pressure; Dual-drive predicts pressure values.

[0010] Preferably, the real-time lightweight evolution module adopts an integrated architecture of cloud-edge collaboration and lightweight federated learning. The cloud-edge collaboration architecture undertakes the distributed storage of data across the entire chain, realizing real-time computing at the edge and collaboration between the global model in the cloud. The lightweight federated learning architecture relies on its computing power and data support to dynamically update model parameters and complete sparsity compression during printing, adapting to the computing power requirements of the edge.

[0011] Preferably, the modular design and adaptive parameter adjustment of the cross-domain collaborative execution module enable adaptation to wide-format inkjet, UV curing, and thermal transfer printing technologies.

[0012] Compared with the prior art, the beneficial effects of the present invention are: 1. In this invention, a terahertz-environment coupling advanced perception architecture is used to achieve full-dimensional collaborative perception of the material's deep mechanical properties, geometric features, environmental parameters, and dynamic deformation trends. Combined with an improved federated filtering algorithm to fuse data, environmental interference is effectively avoided, and perception accuracy is significantly improved. A dual-drive prediction and three-ring dynamic coupling control using digital twin and mechanism model is adopted to replace the traditional post-closed-loop correction mode, which takes into account both high prediction accuracy and fast response speed. This solves the printing defects such as ghosting and uneven ink layer caused by the lag in dynamic deformation response from the root, breaking the passive adaptation technology framework that has long existed in the industry. 2. In this invention, high-speed visual 3D reconstruction and digital twin modeling are used to accurately predict the pressure distribution on the curved surface of irregularly shaped parts. Combined with partitioned flexible pressure units and dynamic adsorption adaptation units, precise adaptation of irregularly shaped materials can be achieved without manual intervention. Equipped with a real-time lightweight evolution module, the adaptation time for special materials and complex environments is significantly shortened through real-time optimization of the model during the printing process. Furthermore, the co-evolution of models from multiple devices further enhances the adaptability and versatility. The deep linkage of cross-domain collaborative execution and multi-dimensional cross-feedback mechanism can stably adapt to flexible, rigid, and irregularly shaped materials of different thicknesses and hardness, as well as complex printing environments, perfectly matching the diversified, high-precision, and efficient core needs of the advertising production industry. Attached Figure Description

[0013] Figure 1 This is a flowchart of a multi-material pressure adaptive closed-loop adjustment system for an advertising printer according to the present invention. Detailed Implementation

[0014] 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 a part of the embodiments of the present invention, and not all of them. 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.

[0015] Reference Figure 1 As shown: A multi-material pressure adaptive closed-loop adjustment system for an advertising printer, comprising a terahertz-environment coupling advanced sensing module, a dual-drive predictive adjustment module, a multi-dimensional cross-feedback module, a cross-domain collaborative execution module, a real-time lightweight evolution module, and a human-computer interaction module; Each module constructs a dynamic interactive network with edge computing nodes through a 5G industrial gateway, forming a full-link active control system with coupled perception, dual-drive prediction, cross feedback, and real-time evolution. The terahertz-environment coupling advanced sensing module collects the characteristics of the printing material, environmental parameters and dynamic deformation trends, and outputs a full-dimensional coupled dataset to the dual-drive prediction and adjustment module. The terahertz-environment coupled advanced sensing module includes a terahertz spectral sensing unit, a high-speed visual prediction unit, an environmental dynamic sensing unit, and an improved coupled fusion processing unit. The terahertz spectral sensing unit is used to penetrate and detect the internal mechanical parameters of the material, the high-speed visual prediction unit is used to collect the outline of the substrate and predict the pressure requirements, and the environmental dynamic perception unit is used to collect environmental parameters such as temperature and humidity in real time. The three units respectively acquire full-dimensional data of material, morphology and environment and transmit them synchronously to the improved coupling fusion processing unit. The improved coupling and fusion processing unit uses an improved federated filtering algorithm to couple and fuse three types of data to output a full-dimensional coupled dataset, completing the advanced perception of the printing path and providing accurate and comprehensive data support for the subsequent dual-drive prediction and adjustment module. The improved federated filtering algorithm is based on the hierarchical fusion architecture of traditional federated filtering. In response to the coupling requirements of multi-source heterogeneous data (mechanical, geometric, and environmental) in this system, it breaks through the limitations of traditional algorithms with single weights and fixed fusion logic. Through three core improvements, it achieves efficient adaptation and accurate fusion of multi-dimensional data. By introducing dual dynamic weights: adding data credibility weight and scenario adaptation weight, the former is dynamically adjusted based on data collection, and the greater the data fluctuation, the lower the weight; the latter is adapted in real time according to the printing scenario, which solves the fusion deviation problem caused by the fixed weight of traditional algorithms. Meanwhile, heterogeneous data normalization preprocessing is added: for the three types of data dimensional differences (mechanical parameters in GPa / dimensionless, geometric parameters in mm, environmental parameters in ℃ / %RH), the data is mapped to the [0,1] interval through a linear normalization formula to eliminate the interference of dimensional differences on the fusion results; Furthermore, by embedding a real-time anomaly suppression mechanism: using the 3σ criterion (normal distribution anomaly determination) to identify outliers in the three types of data in real time (such as sudden jump data from environmental sensors and terahertz detection blind zone data), the outlier data is assigned a temporary weight of 0 to avoid outliers interfering with the fusion results, while triggering data re-acquisition commands to improve the credibility of the fused data.

[0016] The terahertz-environment coupled advanced sensing module is based on mechanical parameter inversion calculation and environment-material coupled deformation prediction calculation. The mechanical parameter inversion calculation formula is as follows: ; In the formula: Terahertz spectral attenuation coefficient, unit: dB / mm. The frequency range is 0.1-10 THz for terahertz waves; a, b, c, and d are empirical coefficients obtained through experimental calibration. Different coefficient values ​​correspond to different materials and different terahertz frequency ranges, and are used to fit the relationship between the spectral attenuation coefficient and mechanical parameters. The unit for the elastic modulus of the substrate is GPa; Poisson's ratio of the substrate reflects the ratio of lateral deformation to longitudinal deformation when the material is subjected to pressure, and is used to predict the deformation trend of the material under printing pressure. The formula for predicting deformation due to environment-material coupling is as follows: ; In the formula: To predict dynamic strain; For printing time; The terahertz spectral attenuation coefficient; For printing stress; This is the temperature influence coefficient; For printing temperature; Humidity influence coefficient; This represents the actual humidity. This is the standard humidity.

[0017] The terahertz spectral sensing unit uses a 0.1-10THz terahertz time-domain spectrometer with a penetration depth of 0.1-5mm; the high-speed visual prediction unit uses a high-frame-rate industrial camera with a resolution of no less than 20 million pixels; the environmental dynamic sensing unit has an accuracy of ±0.1℃ / ±1%RH; the data fusion unit uses an edge computing chip with a built-in improved federated filtering algorithm, which can adapt to substrates with a thickness of 0.001-100mm and a Shore A0-D100 hardness, and is suitable for environments of 5-40℃ and 30%-80%RH. It can effectively adapt to flexible, rigid, irregularly shaped and dynamically deformable materials, reducing printing defect rate and material waste. The dual-drive predictive adjustment module integrates digital twin and mechanism model to generate the optimal pressure adjustment command to the cross-domain collaborative execution module, and simultaneously synchronizes data to the real-time lightweight evolution module. The dual-drive predictive adjustment module includes a twin modeling sub-module, a mechanism prediction sub-module, a dual-drive fusion sub-module, and a dynamic coupling control sub-module. The twin modeling submodule constructs a 1:1 digital twin model based on the coupled sensing dataset, maps the full state of the substrate in real time, and outputs high-precision prediction values; the mechanism prediction submodule generates preliminary prediction values ​​based on material mechanics and environmental laws simultaneously, making up for the time-consuming aspect of twin modeling; the dual-drive fusion submodule receives the outputs of both. The dual-drive predictive adjustment module uses a dual-drive predictive fusion formula. ( / For twin / mechanistic model weights, / To achieve fusion optimization in response to predicted pressure, the output pressure parameters balance accuracy and speed. The dynamic coupling control submodule, combined with the correction signal from the multi-dimensional cross-feedback module, uses the formula... ( The optimal pressure value is... / / To predict / execute / feedback errors, To implement the deviation correction amount, It outputs adjustment commands and feeds back the execution status to the twin modeling submodule to achieve real-time model correction in order to avoid mechanical lag and correction deviation.

[0018] The cross-domain collaborative execution module responds to commands to complete cross-domain collaborative adjustment, and feeds back the execution status to the dual-drive predictive adjustment module. The cross-domain collaborative execution module includes a cross-domain collaborative control unit, a piezoelectric precision adjustment unit, a zoned flexible pressure unit, and a dynamic adsorption adaptation unit. The cross-domain collaborative control unit is the core control unit, with a built-in collaborative algorithm. It is responsible for parsing the optimal pressure adjustment command output by the dual-drive predictive adjustment module, allocating parameters, and verifying the status. It receives feedback from each execution unit in real time, dynamically calibrates the linkage parameters, and ensures that the multi-parameter collaborative error is ≤0.1N. The piezoelectric precision adjustment unit in the cross-domain collaborative execution module enables adaptive adjustment of the printhead spacing to the ±0.0001mm level, matching the different requirements of various printing technologies for the printhead-substrate spacing (e.g., 2-5mm spacing for wide-format inkjet, 1-3mm spacing for UV curing). The partitioned flexible pressure unit (0-300N adjustable, ±0.5N accuracy) can dynamically adapt to the pressure requirements of different technologies (e.g., 150-250N bonding pressure for thermal transfer, 50-100N anti-ink droplet splatter for wide-format inkjet). The cross-domain collaborative control unit has a built-in multi-printing technology adaptation logic library, which can parse the printing parameter instructions of different technologies and synchronously link each execution unit to complete parameter calibration without modifying the core structure of the original equipment. Combined with the various printing technology adaptation parameters stored in the real-time lightweight evolution module, the optimized parameters can be quickly called up during switching to achieve efficient adaptation of multiple technologies and meet the complex printing needs of various advertising production scenarios. At the same time, the system adapts to various printing technologies such as wide-format inkjet, UV curing, and thermal transfer through the modular design and parameter adaptive adjustment of the cross-domain collaborative execution module.

[0019] The multi-dimensional cross-feedback module collects quality, execution, and environmental data (quality data: including actual ink layer thickness, registration deviation, and adhesion; execution data: including actual execution parameters such as printhead spacing, zone pressure, and adsorption force; environmental data: including temperature and humidity of the printing area and airflow parameters), cross-verifies the data, and then feeds it back to the dual-drive prediction and adjustment module and the real-time lightweight evolution module. The multi-dimensional cross-feedback module includes a multi-dimensional detection unit, a cross-verification unit, and a feedback correction unit. The multi-dimensional detection unit serves as the core of data acquisition, integrating a laser interferometric thickness gauge and a high-speed visual inspection instrument to simultaneously acquire and output three types of raw data: quality, execution, and environment. The cross-validation unit receives raw data and removes abnormal data, improving the reliability of the feedback data. The feedback correction unit performs a quality assessment based on the verified data; The quality assessment is based on a quality assessment formula, as follows: ; in, For print quality level, , , and These are the weighting coefficients. This represents the actual ink layer thickness. Standard ink layer thickness, This represents the actual overprinting deviation. For standard registration deviation, For standard adhesion, Standard adhesion, Actual printing pressure, Dual-drive predicts pressure values.

[0020] The real-time lightweight evolution module optimizes the dual-drive prediction model and then empowers the dual-drive prediction adjustment module. The real-time lightweight evolution module adopts an integrated architecture of cloud-edge collaboration and lightweight federated learning, with the two working together in a "support-empowerment" logic. The cloud-edge collaboration architecture includes a local 2TB SSD edge device and a 5G cloud server, undertaking distributed storage of data across the entire link, realizing real-time computing at the edge and cloud computing power collaboration. The edge device prioritizes real-time optimization tasks to ensure response speed, while the cloud is responsible for data aggregation from multiple devices and global model co-evolution. The lightweight federated learning architecture relies on the computing power and data support of cloud-edge collaboration. Through real-time optimization formulas, it dynamically updates model parameters during the printing process at the edge without waiting for printing to complete. Lightweight compression is then performed to sparsely compress the optimized parameters, avoiding excessive consumption of edge computing resources. The two work together to achieve "real-time evolution during printing + lightweight adaptation to edge computing power," ensuring both model optimization accuracy and system response speed, while supporting multi-device model co-evolution and improving the adaptation efficiency for similar materials / environments. The calculation formula for the dual-drive prediction model is as follows: ; In the formula: These are the optimized prediction model parameters; Predicted model parameters before optimization; The learning rate; The gradient of the loss function is calculated based on the parameters before optimization (reflecting the trend of model error changes). Weights for real-time training data; This is the regularization coefficient.

[0021] The human-machine interaction module is equipped with a touch screen and an AR visualization interface. It establishes a two-way data interaction link with the system's core modules (terahertz-environment coupling advanced perception module, dual-drive predictive adjustment module, multi-dimensional cross-feedback module, cross-domain collaborative execution module, and real-time lightweight evolution module) to achieve precise command issuance and full-link visual collaborative functions. On the one hand, it receives operation commands issued by users through the touch screen or AR interface, including preset printing parameters, equipment start-up and shutdown control, manual intervention adjustment, and scene switching. These commands are synchronously transmitted to the corresponding core modules via the 5G industrial gateway, triggering execution actions such as parameter configuration, process initiation, or emergency adjustment. On the other hand, it gathers real-time full-link working status data from each core module, covering material mechanical parameters and environmental data of the perception layer, digital twin models and pressure prediction curves of the predictive layer, real-time parameters of pressure / spacing / adhesion force of the execution layer, quality level and error data of the feedback layer, and model optimization status of the evolution layer. This data is dynamically displayed and traced through the AR visualization interface, and anomaly warnings (such as quality level Q < 90, equipment failure) are triggered simultaneously, along with solution push notifications, ensuring ease of operation and full-link status traceability.

[0022] The working principle of this solution: The core of this system follows the closed-loop working principle of coupled sensing, dual-drive prediction, cross-domain execution, cross feedback and real-time evolution. Each module works together to build an active pressure regulation system. The terahertz-environment coupled advanced sensing module adopts a three-in-one architecture of terahertz deep detection, visual contour capture and dynamic environmental acquisition. It simultaneously captures material mechanical parameters, geometric features, environmental parameters and dynamic deformation trends. Then, it is fused and processed by an improved federated filtering algorithm to output a full-dimensional coupled dataset, which provides accurate data support for the subsequent prediction stage and effectively solves the pain points of traditional sensing with single dimensions and neglect of environmental interference. The dual-drive predictive adjustment module constructs a dual-drive predictive system combining a digital twin and a mechanistic model. The digital twin modeling submodule, based on a coupled dataset, builds a 1:1 high-precision digital twin model that maps the entire state of the substrate in real time. The mechanistic prediction submodule, relying on material mechanics, quickly outputs preliminary prediction values. After calculation using a fusion formula, the two modules output a pressure prediction value that balances accuracy and speed. Based on this, and combined with error signals transmitted by the cross-feedback module, an optimal pressure adjustment command is generated through a three-loop coupled correction formula, fundamentally avoiding mechanical lag and correction deviation problems. The execution layer focuses on cross-domain linkage. The cross-domain collaborative execution module responds to the optimal adjustment command. The micro-nano level control of the nozzle spacing is achieved through the piezoelectric precision adjustment unit. The partitioned flexible pressure unit accurately adapts to the pressure distribution requirements of complex curved surfaces. The dynamic adsorption adaptation unit firmly fixes the material to prevent printing displacement. The cross-domain collaborative control unit ensures the linkage adjustment of multiple parameters, controls the collaborative error within 0.1N, and ensures that the execution accuracy meets the standard. The feedback layer focuses on multi-dimensional verification. The multi-dimensional cross-feedback module simultaneously collects three types of data: quality, execution, and environment. After removing abnormal data interference through the cross-verification formula, it outputs a quality level (Q value) of 0-100 through the quality assessment formula. When Q < 90, it automatically triggers defect warning and correction signals, which are simultaneously fed back to the prediction module and the evolution module to form a closed-loop correction logic, avoiding correction deviations caused by single feedback. The evolution layer adopts a cloud-edge collaborative and lightweight federated learning architecture. At the edge, based on feedback data, the prediction model parameters are dynamically updated through real-time optimization formulas and then adapted to the computing power of the edge after lightweight compression. The cloud aggregates data from multiple devices to achieve global model co-evolution, which in turn empowers the prediction module and continuously improves the system's long-term adaptability and versatility, truly realizing "real-time evolution during printing" without waiting for printing to be completed, and dynamically optimizing the adaptation accuracy. Each module establishes a dynamic interactive network with edge computing nodes through a 5G industrial gateway, keeping data transmission latency within 5ms to ensure real-time response across the entire link, ultimately achieving precise adaptation and efficient printing for multiple materials, environments, and technologies.

[0023] How to use this solution: When using the printing equipment, after turning on the power of each module, start the system through the human-machine interaction module. The system will automatically complete the hardware self-test to ensure that each module is fault-free. Select the printing technology through the AR visualization interface, preset the material standard parameters, environmental benchmark parameters and quality qualification threshold (default Q≥90), and also support importing custom parameter templates to adapt to personalized needs. After the parameters are preset, the system automatically starts the cross-domain collaborative execution module to perform calibration, so that each execution unit returns to zero position. After the calibration accuracy meets the standard, it enters the standby state.

[0024] Place the material to be printed (rigid, flexible, or irregular shape) on the printing platform and activate the dynamic adsorption and adaptation unit to complete the initial fixation, ensuring that the material is flat and wrinkle-free. If it is an irregularly shaped material, it needs to be precisely positioned to fit the curved surface of the platform. Click to start the perception function, and the three perception units of terahertz, vision, and environment will work simultaneously to collect data from all dimensions and complete the fusion to generate a material-specific dataset. The interface will display the core parameters in real time, and after the perception is completed, it will prompt that printing can be started.

[0025] After clicking "Start Printing," the system will automatically execute the dual-drive prediction and cross-domain collaborative adjustment process, output pressure adjustment commands in real time, and each execution unit will respond synchronously. The AR interface will display the full-link status data in real time, including the digital twin model, pressure adjustment curve, dynamic deformation prediction trend, and quality level. If an abnormality occurs, the system will trigger an audible and visual warning, and simultaneously display the fault location and solution for quick handling. If parameters need to be adjusted, the pressure value, nozzle spacing, etc. can be manually modified through the touch screen. The system will respond in real time and complete the calibration. The manual intervention process will not affect the printing continuity.

[0026] When switching printing materials or technologies, the system will automatically call the adaptation parameters stored in the real-time evolution module to quickly complete the parameter adjustment without modifying the equipment structure; after switching, it will automatically perform perception calibration to ensure adaptation accuracy. After printing, the system will automatically store printing parameters, quality data, model optimization logs, and other information, and supports report export. At the same time, the data will be synchronized to the cloud server, and historical data and printing records can be viewed through remote terminals, which facilitates production traceability and process optimization. After printing each day, the terahertz probe, camera lens, and print head must be cleaned to prevent dust from affecting detection and printing accuracy. The cross-domain collaborative execution unit needs to be calibrated monthly, and the empirical coefficients of the mechanism model should be updated based on batch printing data to ensure the long-term operating accuracy of the system. Redundant data stored on the edge SSD should be cleaned up regularly, retaining only core parameters and models to free up storage space and ensure smooth system operation.

[0027] Through the above operations, efficient and accurate printing of multiple materials and in multiple scenarios can be achieved, significantly reducing manual debugging costs, improving production efficiency and printing quality, and fully adapting to the diversified needs of the advertising production industry.

[0028] Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A multi-material pressure adaptive closed-loop adjustment system for an advertising printing printer, characterized in that: It includes a terahertz-environment coupling advanced sensing module, a dual-drive prediction and adjustment module, a multi-dimensional cross-feedback module, a cross-domain collaborative execution module, a real-time lightweight evolution module, and a human-machine interaction module. Each module constructs a dynamic interactive network with edge computing nodes through a 5G industrial gateway, forming a full-link active control system of coupled sensing, dual-drive prediction, cross-feedback, and real-time evolution. The terahertz-environment coupling advanced sensing module collects the characteristics of the printing material, environmental parameters and dynamic deformation trends, and outputs a full-dimensional coupled dataset to the dual-drive prediction and adjustment module. The dual-drive predictive adjustment module integrates digital twin and mechanism model into a dual-drive predictive model, generates the optimal pressure adjustment command to the cross-domain collaborative execution module, and simultaneously synchronizes data to the real-time lightweight evolution module. The cross-domain collaborative execution module responds to commands to complete cross-domain collaborative adjustment, and the execution status is fed back to the dual-drive prediction and adjustment module. The multi-dimensional cross-feedback module collects quality, execution, and environmental data, cross-verifies them, and feeds them back to the dual-drive prediction and adjustment module and the real-time lightweight evolution module. The real-time lightweight evolution module optimizes the dual-drive prediction model and then inversely empowers the dual-drive prediction adjustment module. The human-computer interaction module interacts bidirectionally with the terahertz-environment coupling advanced perception module, dual-drive prediction and adjustment module, multi-dimensional cross-feedback module, cross-domain collaborative execution module, and real-time lightweight evolution module to achieve command issuance and full-link status display.

2. The multi-material pressure adaptive closed-loop adjustment system for advertising production printers according to claim 1, characterized in that, The terahertz-environment coupled advanced sensing module includes a terahertz spectral sensing unit, a high-speed visual prediction unit, an environmental dynamic sensing unit, and an improved coupled fusion processing unit. The terahertz spectral sensing unit is used to penetrate and detect the internal mechanical parameters of the material, the high-speed visual prediction unit is used to collect the outline of the substrate and predict the pressure requirements, and the environmental dynamic perception unit is used to collect temperature and humidity environmental parameters in real time. The three units respectively acquire material, morphology and environmental data in all dimensions and transmit them synchronously to the improved coupling fusion processing unit. The improved coupled fusion processing unit uses an improved federated filtering algorithm. This algorithm introduces dual dynamic weights, adds heterogeneous data normalization preprocessing, and embeds a real-time anomaly suppression mechanism to couple and fuse three types of data to output a full-dimensional coupled dataset, thus completing the advanced perception of the printing path and providing data support for the subsequent dual-drive prediction and adjustment module. The terahertz-environment coupled advanced sensing module is based on mechanical parameter inversion calculation and environment-material coupled deformation prediction calculation. The mechanical parameter inversion calculation formula is as follows: ； In the formula: Terahertz spectral attenuation coefficient, unit: dB / mm; The frequency range is 0.1-10 THz for terahertz waves; a, b, c, and d are empirical coefficients obtained through experimental calibration. Different coefficient values ​​correspond to different materials and different terahertz frequency ranges, and are used to fit the relationship between the spectral attenuation coefficient and mechanical parameters. The elastic modulus of the substrate, in GPa; Poisson's ratio of the substrate reflects the ratio of lateral deformation to longitudinal deformation when the material is subjected to pressure, and is used to predict the deformation trend of the material under printing pressure. The formula for predicting deformation due to environment-material coupling is as follows: ； In the formula: To predict dynamic strain; For printing time; For printing stress; This is the temperature influence coefficient; For printing temperature; Humidity influence coefficient; This represents the actual humidity. This is the standard humidity.

3. The multi-material pressure adaptive closed-loop adjustment system for advertising production printers according to claim 1, characterized in that, The dual-drive predictive adjustment module includes a twin modeling submodule, a mechanism predictive submodule, a dual-drive fusion submodule, and a dynamic coupling control submodule. The twin modeling submodule constructs a 1:1 digital twin model based on the coupled sensing dataset, maps the full state of the substrate in real time and outputs high-precision prediction values. The mechanism prediction submodule generates preliminary prediction values ​​based on material mechanics and environmental laws. The dual-drive fusion submodule receives the outputs of both.

4. The multi-material pressure adaptive closed-loop adjustment system for advertising production printers according to claim 1, characterized in that, The cross-domain collaborative execution module includes a cross-domain collaborative control unit, a piezoelectric precision adjustment unit, a zoned flexible pressure unit, and a dynamic adsorption adaptation unit. The cross-domain collaborative control unit is the core control unit. The core control unit has a built-in collaborative algorithm, which is responsible for parsing the optimal pressure adjustment command output by the dual-drive predictive adjustment module, allocating parameters and verifying the status. It receives feedback from the piezoelectric precision adjustment unit, the partitioned flexible pressure unit and the dynamic adsorption adaptation unit in real time, and dynamically calibrates the linkage parameters. Piezoelectric precision adjustment units are used for nozzle spacing adjustment; The partitioned flexible pressure unit is equipped with multiple independent partitions to adapt to the pressure distribution on the curved surface of irregularly shaped parts; The dynamic adsorption adapter unit consists of multiple electromagnetic adsorption arrays, used to fix flexible or irregularly shaped substrates. Each electromagnetic adsorption array operates synchronously according to the instructions of the cross-domain collaborative control unit, and at the same time, it provides feedback on the real-time execution status.

5. The multi-material pressure adaptive closed-loop adjustment system for advertising production printers according to claim 1, characterized in that, The multi-dimensional cross-feedback module includes a multi-dimensional detection unit, a cross-validation unit, and a feedback correction unit; The multi-dimensional detection unit serves as the core of data acquisition, integrating a laser interferometric thickness gauge and a high-speed visual inspection instrument to simultaneously acquire and output three types of raw data: quality, execution, and environment. The cross-validation unit receives the raw data and removes abnormal data. The feedback correction unit performs a quality assessment based on the verified data; The quality assessment is based on a quality assessment formula, as follows: ； In the formula: For print quality level; , , and These are the weighting coefficients; This refers to the actual ink layer thickness. Standard ink layer thickness; This represents the actual overprinting deviation; Standard overprinting deviation; Standard adhesion; Standard adhesion; Actual printing pressure; Dual-drive predicts pressure values.

6. The multi-material pressure adaptive closed-loop adjustment system for advertising production printers according to claim 1, characterized in that, The real-time lightweight evolution module adopts an integrated architecture of cloud-edge collaboration and lightweight federated learning. The cloud-edge collaboration architecture undertakes the distributed storage of data across the entire chain, enabling real-time computing at the edge and collaboration between the global model in the cloud. The lightweight federated learning architecture, relying on its computing power and data support, dynamically updates model parameters and completes sparsity compression during printing, adapting to the computing power requirements of the edge.

7. The multi-material pressure adaptive closed-loop adjustment system for advertising production printers according to claim 1, characterized in that, Through the modular design and adaptive parameter adjustment of the cross-domain collaborative execution module, it is possible to adapt to wide-format inkjet, UV curing and thermal transfer printing technologies.

Images