Intelligent control system and method for UV printer based on remote task scheduling

Abstract

The application provides a UV printer intelligent control system and method based on remote task scheduling, the system comprises a cloud scheduling center module for global task planning and distribution, network state monitoring, equipment state monitoring, and data storage and analysis; an edge gateway node module for caching task data issued by the cloud scheduling center module, serving as a local scheduling hub in the event of network interruption, and simultaneously performing collaborative control, fault diagnosis and early warning on multiple devices; and a UV printer terminal module for real-time acquisition of state data of the printing equipment, receiving tasks and performing, local data caching and human-computer interaction. The combination of cloud global optimization, edge local collaboration and terminal offline autonomy reduces the dependence on a single network, improves overall production efficiency through multi-device collaboration and load balancing, reduces ink waste and equipment invalid loss based on a prediction model, and optimizes the printing path to reduce energy consumption.

Description

Technical Field

[0001] This invention relates to the field of industrial printing equipment technology, and more particularly to an intelligent control system and method for UV printers based on remote task scheduling. Background Technology

[0002] A UV printer (Ultraviolet LED Inkjet Printer) is a high-tech, plate-free, full-color digital printer. It sprays UV ink onto the substrate through a printhead, then uses a UV lamp to instantly cure and dry the ink, completing the printed image. It is not limited by materials and can perform photo-quality color printing on surfaces such as T-shirts, sliding doors, cabinet doors, glass, sheets, various signs, crystal, PVC, acrylic, metal, plastic, stone, and leather. However, current technology still has significant limitations regarding remote control of UV printers.

[0003] Existing UV printers suffer from problems in remote scheduling, including strong network dependence, low device collaboration efficiency, weak task fault tolerance, and inefficient energy and consumable management. For example, network outages can easily lead to task loss, uneven load distribution among multiple devices, and the inability to dynamically adjust tasks based on real-time device status, resulting in resource waste.

[0004] Therefore, there is an urgent need for an intelligent control system and method for UV printers that can reduce dependence on a single network, enable multi-device collaboration, improve task fault tolerance, and provide precise management of energy consumption and consumables. Summary of the Invention

[0005] To address the aforementioned issues, this invention provides an intelligent control system and method for UV printers based on remote task scheduling, which can reduce network dependence, enable dynamic collaboration among multiple devices, provide offline fault tolerance, and optimize energy consumption and consumables.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] This invention provides an intelligent control system for UV printers based on remote task scheduling, including a cloud scheduling center module, an edge gateway node module, and a UV printer terminal module that are connected in communication.

[0008] The cloud-based scheduling center module is used for global task planning and allocation, network status monitoring, device status monitoring, and data storage and analysis.

[0009] The cloud-based scheduling center module includes a task parsing and priority sorting unit, a device cluster load balancing unit, a network status adaptive scheduling unit, and an energy consumption and consumables optimization unit.

[0010] Task parsing and priority sorting unit: used to receive printing tasks uploaded by users, parse task attributes and assign priorities to tasks according to preset rules, and synchronize the prioritized tasks to the device cluster load balancing unit and user clients.

[0011] Device cluster load balancing unit: used to acquire real-time status data of printing devices, allocate weights according to the degree of impact on task execution and calculate a comprehensive suitability score, and allocate tasks to the optimal device based on the load balancing algorithm;

[0012] Network Status Adaptive Scheduling Unit: Used to monitor the network quality of the time center module and edge gateway node module in the cloud scheduling in real time, and dynamically adjust the data transmission strategy according to the divided network status level;

[0013] Energy consumption and consumables optimization unit: Based on historical data, a prediction model is trained to estimate ink consumption and energy consumption according to task parameters, and compared with the real-time consumables balance data to determine whether the consumables are sufficient. If they are insufficient, a multi-level warning is triggered to replenish the stock in time.

[0014] The edge gateway node module is used to cache task data issued by the cloud scheduling center module, serve as a local scheduling hub when the network is interrupted, and simultaneously perform coordinated control and fault diagnosis and early warning for multiple devices.

[0015] The UV printer terminal module is used to collect real-time status data of the printing device, receive and execute tasks, cache data locally, and interact with the human-machine interface.

[0016] The execution process of the task parsing and priority sorting unit includes the following steps:

[0017] The system automatically generates a unique task ID for print jobs uploaded by users through the client and verifies the validity of the uploaded tasks.

[0018] By using structured parsing algorithms, key attributes are extracted from the validated task data to form a standardized task attribute table;

[0019] A priority determination rule base is set up, with each rule corresponding to an independent weight calculation logic. The rule dimensions include time urgency rules, equipment compatibility rules, and production demand rules.

[0020] The total task weight is obtained by merging the weights of various rules in the calculation. The task priority is then assigned based on the total task weight, and manual adjustment is supported.

[0021] Add a priority level field to the task attribute table and generate priority tags. Then, synchronize the task data with priority tags to the device cluster load balancing unit and user clients.

[0022] The key attributes include basic attributes, process attributes, time attributes, and resource attributes; the basic attributes include task ID, upload time, and user information; the process attributes include material type, printing precision, and color mode; the time attributes include deadline and expected completion time; and the resource attributes include printing area and estimated ink type.

[0023] The execution process of the device cluster load balancing unit includes the following steps:

[0024] The edge gateway node module establishes continuous communication with each UV printer terminal to obtain real-time printer status data, including current task progress data, printhead health data, idle time data, and historical failure rate data.

[0025] The acquired status data is standardized to eliminate the differences in the units of data from different dimensions. Based on the degree of influence of each status data on task execution, weights are assigned to each data dimension to construct a comprehensive device adaptability scoring model.

[0026] The standardized scores of each dimension are multiplied by their corresponding weights and summed to obtain the overall compatibility score for each printer.

[0027] An improved genetic algorithm is adopted, with the addition of two optimization terms: equipment load constraint and task-process matching degree. The optimal task allocation scheme is calculated and output with the goal of minimizing the total completion time of all tasks and minimizing the difference in equipment load.

[0028] The optimal allocation scheme for the output is verified for process matching and load limit. After the verification is passed, the task instruction is sent to the target printer through the edge gateway node module, and the task-device allocation relationship is recorded in the cloud scheduling center module.

[0029] During task execution, the device status update data synchronized by the edge gateway node module is received in real time, and the load distribution is dynamically adjusted.

[0030] The load balancing algorithm is an improved genetic algorithm, and the calculation of the optimal allocation scheme includes the following steps:

[0031] The task-device matching relationship is binary encoded to generate an initial population;

[0032] Using the reciprocal of total completion time and the reciprocal of the standard deviation of equipment load as core indicators, a fitness function is constructed to evaluate the merits of the allocation scheme;

[0033] The roulette wheel selection method is used to select the best solutions with higher fitness values ​​according to their proportion to enter the next generation. The selected solutions are then cross-crossed at a single point, and the cross-crossed solutions are then mutated.

[0034] Repeat the selection, crossover, and mutation operations as described above until the number of iterations reaches a preset value or the fitness value changes by less than 0.001 over 10 consecutive iterations. Output the solution with the highest fitness value as the optimal task allocation solution.

[0035] The execution process of the network state adaptive scheduling unit includes the following steps:

[0036] Obtain network quality parameters for the cloud scheduling center module and the edge gateway node module, including latency, packet loss rate, and bandwidth;

[0037] The collected network quality parameters are standardized and preprocessed, and the network status level is determined according to the preset threshold, which is divided into three levels: good, fluctuating and interrupted.

[0038] Based on the determined network status level, the corresponding transmission strategy is invoked to ensure data transmission efficiency and reliability;

[0039] During the execution of the transmission strategy, the effect of the strategy execution is monitored in real time, and the strategy is adjusted based on the feedback data.

[0040] The execution process of the energy consumption and consumables optimization unit includes the following steps:

[0041] Historical data related to energy consumption and consumables are collected in batches from the historical database of the cloud dispatch center module and the local logs of the UV printer terminal to build a basic dataset, and the dataset is standardized and preprocessed.

[0042] Based on the preprocessed historical dataset, a gradient boosting regression model is used to train a prediction model to accurately predict ink consumption and energy consumption for new tasks, and the model accuracy is continuously optimized through iteration.

[0043] When the cloud scheduling center module receives a new printing task, the energy consumption and consumables optimization unit extracts the task parameters and inputs them into the trained prediction model, outputting the ink consumption and energy consumption prediction results.

[0044] Based on the print pattern size and printer table coordinates of the task, the optimal printhead movement path is planned by combining the contour offset algorithm with path sorting optimization.

[0045] The system acquires real-time data on the remaining consumables of the UV printer terminal, compares it with the estimated ink consumption, and determines whether the consumables are sufficient. If they are insufficient, it triggers multi-level warnings to replenish the stock in a timely manner.

[0046] The edge gateway node module includes a local task caching and resume unit, a device collaborative control unit, and a fault diagnosis and early warning unit.

[0047] Local task caching and resume unit: used to cache task data sent by the cloud scheduling center module. When the network is interrupted, the local scheduling mode is automatically activated and the cached tasks are executed according to priority. After the network is restored, the local execution records are synchronized to the cloud scheduling center module.

[0048] Device collaboration control unit: used to break down large tasks into multiple sub-tasks and control multiple printers in parallel for local collaboration;

[0049] Fault diagnosis and early warning unit: Based on real-time data collected by printer sensors, it identifies abnormal states through edge computing and triggers local early warnings, while pushing fault codes to the cloud dispatch center module.

[0050] The UV printer terminal module includes a task execution and status acquisition unit, an offline autonomous unit, and a human-computer interaction unit;

[0051] Task execution and status acquisition unit: Used to receive task instructions from edge gateway node module or cloud scheduling center module, control nozzle movement and UV lamp curing, and collect real-time status data through sensors;

[0052] Offline Autonomous Unit: It has a built-in local storage module to cache the data of the most recent 100 tasks. When the network between the edge gateway node module and the UV printer terminal is interrupted, it automatically starts the preset rules.

[0053] The preset rules are as follows: prioritize the execution of cached level 1-2 emergency tasks; if the task execution is interrupted, record the breakpoint coordinates and resume from the breakpoint after the network is restored; save the task log locally and continue uploading it after the network is restored.

[0054] Human-machine interaction unit: Equipped with a touch screen and physical buttons, it supports local modification of task parameters and displays task progress, equipment status and fault prompts, facilitating intervention by on-site operators.

[0055] On the other hand, the present invention provides an intelligent control method for a UV printer based on remote task scheduling, comprising the following steps:

[0056] Get the print jobs uploaded by users, perform structured parsing on the jobs, extract the job parameters, and assign priorities to the jobs;

[0057] Real-time monitoring of network quality parameters, classification of network status levels based on thresholds, and matching of corresponding execution strategies based on different levels. The network quality parameters include latency, packet loss rate, and bandwidth.

[0058] Obtain real-time status parameters of printing devices, calculate the comprehensive suitability score of each device, and allocate tasks to the optimal device based on a load balancing algorithm;

[0059] After receiving the task instruction, the device executes the task and provides real-time feedback on the remaining consumables data of the UV printer terminal. It compares the remaining consumables data with the estimated ink consumption output by the prediction model to determine whether replenishment is needed and issues an alert.

[0060] The beneficial effects of this invention are as follows:

[0061] This invention provides an intelligent control system and method for UV printers based on remote task scheduling. The system includes a cloud-based scheduling center module for global task planning and allocation, network status monitoring, device status monitoring, and data storage and analysis; an edge gateway node module for caching task data issued by the cloud-based scheduling center module, serving as a local scheduling hub during network interruptions, and simultaneously coordinating control and fault diagnosis and early warning for multiple devices; and a UV printer terminal module for real-time acquisition of printing device status data, receiving and executing tasks, with local data caching and human-machine interaction. A three-tiered hybrid scheduling model is adopted, combining cloud-based global optimization, edge-based local collaboration, and terminal offline autonomy to reduce dependence on a single network. Tasks can resume locally during network interruptions and seamlessly synchronize after network recovery. An improved genetic algorithm achieves optimal device-task matching, and the efficiency of large tasks is improved by over 40% through splitting and parallel execution. Overall production efficiency is improved through multi-device collaboration and load balancing. Ink waste and ineffective equipment wear are reduced based on predictive models, and energy consumption is reduced by optimizing printing paths. A CNN model is deployed on edge nodes, achieving a fault identification accuracy of ≥92% and a response time of <1 second, reducing downtime. Attached Figure Description

[0062] Figure 1 This is a framework diagram of the intelligent control system for UV printers based on remote task scheduling, as described in this invention.

[0063] Figure 2 This is a flowchart of the intelligent control method for UV printers based on remote task scheduling according to the present invention. Detailed Implementation

[0064] Please see Figure 1 As shown, the present invention provides an intelligent control system for UV printers based on remote task scheduling, including a cloud scheduling center module, an edge gateway node module, and a UV printer terminal module that are connected in communication.

[0065] The cloud-based scheduling center module is used for global task planning and allocation, network status monitoring, device status monitoring, and data storage and analysis.

[0066] The edge gateway node module is used to cache task data issued by the cloud scheduling center module, serve as a local scheduling hub when the network is interrupted, and simultaneously perform coordinated control and fault diagnosis and early warning for multiple devices.

[0067] The UV printer terminal module is used to collect real-time status data of the printing device, receive and execute tasks, cache data locally, and interact with the human-machine interface.

[0068] The cloud-based scheduling center module includes a task parsing and priority sorting unit, a device cluster load balancing unit, a network status adaptive scheduling unit, and an energy consumption and consumables optimization unit.

[0069] Task parsing and priority sorting unit: used to receive printing tasks uploaded by users, parse task attributes and assign priorities to tasks according to preset rules, and synchronize the prioritized tasks to the device cluster load balancing unit and user clients.

[0070] The execution process of the task parsing and priority sorting unit includes the following steps:

[0071] The system automatically generates a unique task ID for print jobs uploaded by users through the client and verifies the validity of the uploaded tasks.

[0072] Users upload print jobs (such as pattern files, material parameters, accuracy requirements, and deadlines) through the client (PC / mobile management system). The system automatically generates a unique task ID (e.g., "UV-Task-20250520-001") and records the upload time and user information, such as operator account and department. The uploaded task file undergoes validity verification. If verification fails, an error message (e.g., "Material parameters are missing; please supplement and re-upload") is generated and sent to the user's client. Simultaneously, the error type and time are recorded in the system log. The receiving process is retried after the user corrects the error. The verification specifically includes three categories:

[0073] File integrity: Verify that the pattern file (such as CAD, PDF, PNG format) is complete and undamaged or missing (confirm that the file transmission was not erroneous by comparing the MD5 value).

[0074] Parameter completeness: Check whether the required task parameters are included, including material parameters (such as metal, glass, plastic), accuracy requirements, deadline, and print size;

[0075] Format compatibility: Determines whether the pattern file format matches the system's supported format library. If the format is incompatible, it automatically triggers format conversion, using the FFmpeg tool to convert the non-standard format to the standard PNG format.

[0076] By using structured parsing algorithms, key attributes are extracted from the validated task data to form a standardized task attribute table;

[0077] The key attributes include basic attributes, process attributes, time attributes, and resource attributes; the basic attributes include task ID, upload time, and user information; the process attributes include material type, printing precision, and color mode; the time attributes include deadline and expected completion time; and the resource attributes include printing area and estimated ink type.

[0078] A priority determination rule base is set up, with each rule corresponding to an independent weight calculation logic. The rule dimensions include time urgency rules, equipment compatibility rules, and production demand rules.

[0079] Time urgency weight (40 points): Calculate the ratio of the remaining available time for the task (remaining available time = deadline - current time) to the estimated execution time. The smaller the ratio, the higher the weight. Specifically, the calculated ratio is divided into different intervals, and each interval corresponds to a weighted score. A ratio between (0, 0.3) has a weight of 40 points, corresponding to an extremely urgent task; a ratio between (0.3, 0.5) has a weight of 32-38 points, corresponding to a highly urgent task; a ratio between (0.5, 0.7) has a weight of 25-31 points, corresponding to a medium-urgent task; a ratio between (0.7, 1.0) has a weight of 20-24 points, corresponding to a low-urgent task; and a ratio greater than 1.0 has a weight of 15 points or less, corresponding to a relaxed task. For example, if there are 2 hours of remaining available time and an estimated execution time of 3 hours, the ratio = 2 / 3 ≈ 0.67, corresponding to a weight of 30 points; if there are 1 hour of remaining available time and an estimated execution time of 3 hours, the ratio ≈ 0.33, corresponding to a weight of 40 points.

[0080] Device compatibility weight (30 points): Based on the matching of task process attributes (material, precision) with the device capability library (material types, maximum precision, ink types supported by each printer). The more matching items, the higher the weight. For example, if the task requires "metal material + 600 DPI + white ink", device A supports all 3 items and gets 30 points; device B only supports "metal material + 600 DPI" and gets 20 points.

[0081] Production demand weight (30 points): The weight is assigned according to the production type of the task. Urgent orders (such as urgent customer repair parts) get 30 points, regular orders get 20 points, and trial production orders (such as new product sampling) get 10 points. If the task is associated with after-sales work orders (such as product quality reprinting), an additional 5 points are added (the weight cap is still 30 points).

[0082] The total task weight is obtained by merging the weights of various rules in the calculation. The task priority is then assigned based on the total task weight, and manual adjustment is supported.

[0083] The tasks are divided into 5 priority levels based on their total weight, with the specific correspondence as follows:

[0084] Level 1 (Urgent): Total weight ≥ 85 points, corresponding to "very short remaining time, fully compatible equipment, after-sales / urgent order";

[0085] Level 2 (High Priority): 70 points ≤ Total Weight < 85 points, corresponding to "Tight Remaining Time, Basic Equipment Compatibility, Regular Urgent Orders";

[0086] Level 3 (Regular): 50 points ≤ total weight < 70 points, corresponding to "sufficient remaining time, equipment compatibility, regular orders";

[0087] Level 4 (Low Priority): 30 points ≤ Total Weight < 50 points, corresponding to "Ample remaining time, partially compatible equipment, non-core orders";

[0088] Level 5 (Pending): Total weight < 30 points, corresponding to "trial production orders, standby tasks without a clear deadline".

[0089] Administrators (such as production supervisors) can manually adjust priorities in special scenarios. For example, if there are too many Level 1 tasks (more than 50% of the current number of idle devices), some "non-after-sales Level 1 tasks" can be downgraded to Level 2. Temporarily inserted VIP customer orders can be manually marked as Level 1, and the adjustment record can be synchronously stored in the system log to ensure traceability.

[0090] Add a priority level field to the task attribute table and generate priority tags. Then, synchronize the task data with priority tags to the device cluster load balancing unit and user clients.

[0091] Priority labels (e.g., red for level 1, blue for level 3) are used for visual identification during subsequent scheduling. Task data with priority labels is synchronized to the device cluster load balancing unit as the core basis for task allocation to specific printers; it is also synchronized to the user client to display the current priority of the task and the estimated start time (e.g., "Level 1 task, estimated to be allocated to the device in 10 minutes"), allowing users to track it in real time.

[0092] Device cluster load balancing unit: used to acquire real-time status data of printing devices, allocate weights according to the degree of impact on task execution and calculate a comprehensive suitability score, and allocate tasks to the optimal device based on the load balancing algorithm;

[0093] The execution process of the device cluster load balancing unit includes the following steps:

[0094] The edge gateway node module establishes continuous communication with each UV printer terminal to obtain real-time printer status data, including current task progress data, printhead health data, idle time data, and historical failure rate data.

[0095] The communication protocol uses the lightweight MQTT protocol, with a sampling frequency of 1 time / 3 seconds to ensure the real-time performance of the data.

[0096] Current task progress data: obtained from the task execution and status acquisition unit of the printer terminal, including the completed printing area / total printing area (e.g., if 0.5㎡ has been completed / 2㎡ total, the progress is 25%), the estimated remaining time of the current task (calculated based on the printing speed). If the printing device has no current task, it is marked as idle and the progress value is set to 0%.

[0097] Printhead health data: The data is collected by combining the temperature sensor, camera and historical records of the printer terminal, including the current temperature of the printhead (normal range 40-60℃, if it exceeds the range, the health score will be reduced), the number of abnormal ink droplet ejection in the past hour (such as the number of ink interruptions caused by clogging, 5 points will be deducted from the health score for each abnormality), and the cumulative working time of the printhead (if it has not been maintained for more than 1000 hours, the health score will be reduced to 80 points). Finally, the printhead health score is output as a percentage.

[0098] Idle time data: Calculated in two scenarios. If the printing device is currently idle, the idle time is the difference between the current time and the completion time of the previous task. If the printing device is executing a task, the idle time is the difference between the estimated completion time of the current task and the current time.

[0099] Historical failure rate data: Extracted from the historical database of the cloud dispatch center module, the number of failures (such as paper jams, UV lamp failures, printhead blockages, etc.) and failure repair time of the printing equipment in the past 30 days are statistically analyzed. The failure frequency (number of failures / 30 days, such as 0.5 times / day) and the average failure repair time (such as 2 hours / time) are calculated. The higher the failure frequency and the longer the repair time, the greater the negative impact of this indicator on equipment matching.

[0100] The acquired status data is standardized to eliminate the differences in the units of data from different dimensions. Based on the degree of influence of each status data on task execution, weights are assigned to each data dimension to construct a comprehensive device adaptability scoring model.

[0101] Min-Max standardization is used to convert various data types into standardized scores of 0-100, avoiding weight bias caused by different data units. Different weights are assigned to each status data point based on its impact on task execution. Specifically, printhead health directly affects print quality (e.g., clogging leading to incomplete patterns), making it a core influencing factor, hence it receives 35%; idle time determines task start speed—shorter idle time means less waiting time, improving overall production efficiency, hence it receives 30%; equipment with low failure rates reduces the risk of task interruption and lowers reprint costs, hence historical failure rate receives 20%; current task progress data only affects whether to prioritize equipment nearing completion of the current task—if the current task progress is 90%, shorter tasks can be prioritized, hence current task progress receives 15%.

[0102] The standardized scores of each dimension are multiplied by their corresponding weights and summed to obtain the overall compatibility score for each printer.

[0103] An improved genetic algorithm is adopted, with the addition of two optimization terms: equipment load constraint and task-process matching degree. The optimal task allocation scheme is calculated and output with the goal of minimizing the total completion time of all tasks and minimizing the difference in equipment load.

[0104] The load balancing algorithm is an improved genetic algorithm, and the calculation of the optimal allocation scheme includes the following steps:

[0105] The task-device matching relationship is binary encoded to generate an initial population;

[0106] Suppose there are 5 tasks to be assigned (T1-T5) and 3 printers (D1-D3). Code 100 indicates that the task is assigned to D1, 010 indicates that it is assigned to D2, and 001 indicates that it is assigned to D3. 50 sets of codes are randomly generated as the initial population. Each set of codes corresponds to one task assignment scheme to ensure that the initial scheme covers different combinations of device assignments.

[0107] Using the reciprocal of total completion time and the reciprocal of the standard deviation of equipment load as core indicators, a fitness function is constructed to evaluate the merits of the allocation scheme. The formula is as follows:

[0108] Fitness value = (1 / total completion time) × 0.6 + (1 / standard deviation of equipment load) × 0.4;

[0109] The total completion time is the sum of the execution times of the tasks assigned to each device (calculated by adding the device's idle time to the estimated execution time of the task).

[0110] Equipment load standard deviation = standard deviation of the total time of task allocation for each device (the smaller the standard deviation, the more balanced the equipment load).

[0111] For example, Option A has a total completion time of 120 minutes and a standard deviation of equipment load of 10; Option B has a total completion time of 150 minutes and a standard deviation of equipment load of 5. Then, the fitness value of Option A is (1 / 120) × 0.6 + (1 / 10) × 0.4 ≈ 0.005 + 0.04 = 0.045, and the fitness value of Option B is (1 / 150) × 0.6 + (1 / 5) × 0.4 ≈ 0.004 + 0.08 = 0.084. Option B is better.

[0112] The roulette wheel selection method is used to select the best solutions with higher fitness values ​​according to their proportion to enter the next generation. The selected solutions are then cross-crossed at a single point, and the cross-crossed solutions are then mutated.

[0113] The higher the fitness value, the greater the probability of being selected. In this embodiment, the top 20 fitness values ​​are selected to enter the next generation. A single-point crossover is performed on the selected schemes. For example, two schemes are coded as "T1:D1,T2:D2,T3:D3" and "T1:D2,T2:D1,T3:D3", and crossed at position T2 to obtain the new scheme "T1:D1,T2:D1,T3:D3". A 1% probability mutation is then performed on the crossed schemes (e.g., mutating "T1:D1" to "T1:D2") to avoid the algorithm getting trapped in local optima.

[0114] Repeat the selection, crossover, and mutation operations as described above until the number of iterations reaches a preset value or the fitness value changes by less than 0.001 over 10 consecutive iterations. Output the solution with the highest fitness value as the optimal task allocation solution.

[0115] The optimal allocation scheme for the output is verified for process matching and load limit. After the verification is passed, the task instruction is sent to the target printer through the edge gateway node module, and the task-device allocation relationship is recorded in the cloud scheduling center module.

[0116] Process matching verification: The process attributes from the task parsing stage are retrieved and matched with the device capability library of the printer terminal. For example, if the task requires "glass material + 600DPI + UV soft ink", and the assigned printer only supports metal material, the process is determined to be mismatched, and the algorithm is triggered to recalculate (the device is removed and the optimal solution is regenerated from the remaining devices).

[0117] Load limit verification: Check whether the total duration of tasks assigned to each device exceeds the maximum daily load threshold (preset value, such as 8 hours / day). If the total duration of tasks assigned to a device exceeds the threshold by 9 hours, reduce the number of long-duration tasks assigned to that device by 1 (e.g., adjust the task estimated to take 2 hours to other idle devices).

[0118] After successful verification, the task instructions (including pattern compression files, process parameters, and execution priority) are sent to the target printer through the edge gateway node module. At the same time, the task-device allocation relationship is recorded in the cloud scheduling center module (e.g., "Task T1 is allocated to D2, issuance time 2025-05-20 10:00") for subsequent status tracking.

[0119] During task execution, the device status update data synchronized by the edge gateway node module is received in real time, and the load distribution is dynamically adjusted.

[0120] Device failure trigger adjustment: If a device suddenly fails during task execution (such as a UV lamp failure), the edge gateway node module will report a fault code. The device cluster load balancing unit will immediately mark the unfinished task of the device as "to be reassigned", call the load balancing algorithm to reassign it to the idle device with the highest overall adaptability, and push a task transfer notification to the administrator terminal (such as "task T1 is transferred from D2 to D3 because D2 UV lamp failure").

[0121] Task priority change trigger adjustment: If the cloud scheduling center module receives a high-priority task insertion (such as a level 1 emergency task), the device cluster load balancing unit recalculates the overall adaptability of all devices and prioritizes assigning high-priority tasks to the devices with the shortest idle time and the highest nozzle health. If it is necessary to preempt a device with a low-priority task (such as D3 executing a level 5 task), the low-priority task of D3 is paused first (breakpoint is saved), and the high-priority task is issued. After the high-priority task is completed, the low-priority task is resumed to execute from the breakpoint.

[0122] Network Status Adaptive Scheduling Unit: Used to monitor the network quality between the cloud scheduling center module and the edge gateway node module in real time, and dynamically adjust the data transmission strategy according to the divided network status levels;

[0123] The execution process of the network state adaptive scheduling unit includes the following steps:

[0124] Obtain network quality parameters for the cloud scheduling center module and the edge gateway node module, including latency, packet loss rate, and bandwidth;

[0125] Delay collection: Using the ICMP echo request mechanism, the cloud scheduling center module sends a fixed-size test data packet to the edge gateway node module, and records the round-trip time (RTT) from the sending of the data packet to the receiving edge gateway node module's response. After collecting 10 consecutive times, the average value is taken as the current delay value. If there is no response in a certain collection (timeout > 1000ms), it is marked as an abnormal delay and included in the subsequent packet loss rate statistics.

[0126] Packet loss rate collection: Based on the data packet transmission results during the above delay collection process, the ratio of "number of test data packets without response / total number of data packets sent" is calculated. At the same time, the transmission acknowledgment frames (ACKs) of actual task data between the cloud scheduling center module and the edge gateway node module are monitored. If no ACK is received for 3 consecutive times, it is determined as data packet loss and is added to the packet loss rate statistics to ensure data accuracy.

[0127] Bandwidth Acquisition: Using the sliding window method, a temporary data transmission channel is established between the cloud scheduling center module and the edge gateway node module. A fixed-size test file is transmitted continuously for 10 seconds, and the actual transmission completion time is recorded. The real-time bandwidth is calculated using the formula "actual bandwidth = file size × 8 / transmission time" (e.g., if a 100MB file is transmitted for 10 seconds, the actual bandwidth = 100 × 8 / 10 = 80Mbps). At the same time, the bandwidth usage of currently transmitted task data is monitored to avoid interference with actual business transmission caused by test bandwidth.

[0128] The collected network quality parameters are standardized and preprocessed, and the network status level is determined according to the preset threshold, which is divided into three levels: good, fluctuating and interrupted.

[0129] Based on the preprocessed parameters, the network status is divided into three levels: good, fluctuating, and interrupted, as detailed below:

[0130] Good rating: Latency < 50ms, packet loss rate < 1%, actual bandwidth ≥ 120% of the task's required bandwidth. All three parameters meet the threshold requirements, and the network is judged to be good.

[0131] Fluctuation level: 50ms≤latency≤200ms, 1%≤packet loss rate≤5%, 80% of the bandwidth required by the task≤actual bandwidth<120%. If any parameter exceeds the threshold of the good level but does not reach the standard of the interruption level, it is judged as network fluctuation.

[0132] Interruption level: If the latency is greater than 200ms or there is no response after timeout, the packet loss rate is greater than 5%, the actual bandwidth is less than 80% of the bandwidth required by the task or there is no data transmission, the network is judged to be interrupted if any of the parameters reach the interruption threshold.

[0133] Based on the determined network status level, the corresponding transmission strategy is invoked to ensure data transmission efficiency and reliability;

[0134] Network optimization strategy (latency < 50ms, packet loss rate < 1%): Data transmission adopts real-time synchronous transmission. The cloud scheduling center module directly sends complete task data (including uncompressed high-definition pattern files and all process parameters) to the edge gateway node module, while simultaneously receiving printer status data from the edge gateway node module in real time. The transmission frequency is consistent with the data acquisition frequency (1 second / time). CRC (cyclic redundancy check) is added only when transmitting critical parameters to ensure that core instructions are error-free. No additional redundancy is added to non-critical data, improving transmission efficiency.

[0135] Network fluctuation strategy (50ms ≤ latency ≤ 200ms, 1% ≤ packet loss rate ≤ 5%): The LZ77 compression algorithm is used to compress the task pattern file, reducing the file size to 40% of the original size, thus reducing data transmission volume; only key task parameters are transmitted, omitting unnecessary data. A "segmented transmission + retransmission mechanism" is adopted, dividing the compressed task data into 10MB / segment data packets. After transmitting each segment, an ACK confirmation is awaited from the edge gateway node module. If no ACK is received, the segment is retransmitted (up to 3 times), avoiding the loss of overall task data due to packet loss; the frequency of status data transmission is reduced to 5 seconds / time, reducing network load.

[0136] Network interruption strategy (latency > 200ms or packet loss rate > 5%): The cloud scheduling center module immediately triggers the "edge gateway takeover" command, records the status of currently pending task data and already issued tasks, and synchronously marks them as "interrupted and pending," ceasing active data transmission to the edge gateway node module; upon receiving the takeover command, the edge gateway node module automatically activates local scheduling mode, calls locally cached task data (including complete pattern files and parameter configurations), and executes tasks according to priority. The cloud scheduling center module records in real time the task allocation status before the interruption (e.g., "Task T1 has been issued to the edge gateway node module, but no execution feedback has been received"), network interruption time and reason (e.g., "2025-05-20 11:30, latency 280ms, packet loss rate 7%), and only transmits newly added task data and status update commands during the interruption (incremental synchronization) to the edge gateway node module after network recovery, avoiding duplicate transmission.

[0137] During the execution of the transmission strategy, the effect of the strategy execution is monitored in real time, and the strategy is adjusted based on the feedback data.

[0138] The edge gateway node module feeds back key metrics after policy execution to the cloud scheduling center module, including task data transmission completion time (e.g., transmission time of an 80MB compressed file under a fluctuation policy), number of data retransmissions (e.g., a data packet retransmitted twice under a fluctuation policy), and whether local task execution is normal (e.g., whether the edge gateway node module successfully starts a local task under an interruption policy). The cloud scheduling center module statistically analyzes the feedback data. If the number of data retransmissions under a certain policy is greater than 5 times / hour, or the task transmission completion time exceeds expectations (e.g., transmission time of a 200MB file greater than 30 seconds under a good policy), it is marked as "policy adaptation anomaly". For policy adaptation anomaly scenarios, policy parameters are fine-tuned to adapt to the actual network environment. All optimization adjustments are recorded in the policy log, including the adjustment time, adjustment parameters, and adjustment reasons, facilitating subsequent traceability and policy iteration.

[0139] Energy consumption and consumables optimization unit: Based on historical data, a prediction model is trained to estimate ink consumption and energy consumption according to task parameters, and compared with the real-time consumables balance data to determine whether the consumables are sufficient. If they are insufficient, a multi-level warning is triggered to replenish the stock in time.

[0140] The execution process of the energy consumption and consumables optimization unit includes the following steps:

[0141] Historical data related to energy consumption and consumables are obtained from the historical database of the cloud dispatch center module and the local logs of the UV printer terminal to construct a basic dataset, and the dataset is standardized and preprocessed.

[0142] Task parameter data: Collect process parameters for all printing tasks within the past 12 months, including printing area, color saturation, printing accuracy, material type, and UV lamp power level;

[0143] Consumable consumption data: Collect the actual ink usage for the corresponding task (statistics are compiled separately for CMYK four colors and white ink, such as 15ml of cyan ink and 20ml of black ink), printhead cleaning times and ink consumption for cleaning;

[0144] Energy consumption data: Collect the total power consumption of the corresponding task, and break it down into UV lamp energy consumption (accounting for 60%-70% of the total energy consumption), nozzle drive energy consumption (accounting for 15%-20%), and motor operation energy consumption (accounting for 10%-15%).

[0145] Data is collected sequentially according to task dimensions, with each task corresponding to a complete data record to ensure a one-to-one match between data and tasks. Outlier removal and standardization are performed on the collected data, and the dataset is divided into sub-datasets based on material type, such as metal, glass, and plastic. Each sub-dataset is further subdivided according to printing precision to ensure targeted training of subsequent models.

[0146] Based on the preprocessed historical dataset, a gradient boosting regression (GBR) model is used to train a prediction model to accurately predict ink consumption and energy consumption for new tasks, and the model accuracy is continuously optimized through iteration.

[0147] The model input features (independent variables) are "printing area, color saturation, printing precision, material type, and UV lamp power level". The material type is converted into a numerical feature through one-hot encoding (e.g., metal=100, glass=010, plastic=001). The model output targets (dependent variables) are "CMYK four-color ink usage, white ink usage, and total power consumption". 70% of the historical dataset is used as the training set and 30% as the test set. The gradient boosting algorithm is used to minimize the mean squared error (MSE) between the predicted and actual values. The model hyperparameters (e.g., learning rate, number of decision trees) are iteratively adjusted until the prediction error on the test set is <5% (i.e., the deviation between the estimated ink usage and the actual usage is ≤5%).

[0148] When the cloud scheduling center module receives a new printing task, the energy consumption and consumables optimization unit extracts the task parameters and inputs them into the trained prediction model, outputting the ink consumption and energy consumption prediction results.

[0149] The task parameters are obtained by parsing the task and input into the trained prediction model to obtain the estimated results of energy consumption and consumables. The estimated data is synchronized to the cloud scheduling center module and the edge gateway node module as the basis for printing path optimization and consumable replenishment judgment.

[0150] Consumables estimation: The ink consumption is broken down by color (e.g., cyan 22ml, magenta 20ml, yellow 18ml, black 25ml, white 30ml), the estimated number of cleaning cycles and the ink consumption for cleaning (e.g., 3ml of ink is consumed for the first cleaning before printing), and the total ink consumption for the task is summed up (e.g., 22+20+18+25+30+3=118ml).

[0151] Energy consumption estimate: Break down the energy consumption of UV lamp (e.g., for 3㎡ metal printing, if the UV lamp runs at 80% power for 2 hours, the energy consumption = 0.8kW × 2h = 1.6kWh), the energy consumption of printhead drive (0.3kWh), and the energy consumption of motor operation (0.2kWh), and sum them up to get the total energy consumption (1.6 + 0.3 + 0.2 = 2.1kWh).

[0152] Based on the print pattern size and printer table coordinates of the task, the optimal printhead movement path is planned by combining the contour offset algorithm with path sorting optimization.

[0153] Image recognition is performed on the pattern file for the task to extract the outer contour coordinates of the pattern (e.g., the coordinates of the four vertices of a rectangular pattern), and blank areas (areas not requiring printing) inside the pattern are marked to prevent the printhead from moving aimlessly in these areas. A contour offset algorithm is used to generate contour offset paths, i.e., generating printing paths layer by layer from the outer contour of the pattern inwards according to the printhead width, ensuring that the path covers the entire printing area and that the distance between adjacent paths is equal to the printhead width, avoiding duplicate printing or omissions. A greedy algorithm is used to sort the generated printing paths, prioritizing the printing order of adjacent areas to reduce long-distance printhead movement in non-printing areas. For example, if the pattern contains three independent printing areas A, B, and C, and A is adjacent to B and B is adjacent to C, then the planned path is "A→B→C" instead of "A→C→B", reducing the long-distance aimless movement from A to C. The optimized path coordinates (including the printing path and necessary aimless movement paths) are converted into G-code that the printer can recognize and sent to the task execution unit of the UV printer terminal to control the printhead to move along the optimized path, reducing aimless movement distance and lowering unnecessary energy consumption and printhead wear.

[0154] The system acquires real-time data on the remaining consumables of the UV printer terminal, compares it with the estimated ink consumption, and determines whether the consumables are sufficient. If they are insufficient, it triggers multi-level warnings to replenish the stock in a timely manner.

[0155] The ink tank weight sensor (or level sensor) on the UV printer terminal collects the current remaining ink level of each color in real time and synchronizes it to the energy consumption and consumables optimization unit via the edge gateway node module. The current remaining ink level of each color is compared with the estimated ink consumption for the new task, and the following calculation is made: "Ink Level Adequacy Ratio = (Current Remaining Ink Level - Estimated Ink Consumption) / Estimated Ink Consumption × 100%". If the adequacy ratio > 20%, the consumables are considered sufficient, and no warning is needed; if 0% ≤ adequacy ratio ≤ 20%, the consumables are considered critically insufficient, requiring attention; if the adequacy ratio < 0%, the consumables are considered absolutely insufficient and cannot meet the current task's requirements, necessitating immediate replenishment.

[0156] When the consumables are deemed insufficient, multi-level alerts are triggered, and all alert information (including trigger time, alert level, and processing result) is recorded in the cloud logs for easy tracking of consumable usage cycles and replenishment efficiency.

[0157] Level 1 Warning (Adequacy < 0%): Immediately push a red warning message to the administrator terminal (PC / mobile APP), including "Task ID, missing ink color, current remaining ink, estimated amount of ink to be replenished (e.g., white ink needs to be replenished 20ml)", and at the same time suspend the issuance of the current task until the consumables are replenished;

[0158] Level 2 Warning (0%≤Adequacy Rate≤20%): A yellow alert message is pushed to the administrator terminal, which includes the message "Consumables are about to run out, it is recommended to replenish within 24 hours" and the ink tank is marked as "Pending Replenishment" in the consumables management interface of the cloud dispatch center.

[0159] The edge gateway node module includes a local task caching and resume unit, a device collaborative control unit, and a fault diagnosis and early warning unit.

[0160] Local task caching and resume unit: used to cache task data sent by the cloud scheduling center module. When the network is interrupted, the local scheduling mode is automatically activated and the cached tasks are executed according to priority. After the network is restored, the local execution records are synchronized to the cloud scheduling center module.

[0161] Device collaboration control unit: used to break down large tasks into multiple sub-tasks and control multiple printers in parallel for local collaboration;

[0162] For example, a 3m×6m glass printing task can be split into three 2m×3m sub-tasks and assigned to three adjacent printers for simultaneous printing. The edge gateway calibrates the printing coordinates of each device in real time (error ≤0.1mm) to ensure splicing accuracy.

[0163] Fault diagnosis and early warning unit: Based on real-time data collected by printer sensors, it identifies abnormal states through edge computing and triggers local early warnings, while pushing fault codes to the cloud dispatch center module.

[0164] The UV printer terminal continuously collects core data related to its operating status through hardware sensors, including printhead temperature (collected by a temperature sensor, range 0-100℃, accuracy ±0.5℃), UV lamp power (collected by a power sensor, range 0-100% rated power, accuracy ±1%), and motor current (collected by a current sensor, range 0-5A, accuracy ±0.01A). It also correlates with auxiliary data such as droplet ejection frequency and printing speed. The collected data undergoes standardized preprocessing to eliminate noise interference.

[0165] Two key features are extracted: one is "real-time absolute value features" (such as current printhead temperature 55℃, UV lamp power 80%), and the other is "trend change features" (such as printhead temperature rise rate within 5 seconds, motor current fluctuation amplitude within 10 seconds). Among them, the frequency fluctuation value is calculated for the ink droplet ejection frequency: frequency fluctuation value = (current frequency - historical average frequency) / historical average frequency × 100%, which is used to identify early signs of printhead clogging (fluctuation > 5% is marked as a high-risk feature).

[0166] The edge gateway node module deploys a lightweight CNN model (based on MobileNet architecture compression, parameter size < 5MB, inference time < 0.1 seconds) to infer the preprocessed feature data and determine whether the device is in an abnormal state. Specifically, the extracted "real-time absolute value features + trend change features" are combined into a 128-dimensional feature vector and input into the CNN model. The model outputs three status labels: "normal", "minor abnormality", and "serious abnormality", along with the corresponding fault type probabilities (e.g., "printer head clogging: 92%", "UV lamp aging: 5%"). If the model outputs a "normal" label and the probability of all fault types is < 10%, the device is considered to be operating normally. If it outputs "minor abnormality" (e.g., ink droplet ejection frequency fluctuation of 4.5%, printer head clogging probability of 35%), it is marked as "to be observed" and data changes are continuously tracked. If it outputs "serious abnormality" (e.g., printer head temperature of 75℃, exceeding the normal range of 40-60℃, fault probability > 80%), or if key parameters trigger preset thresholds (e.g., UV lamp power < 30% of rated power), it is directly determined to be in a fault state. Based on the anomaly identification results, corresponding local alerts are triggered and synchronized with the cloud to ensure that the faults are handled in a timely manner. After the alert is triggered, fault-related information is automatically recorded and subsequent traceability is supported.

[0167] The UV printer terminal module includes a task execution and status acquisition unit, an offline autonomous unit, and a human-computer interaction unit;

[0168] Task execution and status acquisition unit: Used to receive task instructions from edge gateway node module or cloud scheduling center module, control nozzle movement and UV lamp curing, and collect real-time status data through sensors;

[0169] Offline Autonomous Unit: It has a built-in local storage module to cache the data of the most recent 100 tasks. When the network between the edge gateway node module and the UV printer terminal is interrupted, it automatically starts the preset rules.

[0170] The preset rules are as follows: prioritize the execution of cached level 1-2 emergency tasks; if the task is interrupted during execution, record the breakpoint coordinates and resume printing from the breakpoint after the network is restored; save the task log locally (including printing time, consumable usage, and fault records) and continue uploading it after the network is restored;

[0171] Human-machine interaction unit: Equipped with a touch screen and physical buttons, it supports local modification of task parameters and displays task progress, equipment status and fault prompts, facilitating intervention by on-site operators.

[0172] System workflow: Users upload print jobs via client (PC / mobile phone), the cloud scheduling center module parses the jobs and assigns priorities; the cloud scheduling center module allocates jobs to target printers based on device load and network status, and distributes job data (including compressed pattern files and parameters) through the edge gateway node module; the edge gateway node module and the printer terminal cache job data respectively to ensure local execution in case of network interruption; after receiving the job instruction, the printer starts printing according to the job instruction, while simultaneously collecting status data in real time and feeding it back to the edge gateway node module, which then synchronizes the data to the cloud scheduling center module.

[0173] Furthermore, during data transmission, if the network is interrupted, the edge gateway node module takes over the scheduling, the printer executes local cached tasks, and records the breakpoint; if the device malfunctions, the edge gateway node module triggers a local alert, and the cloud scheduling center reassigns the task to a backup device. After the task is completed, the printer reports the task results (including the finished product image and consumable usage), the cloud scheduling center generates a production report, and the edge gateway node module clears the cache of completed tasks.

[0174] On the other hand, please see Figure 2 As shown, this invention provides an intelligent control method for UV printers based on remote task scheduling, comprising the following steps:

[0175] Get the print jobs uploaded by users, perform structured parsing on the jobs, extract the job parameters, and assign priorities to the jobs;

[0176] Real-time monitoring of network quality parameters, classification of network status levels based on thresholds, and matching of corresponding execution strategies based on different levels. The network quality parameters include latency, packet loss rate, and bandwidth.

[0177] Obtain real-time status parameters of printing devices, calculate the comprehensive suitability score of each device, and allocate tasks to the optimal device based on a load balancing algorithm;

[0178] After receiving the task instruction, the device executes the task and provides real-time feedback on the remaining consumables data of the UV printer terminal. It compares the remaining consumables data with the estimated ink consumption output by the prediction model to determine whether replenishment is needed and issues an alert.

[0179] The above embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A UV printer intelligent control system based on remote task scheduling, characterized in that, This includes a cloud-based dispatch center module for communication connectivity, an edge gateway node module, and a UV printer terminal module; The cloud-based scheduling center module is used for global task planning and allocation, network status monitoring, device status monitoring, and data storage and analysis. The cloud-based scheduling center module includes a task parsing and priority sorting unit, a device cluster load balancing unit, a network status adaptive scheduling unit, and an energy consumption and consumables optimization unit. Task parsing and priority sorting unit: used to receive printing tasks uploaded by users, parse task attributes and assign priorities to tasks according to preset rules, and synchronize the prioritized tasks to the device cluster load balancing unit and user clients. Device cluster load balancing unit: used to acquire real-time status data of printing devices, allocate weights according to the degree of impact on task execution and calculate a comprehensive suitability score, and allocate tasks to the optimal device based on the load balancing algorithm; Network Status Adaptive Scheduling Unit: Used to monitor the network quality between the cloud scheduling center module and the edge gateway node module in real time, and dynamically adjust the data transmission strategy according to the divided network status levels; Energy consumption and consumables optimization unit: Based on historical data, a prediction model is trained to estimate ink consumption and energy consumption according to task parameters, and compared with the real-time consumables balance data to determine whether the consumables are sufficient. If they are insufficient, a multi-level warning is triggered to replenish the stock in time. The edge gateway node module is used to cache task data issued by the cloud scheduling center module, serve as a local scheduling hub when the network is interrupted, and simultaneously perform coordinated control and fault diagnosis and early warning for multiple devices. The UV printer terminal module is used to collect the status data of the printing device in real time, receive and execute tasks, cache data locally, and interact with the human-machine interface. The execution process of the device cluster load balancing unit includes the following steps: The edge gateway node module establishes continuous communication with each UV printer terminal to obtain real-time printer status data, including current task progress data, printhead health data, idle time data, and historical failure rate data. The acquired status data is standardized to eliminate the differences in the units of data from different dimensions. Based on the degree of influence of each status data on task execution, weights are assigned to each data dimension to construct a comprehensive device adaptability scoring model. The standardized scores of each dimension are multiplied by their corresponding weights and summed to obtain the overall compatibility score for each printer. An improved genetic algorithm is adopted, with the addition of two optimization terms: equipment load constraint and task-process matching degree. The optimal task allocation scheme is calculated and output with the goal of minimizing the total completion time of all tasks and minimizing the difference in equipment load. The optimal allocation scheme for the output is verified for process matching and load limit. After the verification is passed, the task instruction is sent to the target printer through the edge gateway node module, and the task-device allocation relationship is recorded in the cloud scheduling center module. During task execution, the device status update data synchronized by the edge gateway node module is received in real time, and the load distribution is dynamically adjusted. The load balancing algorithm is an improved genetic algorithm, and the calculation of the optimal allocation scheme includes the following steps: The task-device matching relationship is binary encoded to generate an initial population; Using the reciprocal of total completion time and the reciprocal of the standard deviation of equipment load as core indicators, a fitness function is constructed to evaluate the merits of the allocation scheme; The roulette wheel selection method is used to select the best solutions with higher fitness values ​​according to their proportion to enter the next generation. The selected solutions are then cross-crossed at a single point, and the cross-crossed solutions are then mutated. Repeat the selection, crossover, and mutation operations as described above until the number of iterations reaches a preset value or the fitness value changes by less than 0.001 over 10 consecutive iterations. Output the solution with the highest fitness value as the optimal task allocation solution.

2. The intelligent control system for UV printers based on remote task scheduling according to claim 1, characterized in that, The execution process of the task parsing and priority sorting unit includes the following steps: The system automatically generates a unique task ID for print jobs uploaded by users through the client and verifies the validity of the uploaded tasks. By using structured parsing algorithms, key attributes are extracted from the validated task data to form a standardized task attribute table; A priority determination rule base is set up, with each rule corresponding to an independent weight calculation logic. The rules include time urgency rules, equipment adaptability rules, and production demand rules. The total task weight is obtained by integrating the weights of various rules, and the task priority is divided based on the total task weight, with the option for manual adjustment. Add a priority level field to the task attribute table and generate priority tags. Then, synchronize the task data with priority tags to the device cluster load balancing unit and user clients.

3. The intelligent control system for UV printers based on remote task scheduling according to claim 2, characterized in that, The key attributes include basic attributes, process attributes, time attributes, and resource attributes; the basic attributes include task ID, upload time, and user information; the process attributes include material type, printing precision, and color mode; the time attributes include deadline and expected completion time; and the resource attributes include printing area and estimated ink type.

4. The intelligent control system for UV printers based on remote task scheduling according to claim 1, characterized in that, The execution process of the network state adaptive scheduling unit includes the following steps: Obtain network quality parameters for the cloud scheduling center module and the edge gateway node module, including latency, packet loss rate, and bandwidth; The collected network quality parameters are standardized and preprocessed, and the network status level is determined according to the preset threshold, which is divided into three levels: good, fluctuating and interrupted. Based on the determined network status level, the corresponding transmission strategy is invoked to ensure data transmission efficiency and reliability; During the execution of the transmission strategy, the effect of the strategy execution is monitored in real time, and the strategy is adjusted based on the feedback data.

5. The intelligent control system for a UV printer based on remote task scheduling according to claim 1, characterized in that, The execution process of the energy consumption and consumables optimization unit includes the following steps: Historical data related to energy consumption and consumables are collected in batches from the historical database of the cloud dispatch center module and the local logs of the UV printer terminal to build a basic dataset, and the dataset is standardized and preprocessed. Based on the preprocessed historical dataset, a gradient boosting regression model is used to train a prediction model to accurately predict ink consumption and energy consumption for new tasks, and the model accuracy is continuously optimized through iteration. When the cloud scheduling center module receives a new printing task, the energy consumption and consumables optimization unit extracts the task parameters and inputs them into the trained prediction model, outputting the ink consumption and energy consumption prediction results. Based on the print pattern size and printer table coordinates of the task, the optimal printhead movement path is planned by combining the contour offset algorithm with path sorting optimization. The system acquires real-time data on the remaining consumables of the UV printer terminal, compares it with the estimated ink consumption, and determines whether the consumables are sufficient. If they are insufficient, it triggers multi-level warnings to replenish the stock in a timely manner.

6. The intelligent control system for a UV printer based on remote task scheduling according to claim 1, characterized in that, The edge gateway node module includes a local task caching and resume unit, a device collaborative control unit, and a fault diagnosis and early warning unit. Local task caching and resume unit: used to cache task data sent by the cloud scheduling center module. When the network is interrupted, the local scheduling mode is automatically activated and the cached tasks are executed according to priority. After the network is restored, the local execution records are synchronized to the cloud scheduling center module. Device collaboration control unit: used to break down large tasks into multiple sub-tasks and control multiple printers in parallel for local collaboration; Fault diagnosis and early warning unit: Based on real-time data collected by printer sensors, it identifies abnormal states through edge computing and triggers local early warnings, while pushing fault codes to the cloud dispatch center module.

7. The intelligent control system for a UV printer based on remote task scheduling according to claim 1, characterized in that, The UV printer terminal module includes a task execution and status acquisition unit, an offline autonomous unit, and a human-computer interaction unit; Task execution and status acquisition unit: Used to receive task instructions from edge gateway node module or cloud scheduling center module, control nozzle movement and UV lamp curing, and collect real-time status data through sensors; Offline Autonomous Unit: It has a built-in local storage module to cache the data of the most recent 100 tasks. When the network between the edge gateway node module and the UV printer terminal is interrupted, it automatically starts the preset rules. The preset rule is as follows: prioritize the execution of cached level 1-2 emergency tasks; if the task execution is interrupted, record the breakpoint coordinates and resume from the breakpoint after the network is restored; Save the task log locally and continue uploading it once the network is restored; Human-machine interaction unit: Equipped with a touch screen and physical buttons, it supports local modification of task parameters and displays task progress, equipment status and fault prompts, facilitating intervention by on-site operators.

8. A UV printer intelligent control method based on remote task scheduling, applied to the UV printer intelligent control system based on remote task scheduling as described in any one of claims 1-7, characterized in that, Includes the following steps: Get the print jobs uploaded by users, perform structured parsing on the jobs, extract the job parameters, and assign priorities to the jobs; Real-time monitoring of network quality parameters, classification of network status levels based on thresholds, and matching of corresponding execution strategies based on different levels. The network quality parameters include latency, packet loss rate, and bandwidth. Obtain real-time status parameters of printing devices, calculate the comprehensive suitability score of each device, and allocate tasks to the optimal device based on a load balancing algorithm; After receiving the task instruction, the device executes the task and provides real-time feedback on the remaining consumables data of the UV printer terminal. It compares the remaining consumables data with the estimated ink consumption output by the prediction model to determine whether replenishment is needed and issues an alert.

Images