POD printing monitoring method and system based on load prediction

Abstract

The invention discloses a POD printing monitoring method and system based on load prediction. The method comprises the following steps: constructing a hot folder of a POD printer and performing real-time data monitoring to obtain printing file data; load prediction is carried out on the POD printer according to the data, and a load characteristic spectrum is generated; and according to the atlas, the printing files in the hot folder are processed, and corresponding printing strategies are formulated under different loads. And after the printing strategy is executed, the system monitors the printing process, carries out fault analysis and evaluates the fault-tolerant capability of the POD printer. And according to the fault-tolerant capability, a fault-tolerant compensation scheme is generated and implemented, and the stability of the printing process is ensured. Through load prediction and real-time monitoring, potential problems can be recognized in advance, the printing strategy is optimized, the fault risk is reduced, and the printing efficiency and stability are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of print monitoring, in particular to a POD print monitoring method and system based on load prediction. BACKGROUND

[0002] With the rapid development of digital printing technology, POD (Print On Demand) printing technology has been widely applied in book publishing, enterprise document printing, nautical chart printing and other fields. POD printers have the characteristics of high efficiency, flexibility and customization, and can produce printed matter in real time according to customer needs. However, with the increase in the amount of printing tasks and the diversification of printing task types, POD printers face many challenges in actual application, especially in load management, fault prevention and fault tolerance.

[0003] Traditional POD print monitoring methods are usually based on static load management models or simple fault monitoring mechanisms. These methods often lack the ability to predict changes in printing task loads and cannot achieve real-time load adjustment and effective fault prevention. Existing technologies do not fully consider the adjustment of printing strategies under different load conditions when dealing with complex printing tasks, which can easily affect the printing quality, efficiency and system stability, and even cause system failure or loss of printing tasks. In addition, existing fault diagnosis and fault tolerance mechanisms rely heavily on manual intervention and cannot achieve automated fault analysis and real-time fault tolerance compensation, limiting the autonomous operation capability and efficiency of POD printers.

[0004] Therefore, how to realize dynamic monitoring and adjustment of POD printers based on load prediction, identify potential printing problems in advance, and automatically compensate for faults when they occur, is a major problem in current technology. To solve this problem, the present application proposes a POD print monitoring method and system based on load prediction, which effectively improves the working efficiency, stability and fault tolerance of POD printers through the combination of real-time load prediction, fault monitoring and fault tolerance compensation mechanisms. SUMMARY

[0005] To solve at least one of the above technical problems, the present application proposes a POD print monitoring method and system based on load prediction.

[0006] The first aspect of the present application provides a POD print monitoring method based on load prediction, comprising: constructing a hot folder of a POD printer, performing data monitoring operations on the hot folder, and obtaining printing file data of the hot folder in real time; performing load prediction on the POD printer according to the printing file data, and obtaining a load feature map; According to the load feature map, the print file data in the hot folder is printed, and a print strategy of a POD printer under different loads is constructed. The print strategy is executed, monitored, and analyzed for fault of the POD printer, the fault tolerance capability of the POD printer is determined, the fault tolerance compensation scheme of the POD printer is determined according to the fault tolerance capability.

[0007] In the scheme, the hot folder of the POD printer is constructed, and the data monitoring operation is performed on the hot folder to obtain the print file data of the hot folder in real time, specifically: The historical print task peak value data of the POD printer is obtained, the initial capacity of the hot folder is divided in the local storage device according to the historical print task peak value data, and the hot folder of the POD printer is constructed according to the initial capacity; A storage pressure prediction model of the hot folder is constructed, the residual storage space, file writing rate, and file quantity increment of the hot folder are obtained according to the storage pressure prediction model, and the historical storage pressure data of the same time period in a historical preset time period is obtained, including storage space occupation peak value, file writing rate fluctuation curve, and expansion operation record; The historical storage pressure features of the historical storage pressure data are extracted according to a sliding time window algorithm, including storage space consumption rate and writing rate fluctuation; The historical storage pressure features and the residual storage space, file writing rate, and file quantity increment of the hot folder are imported into the storage pressure prediction model, the current residual storage space of the hot folder is predicted by exponential smoothing method, and the storage space depletion time point of the hot folder is determined; The expansion time point of the hot folder is determined according to the storage space depletion time point, the expansion capacity of the hot folder in a preset time period is determined according to the file quantity increment, and the capacity expansion scheme of the hot folder is obtained; According to the capacity expansion scheme, the hot folder is expanded, and the data monitoring operation on the hot folder is performed in real time, the hot folder is abnormally intercepted, and the print file data of the hot folder is obtained.

[0008] In the scheme, the data monitoring operation on the hot folder is performed in real time, and the hot folder is abnormally intercepted, specifically: A multi-thread file data monitoring service is deployed at a data input port of a hot folder, a file writing event is captured in real time by a data monitoring thread of the hot folder, attribute characteristic data of a currently written file is extracted, including file format identifier, byte size, hash fingerprint and writing time interval; the attribute characteristic data is matched with a preset interception rule library, the interception rule library includes format type whitelist, single file capacity threshold, hash blacklist and time window writing frequency threshold; When it is detected that the writing file format identifier is not in the format type whitelist, the writing file is intercepted, the intercepted file is transferred to a sandbox environment for format analysis, if a valid print instruction field is extracted after analysis, a standardized print file is regenerated according to the instruction field and returned to the hot folder, otherwise it is marked as a format exception file and deleted, obtaining a first interception strategy; If the byte size of the current writing file exceeds the single file capacity threshold, the writing file is split into multiple sub-files according to a preset size and a sharding print order identifier is attached, the writing delay time window of the split sub-file is calculated according to the number of current unprocessed tasks in the hot folder, and the sub-file is injected into the hot folder in batches according to the delay time window, obtaining a second interception strategy; When it is detected that the writing frequency of the same hash fingerprint within a preset time window exceeds the writing frequency threshold or exists in the hash blacklist, the writing process of the writing file is frozen and a repeated task alert is generated, based on the task rollback record of the hash fingerprint in the historical interception log, the risk level of the file is evaluated, if the risk level exceeds a preset threshold, the hash fingerprint is permanently added to the hash blacklist, otherwise the writing frequency is temporarily limited and a temporary interception lock is generated, obtaining a third interception strategy.

[0009] In the scheme, the POD printer is predicted according to the print file data to obtain a load characteristic map, specifically: The print file data is monitored in real time, a print file queue is determined according to the upload timestamp of the print file, and file attribute characteristics of the print file queue are obtained, the file attribute characteristics including file page number, color mode, medium type and print quality parameter; The historical print log data of the POD printer in a preset time period is obtained, historical file attribute characteristics and task time consumption data of the historical print log data are extracted, the historical file attribute characteristics and the task time consumption data are analyzed for correlation, the influence of different file attribute characteristics on the POD printing time consumption is determined, and influence data is obtained; obtain historical print file queue change data of a second preset time period, analyze the historical print file queue change data according to the influence data, determine task time consumption change data of the POD printer in the second preset time period, and determine POD printer load change data in the second preset time period according to the task time consumption change data; input the POD printer load change data into an LSTM-based load prediction model for training, the LSTM adopts a gating mechanism to capture the time sequence dependence of the printer load change, and extracts load fluctuation features of different time scales through stacked bidirectional LSTM layers; obtain print file queue change data of a current preset time period, determine task time consumption change of the print file queue change data of the current preset time period according to the influence data, and determine current load change data; input the current load change data into the trained load prediction model, predict the load in a future preset time period, and obtain a load change prediction result; perform clustering analysis on the load intensity according to the prediction result, identify peak load mode, stable load mode and idle load mode, perform graph modeling on file attribute features and time distribution features corresponding to different load modes, and generate a load feature graph containing a different time load intensity heat map.

[0010] In this scheme, the print file data in the hot folder is printed according to the load feature graph, and a print strategy of the POD printer under different loads is constructed, specifically as follows: obtain print failure occurrence data of the POD printer under different print durations in each load mode, determine failure probability distribution data under different print durations in each load mode according to the failure occurrence data, and establish an association model of print duration and failure probability in each load mode according to the failure probability distribution data; extract a print task queue corresponding to the current load intensity from the load feature graph, obtain task attributes and estimated print time consumption of each print file in the print task queue, and determine a continuous print duration of the POD printer for the print task queue according to the estimated print time consumption; set a preset print failure safety threshold, input the print failure safety threshold into the association model for matching, and determine a safe continuous print duration of the POD; if the continuous print duration exceeds the safe continuous print duration, divide the print files in the print task queue into a plurality of print groups within the safe continuous print duration according to the estimated print time consumption of each print file in the print task queue; When the printing of a print group is completed, the working temperature data of the POD printer and the optimal working temperature range are obtained, the required time length for the working temperature of the printer to recover to the optimal working temperature range is calculated according to the working temperature data, and the printing interval time length of the print group is obtained; The print group and the printing interval time length are constructed into a print strategy in a future preset time period.

[0011] In the scheme, the print strategy is executed, a monitoring operation is performed on the print strategy, a fault analysis is performed on the POD printer, a fault tolerance capability of the POD printer is determined, a fault tolerance compensation scheme of the POD printer is determined according to the fault tolerance capability, and specifically: The print strategy is executed, and the printing state data of the POD printer, including the printing head temperature fluctuation value, the paper transmission rate deviation value and the abnormal increment of ink consumption, is collected in real time, the printing state data is input into a preset fault prediction model, and the real-time fault probability of the current printing task within the remaining printing time length is calculated; When it is detected that the real-time fault probability exceeds a preset fault warning threshold, the remaining number of pages of the current printing task and the hash check value of the printed pages are obtained, a redundant printing channel matched with the current printing task is constructed according to the file fragment features of the unexecuted tasks in the printing task queue, and the remaining number of pages and the unexecuted task fragments are synchronously distributed to the redundant printing channel; A virtual printing queue parallel to the main printing channel is deployed in the redundant printing channel, the page generation logs of the main printing channel and the redundant printing channel are compared in real time, when the number of continuous inconsistencies between the page hash check values of the main printing channel and the redundant printing channel exceeds a preset fault tolerance threshold, the current task of the main printing channel is frozen, the printing control right is switched to the redundant printing channel, and a compensation printing task is regenerated by extracting the undamaged fragments from the virtual printing queue according to the fragment identifiers corresponding to the fault pages of the main printing channel, to obtain a fault tolerance printing scheme.

[0012] The second aspect of the present application also provides a POD printing monitoring system based on load prediction, which comprises a memory and a processor, the memory comprises a POD printing monitoring method program based on load prediction, and the POD printing monitoring method program based on load prediction is executed by the processor to realize the following steps: A hot folder of the POD printer is constructed, a data monitoring operation is performed on the hot folder, and the printing file data of the hot folder is obtained in real time; The POD printer is load predicted according to the printing file data, and a load feature map is obtained; The printing file data in the hot folder is printed according to the load feature map, and a print strategy of the POD printer under different loads is constructed. execute the printing strategy, monitor the printing strategy, analyze the fault of the POD printer, determine the fault tolerance capability of the POD printer, and determine a fault tolerance compensation scheme of the POD printer according to the fault tolerance capability.

[0013] The application discloses a POD printing monitoring method and system based on load prediction. The method obtains printing file data by constructing a hot folder of a POD printer and real-time data monitoring; then, the POD printer is subjected to load prediction according to the data, and a load feature map is generated. According to the map, the printing files in the hot folder are processed, and corresponding printing strategies are formulated under different loads. After the printing strategies are executed, the system monitors the printing process, analyzes the fault, and evaluates the fault tolerance capability of the POD printer. According to the fault tolerance capability, a fault tolerance compensation scheme is generated and implemented, so as to ensure the stability of the printing process. Through load prediction and real-time monitoring, the application can identify potential problems in advance, optimize the printing strategy, reduce the fault risk, and improve the printing efficiency and stability. BRIEF DESCRIPTION OF DRAWINGS

[0014] Figure 1 A flow chart of the POD printing monitoring method based on load prediction is shown. Figure 2 A flow chart of constructing the printing strategy of the POD printer is shown. Figure 3 A flow chart of determining the fault tolerance compensation scheme of the POD printer is shown. Figure 4 A block diagram of the POD printing monitoring system based on load prediction is shown. DETAILED DESCRIPTION

[0015] In order to more clearly understand the above-mentioned purposes, features and advantages of the application, the application will be further described in detail below with reference to the drawings and specific embodiments. It should be noted that the embodiments of the application and the features in the embodiments can be combined with each other without conflict.

[0016] In the following description, many specific details are set forth in order to provide a thorough understanding of the application, but the application can also be practiced without the other ways different from those described herein, therefore, the scope of protection of the application is not limited by the specific embodiments disclosed below.

[0017] Figure 1 A flow chart of the POD printing monitoring method based on load prediction is shown.

[0018] As shown in Figure 1 The first aspect of the application provides a POD printing monitoring method based on load prediction, which comprises: S102, a hot folder of the POD printer is constructed, a data monitoring operation is performed on the hot folder, and print file data of the hot folder is acquired in real time; S104, load prediction is performed on the POD printer according to the print file data, and a load feature map is obtained; S106, print processing is performed on the print file data in the hot folder according to the load feature map, and a print strategy of the POD printer under different loads is constructed; S108, the print strategy is executed, a monitoring operation is performed on the print strategy, fault analysis is performed on the POD printer, fault tolerance capability of the POD printer is determined, and a fault tolerance compensation scheme of the POD printer is determined according to the fault tolerance capability.

[0019] It should be noted that by dynamically constructing a hot folder to monitor storage pressure in real time, predicting expansion requirements and deploying a multi-level abnormal interception mechanism, storage overflow and malicious attacks can be effectively prevented. The load prediction technology based on the LSTM model analyzes the correlation between file attributes and historical task time consumption, generates a load intensity heat map to accurately predict the load trend of the equipment, and supports dynamic scheduling of the task queue. In combination with the fault probability model under different load modes, continuous tasks are split into grouped printing and inserted into temperature control intervals, which reduces the risk of equipment overheating while ensuring efficiency and prolongs the service life of core components. In the execution stage, print status data is collected in real time, and through the redundant channel parallel verification and fragmentation compensation mechanism, seamless switching and uninterrupted task printing are realized when the main channel fails, which significantly reduces material waste. Finally, a full-link closed-loop management from storage optimization, load prediction, strategy execution to fault tolerance compensation is formed, and the stability of the equipment and the resource utilization rate are comprehensively improved.

[0020] According to an embodiment of the present application, the hot folder of the POD printer is constructed, the data monitoring operation is performed on the hot folder, and the print file data of the hot folder is acquired in real time, specifically: The historical print task peak data of the POD printer is acquired, the initial capacity of the hot folder is divided in the local storage device according to the historical print task peak data, and the hot folder of the POD printer is constructed according to the initial capacity; A storage pressure prediction model of the hot folder is constructed, the residual storage space, file write rate and file quantity increment of the hot folder are acquired according to the storage pressure prediction model, and historical storage pressure data of the same time period in a historical preset time period is also acquired, including storage space occupation peak value, file write rate fluctuation curve and expansion operation record; The historical storage pressure characteristics of the historical storage pressure data are extracted according to a sliding time window algorithm, including storage space consumption rate and write-out rate fluctuation; The historical storage pressure feature and the remaining storage space, file writing rate and file quantity increment of the hot folder are imported into a storage pressure prediction model, the current remaining storage space of the hot folder is predicted by exponential smoothing method, and a storage space depletion time point of the hot folder is determined; According to the storage space depletion time point, a capacity expansion time point of the hot folder is determined, and according to the file quantity increment, an expanded capacity of the hot folder in a preset time period is determined, so that a capacity expansion scheme of the hot folder is obtained; According to the capacity expansion scheme, the hot folder is expanded, and data monitoring operation is performed on the hot folder in real time, the hot folder is abnormally intercepted, and print file data of the hot folder is obtained.

[0021] It should be noted that the hot folder management of the traditional POD printing system adopts a fixed capacity or passive expansion mechanism, which is difficult to cope with sudden task loads, often causes task interruption due to insufficient storage space, and has obvious expansion lag (such as responding only when the storage threshold is triggered), causing task accumulation during peak periods. In addition, the time sequence characteristics of historical storage pressure data are not effectively mined, the correlation analysis of storage consumption rate and writing behavior is missing, the expansion time and capacity planning lack scientific basis, and the influx of abnormal files (such as format errors, repeated submission or malicious large files) aggravates the storage pressure, further reducing the system stability. Therefore, the present application dynamically constructs a storage pressure prediction model, initializes the hot folder capacity based on historical task peak data, extracts historical storage pressure features (such as space consumption rate and writing fluctuation) by using a sliding time window algorithm, and uses exponential smoothing method to predict the decay of real-time remaining storage space, accurately calculates the storage depletion time point, actively plans the expansion time and capacity, and breaks through the lag of traditional passive expansion. At the same time, after expansion, the hot folder is monitored in real time, abnormal files are intercepted through multi-dimensional rules such as format whitelist, capacity threshold splitting and hash blacklist, and invalid tasks are prevented from occupying resources. This scheme realizes dynamic optimization of storage resources and active control of abnormal writing, significantly reduces the risk of task interruption, and improves the system resource utilization and task processing stability in high-concurrency scenarios.

[0022] According to the present application, the data monitoring operation on the hot folder is performed in real time, and the hot folder is abnormally intercepted, specifically: A multi-thread file data monitoring service is deployed at a data transmission input port of the hot folder, a file writing event is captured in real time through a data monitoring thread of the hot folder, attribute feature data of the current writing file is extracted, including file format identifier, byte size, hash fingerprint and writing time interval; the attribute feature data is matched with a preset interception rule library, and the interception rule library includes a format type whitelist, a single file capacity threshold, a hash blacklist and a time window writing frequency threshold; When it is detected that the write file format identifier is not in the format type whitelist, the write file is intercepted, the intercepted file is transferred to a sandbox environment for format analysis, if a valid print instruction field is extracted after analysis, a standardized print file is regenerated according to the instruction field and returned to the hot folder, otherwise it is marked as a format exception file and deleted, obtaining a first interception strategy; If the byte size of the current write file exceeds the single file capacity threshold, the write file is split into multiple sub-files according to the preset size and the sub-file print order identifier is attached, the write delay time window of the split sub-file is calculated according to the number of current unprocessed tasks in the hot folder, and the sub-file is injected into the hot folder in batches according to the delay time window, obtaining a second interception strategy; When it is detected that the write frequency of the same hash fingerprint within a preset time window exceeds the write frequency threshold or exists in the hash blacklist, the write process of the write file is frozen and a repeated task alert is generated, the risk level of the file is evaluated based on the task rollback record of the hash fingerprint in the historical interception log, if the risk level exceeds the preset threshold, the hash fingerprint is permanently added to the hash blacklist, otherwise the write frequency is temporarily limited and a temporary interception lock is generated, obtaining a third interception strategy.

[0023] It should be noted that the direct discarding of format error files in the hot folder may misinjure the task containing valid instructions, causing the loss of the task; the task of a large file or high-frequency repeated submission is prone to cause storage transient overload or malicious attack due to the lack of intelligent fragmentation and risk assessment mechanism; therefore, by deploying a multi-threaded data monitoring service to capture file attribute characteristics in real time, a multi-dimensional interception rule library and dynamic processing strategy are constructed, which specifically shows that: for format abnormal files, the standardized print file is parsed and reconstructed in a sandbox environment to avoid misdeletion of valid tasks, and invalid format files are intercepted; the large file is intelligently fragmented and attached with a sequence identifier, and the current unprocessed task amount is combined to dynamically calculate the sub-file delay writing window, so as to relieve the storage pressure peak; for high-frequency repeated or blacklisted hash files, the risk level is dynamically evaluated based on the historical rollback record, and the permanent interception or frequency limiting mechanism is implemented to accurately block malicious attacks. The scheme ensures the integrity of the task while significantly reducing the occupation of invalid tasks on storage resources, improving the active defense capability of the system to abnormal writing, and optimizing the task queue load balancing through the fragmentation and delay injection of large files, enhancing the resource utilization and system fault tolerance in high-concurrency scenarios. The sandbox environment is an isolated, controlled and safe execution space, which can make suspicious files or programs be safely parsed, tested or executed without affecting the security of the main system, so as to realize intelligent processing of abnormal files while ensuring system security. The effective print instruction field refers to a specific code segment or structured data that meets the standardized print control command or data format recognizable and executable by a POD printer. These fields are usually included in print files to accurately describe various parameter requirements of print tasks, such as page size, color mode (CMYK / RGB), resolution (DPI), media type (paper / film), binding method and other key print attributes.

[0024] According to the embodiment of the present application, the load of the POD printer is predicted according to the print file data to obtain a load feature map, specifically: The print file data is monitored in real time, the print file queue is determined according to the upload timestamp of the print file, and the file attribute characteristics of the print file queue are obtained, the file attribute characteristics including file page number, color mode, media type and print quality parameters; The historical print log data of the POD printer in a preset time period is obtained, the historical file attribute characteristics and task time consumption data of the historical print log data are extracted, the historical file attribute characteristics and task time consumption data are analyzed for correlation, the influence of different file attribute characteristics on the time consumption of the POD printing is determined, and the influence data is obtained; acquire historical print file queue change data of a second preset time period, analyze the historical print file queue change data according to the influence data, determine task time consumption change data of the POD printer in the second preset time period, and determine POD printer load change data in the second preset time period according to the task time consumption change data; input the POD printer load change data into an LSTM-based load prediction model for training, the LSTM adopts a gating mechanism to capture the time sequence dependence of the printer load change, and different time scale load fluctuation features are extracted through stacked bidirectional LSTM layers; acquire print file queue change data of a current preset time period, determine task time consumption change of the print file queue change data of the current preset time period according to the influence data, and determine current load change data; input the current load change data into the trained load prediction model, predict the load in a future preset time period, and obtain a load change prediction result; perform cluster analysis on the load intensity according to the prediction result, identify peak load mode, stable load mode and idle load mode, perform graph modeling on file attribute features and time distribution features corresponding to different load modes, and generate a load feature graph containing a different time load intensity heat map.

[0025] It should be noted that the gating mechanism and time sequence memory capability of the LSTM neural network are used to accurately capture long-term dependence and nonlinear features in the printer load change, multi-level extraction of load fluctuation features of different time scales (such as minute-level task accumulation and hour-level business peak) is realized through stacked bidirectional LSTM layers; secondly, historical file attribute features (such as page number and color mode) are deeply analyzed in correlation with task time consumption, a quantitative evaluation model is established, and the prediction result can accurately reflect the influence weight of different print task types on the actual load; then, by inputting the current queue data in real time, the load change trend in the future time window is dynamically predicted, and the static limitation of the traditional threshold alarm is broken through; finally, based on the prediction result, mode clustering (peak / stable / idle) is performed and a load heat map is generated, the map not only directly shows the load intensity distribution of each period, but also forms an interpretable decision basis through the association of file attribute features, so that the system can identify potential overload risks in advance, provide data support for subsequent dynamic adjustment of print strategies (such as task fragmentation and resource scheduling), thereby significantly improving the throughput efficiency and stability of the print system and reducing the failure rate in high load periods.

[0026] Figure 2 A flowchart of constructing a POD printer print strategy of the application is shown.

[0027] According to the embodiment of the present application, the print file data in the hot folder is printed according to the load feature map, and a print strategy of the POD printer under different loads is constructed, specifically: S202, obtaining print failure occurrence data of the POD printer under different print durations in each load mode, determining failure probability distribution data under different print durations in each load mode according to the failure occurrence data, and establishing a correlation model of print duration and failure probability in each load mode according to the failure probability distribution data; S204, extracting a print task queue corresponding to the current load intensity from the load feature map, obtaining task attributes and estimated print time of each print file in the print task queue, and determining a continuous print duration of the POD printer to the print task queue according to the estimated print time; S206, presetting a print failure safety threshold, importing the print failure safety threshold into the correlation model for matching, and determining a safe continuous print duration of the POD; S208, if the continuous print duration exceeds the safe continuous print duration, dividing the print files in the print task queue into a plurality of print groups within the safe continuous print duration according to the estimated print time of each print file in the print task queue; S210, when the printing of a print group is completed, obtaining working temperature data and an optimal working temperature interval of the POD printer, calculating a required duration for the working temperature of the printer to recover to the optimal working temperature interval according to the working temperature data, and obtaining a print interval duration of the print group; S212, constructing a print strategy in a future preset time period according to the print group and the print interval duration.

[0028] It needs to be explained that by constructing a dynamic self-adaptive printing strategy, active fault prevention and resource optimization scheduling of the POD printer under a complex load scenario are realized, and specific technical effects are embodied as: based on a fault probability correlation model of different modes (peak / stable / idle) in a load feature map, the printing time length is quantitatively bound with the device fault risk, the over-limit risk of the current task queue is accurately identified by real-time matching of a safe continuous printing time length threshold; when it is detected that the continuous printing time length exceeds the safe threshold, intelligent fragmentation is performed in combination with file estimated time consumption, a long-time task is disassembled into multiple printing groups conforming to the safe time length, overheat wear of mechanical parts caused by continuous high-load operation (such as service life attenuation of a print head) is avoided, and through dynamic insertion of a temperature recovery interval (cooling time length is calculated according to real-time temperature data) between groups, the device is forced to operate within an optimal working temperature range, significantly reducing sudden faults caused by thermal stress; at the same time, through time window planning of the grouping strategy, a large task is discretized into multiple sub-task units that can be processed in parallel, in combination with resource utilization rate prediction of different time periods in the load feature map, dynamic matching of the task queue and the printer working state is realized, resource idling during a low-load period is avoided, and overload conflicts during a peak period are avoided, and finally a dynamic printing strategy with balanced load, controllable temperature and optimized fault rate is formed, improving the overall throughput efficiency and the service life of the device.

[0029] Figure 3 A flowchart of determining a fault-tolerant compensation scheme of the POD printer is shown.

[0030] According to the embodiment of the present application, the printing strategy is executed, a monitoring operation is performed on the printing strategy, fault analysis is performed on the POD printer, the fault tolerance capability of the POD printer is determined, and the fault-tolerant compensation scheme of the POD printer is determined according to the fault tolerance capability, specifically: S302, the printing strategy is executed, printing state data of the POD printer is collected in real time, including a print head temperature fluctuation value, a paper transmission rate deviation value and an abnormal increment of ink consumption, the printing state data is input into a preset fault prediction model, and the real-time fault probability of the current printing task within the remaining printing time length is calculated; S304, when it is detected that the real-time fault probability exceeds a preset fault warning threshold, the remaining number of pages of the current printing task and the hash check value of the printed pages are acquired, a redundant printing channel matched with the current printing task is constructed according to the file fragmentation features of the unexecuted tasks in the printing task queue, and the remaining number of pages and the unexecuted task fragments are synchronously distributed to the redundant printing channel; S306, deploying a virtual print queue in parallel with the main print channel in the redundant print channel, comparing the page generation logs of the main print channel and the redundant print channel in real time, when the number of consecutive inconsistencies between the page hash check values of the main print channel and the redundant print channel exceeds the preset fault tolerance threshold, freezing the current task of the main print channel and switching the print control to the redundant print channel, and regenerating a compensation print task according to the undamaged fragments corresponding to the fault pages of the main print channel from the virtual print queue to obtain a fault-tolerant print scheme.

[0031] It should be noted that by collecting multi-dimensional state data such as print head temperature fluctuation value, paper transmission rate deviation value and abnormal increment of ink consumption in real time, inputting a preset fault prediction model (learning the historical fault state of the printer state data through a convolutional neural network to construct a fault prediction model) to dynamically calculate the real-time fault probability of the current task, when the probability exceeds the warning threshold, the fault tolerance mechanism is triggered immediately: first, analyze the remaining number of pages and the hash check value (such as generating a unique identifier by SHA-256 algorithm) of the printed pages of the current task, and construct a redundant print channel according to the file fragment characteristics (such as metadata of page code or color layer fragments) of the unexecuted task; secondly, deploy a virtual print queue in the redundant channel which is completely consistent with the main channel, compare the page hash values generated by the main and standby channels in real time, if the number of consecutive inconsistencies exceeds the threshold, it is determined that the main channel fails and its task is frozen, and it is switched to the redundant channel for continuous execution; finally, according to the fragment identifier (such as the ID of the damaged page) corresponding to the fault page of the main channel, extract the undamaged fragments from the virtual queue to regenerate the compensation task, and only reprint the fault fragments instead of the entire task. Through this process, the system realizes accurate fault isolation (only processing abnormal fragments), dynamic redundancy switching (seamless connection of main and standby channels) and minimization of resource waste (avoiding full task restart), while combining the intelligent matching of fragment characteristics and the hash check mechanism to ensure the consistency (such as color accuracy, binding order) and continuity of the printing results during the fault tolerance process, ultimately enabling the POD system to maintain high availability under complex loads, significantly reducing the risk of order delay or cancellation due to hardware failure. The file fragment characteristics include file fragment size, fragment page code sequence, fragment data check code, and print parameter configuration (including resolution, color mode and ink coverage) corresponding to the print task; the redundant print channel is a redundant POD printer.

[0032] According to the embodiment of the application, the method further comprises: A hot folder write traffic real-time monitoring module is constructed, and a sliding time window algorithm is used to capture the current file write rate and file quantity increment, and the peak write rate of the same period in history is obtained; A capacity expansion urgency coefficient is calculated based on a deviation value of a current write rate and a historical peak write rate, when the capacity expansion urgency coefficient exceeds a preset threshold, a high-priority printing task is identified according to a task attribute feature of a current unprocessed file queue, and a capacity pre-allocation scheme is generated; In the process of expanding the hot folder capacity, the temporary buffer storage area is used to receive the file write request exceeding the current capacity, and the buffered files are migrated back to the hot folder according to the task priority order after the expansion is completed. The monitoring frequency of the sliding time window is dynamically adjusted according to the real-time write rate change.

[0033] It should be noted that during the operation of the POD printing system, when encountering a promotional activity or a sudden large-scale order, the static capacity management mechanism of the hot folder may have serious lag. The expansion mode of the hot folder cannot effectively cope with the nonlinear surge of write traffic, resulting in file write queue congestion, high-priority task delay, and even data loss. At the same time, the service interruption during the expansion process will cause task submission failure on the user side, and the lack of intelligent identification of task priority may cause key business order processing lag, therefore, by monitoring the file write traffic in real time and dynamically evaluating the expansion demand combined with historical peak data, the sliding window algorithm is used to accurately capture the traffic fluctuation trend, the high-priority task is intelligently identified, and the pre-allocation strategy is generated, at the same time, the temporary buffer storage area is created to realize seamless connection of business during the expansion period, the adaptive monitoring frequency adjustment mechanism is adopted to build a closed-loop control system, which effectively guarantees the timeliness of file processing in large-scale concurrent scenarios, significantly improves the system resource utilization, realizes the priority processing of key printing tasks and the stable operation of high-throughput jobs, and fundamentally solves the capacity expansion lag and business continuity damage problems in high-concurrency scenarios.

[0034] Figure 4 A block diagram of a POD printing monitoring system based on load prediction is shown.

[0035] The second aspect of the present application also provides a POD printing monitoring system 4 based on load prediction, which comprises a memory 41 and a processor 42, wherein the memory comprises a POD printing monitoring method program based on load prediction, and the POD printing monitoring method program based on load prediction is executed by the processor to realize the following steps: A hot folder of the POD printer is constructed, and a data monitoring operation is performed on the hot folder to obtain printing file data of the hot folder in real time; Load prediction is performed on the POD printer according to the printing file data to obtain a load feature map; The printing file data in the hot folder is processed according to the load feature map to construct a printing strategy of the POD printer under different loads; The printing strategy is executed, the printing strategy is monitored, the POD printer is analyzed for faults, the fault tolerance capability of the POD printer is determined, and a fault tolerance compensation scheme for the POD printer is determined according to the fault tolerance capability.

[0036] The application discloses a POD printing monitoring method and system based on load prediction. The method obtains printing file data by constructing a hot folder of a POD printer and real-time data monitoring; then, the POD printer is predicted for load according to the data, and a load feature map is generated. According to the map, the printing files in the hot folder are processed, and corresponding printing strategies are formulated under different loads. After the printing strategies are executed, the system monitors the printing process, analyzes faults, and evaluates the fault tolerance capability of the POD printer. According to the fault tolerance capability, a fault tolerance compensation scheme is generated and implemented, so as to ensure the stability of the printing process. Through load prediction and real-time monitoring, the application can identify potential problems in advance, optimize the printing strategy, reduce the fault risk, and improve the printing efficiency and stability.

[0037] In several embodiments provided in the present application, it should be understood that the disclosed devices and methods can be implemented in other manners. The above-described device embodiments are only schematic. For example, the division of the units is only a logical function division. There can be another division manner for the actual implementation, for example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. In addition, the coupling or direct coupling or communication connection between the above-described components can be indirect coupling or communication connection through some interfaces, devices or units, and can be electrical, mechanical or other forms.

[0038] The units described as separate components can or can not be physically separate, and the components shown as units can or can not be physical units; they can be located in one place or distributed on multiple network units; and part or all of the units can be selected according to actual needs to achieve the purpose of the embodiment.

[0039] In addition, each functional unit in each embodiment of the application can be integrated into one processing unit, or each unit can be a separate unit, or two or more units can be integrated into one unit; the integrated unit can be realized in the form of hardware or in the form of hardware plus software functional unit.

[0040] Those skilled in the art can understand that all or part of the steps of the above-mentioned method embodiments can be completed by program instruction related hardware, the foregoing program can be stored in a computer readable storage medium, and the program executes the steps of the method embodiments when executed; and the foregoing storage medium includes a mobile storage device, a read-only memory (ROM), a random access memory (RAM), a magnetic disc or an optical disc, and various storage medium capable of storing program codes.

[0041] Alternatively, the integrated unit of the present application can be stored in a computer readable storage medium if it is realized in the form of a software function module and sold or used as an independent product. Based on such understanding, the technical solutions of the embodiments of the present application can be embodied in the form of a software product, and the computer software product is stored in a storage medium, includes a plurality of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the methods described in the embodiments of the present application. The foregoing storage medium includes a mobile storage device, a ROM, a RAM, a magnetic disc or an optical disc, and various storage medium capable of storing program codes.

[0042] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto, any person skilled in the art can easily think of changes or replacements within the technical range disclosed in the present application, which should be covered in the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A POD print monitoring method based on load prediction, characterized by, The method comprises the following steps: constructing a hot folder of the POD printer, performing a data monitoring operation on the hot folder, and acquiring print file data of the hot folder in real time; performing load prediction on the POD printer according to the print file data to obtain a load feature map; performing print processing on the print file data in the hot folder according to the load feature map, and constructing a print strategy of the POD printer under different loads; executing the print strategy, monitoring the print strategy, analyzing faults of the POD printer, determining fault tolerance capability of the POD printer, and determining a fault tolerance compensation scheme of the POD printer according to the fault tolerance capability.

2. The POD print monitoring method based on load prediction according to claim 1, characterized in that, The method of constructing the hot folder of the POD printer and performing the data monitoring operation on the hot folder to acquire the print file data of the hot folder in real time specifically comprises: acquiring historical print task peak value data of the POD printer, dividing an initial capacity of the hot folder in a local storage device according to the historical print task peak value data, and constructing the hot folder of the POD printer according to the initial capacity; constructing a storage pressure prediction model of the hot folder, acquiring residual storage space, file writing rate, and file quantity increment of the hot folder according to the storage pressure prediction model, and simultaneously acquiring historical storage pressure data of the same time period in a historical preset time period, including a storage space occupation peak value, a file writing rate fluctuation curve, and an expansion operation record; extracting historical storage pressure features of the historical storage pressure data according to a sliding time window algorithm, including a storage space consumption rate and a writing rate fluctuation; importing the historical storage pressure features and the residual storage space, the file writing rate, and the file quantity increment of the hot folder into the storage pressure prediction model, performing decay prediction on the current residual storage space of the hot folder by an exponential smoothing method, and determining a storage space depletion time point of the hot folder; determining an expansion time point of the hot folder according to the storage space depletion time point, determining an expansion capacity of the hot folder in the preset time period according to the file quantity increment, and obtaining a capacity expansion scheme of the hot folder; expanding the hot folder according to the capacity expansion scheme, performing a data monitoring operation on the hot folder in real time, intercepting an exception of the hot folder, and acquiring print file data of the hot folder.

3. The POD print monitoring method based on load prediction according to claim 2, characterized in that, The method of performing the data monitoring operation on the hot folder in real time and intercepting an exception of the hot folder specifically comprises: deploying a multi-thread file data monitoring service at a data transmission input port of the hot folder, capturing a file writing event in real time through a data monitoring thread of the hot folder, extracting attribute feature data of a currently written file, including a file format identifier, a byte size, a hash fingerprint, and a writing time interval; and matching the attribute feature data with a preset interception rule library, wherein the interception rule library contains a format type whitelist, a single file capacity threshold, a hash blacklist, and a time window writing frequency threshold. When it is detected that the write file format identifier is not in the format type whitelist, the write file is intercepted, the intercepted file is transferred to a sandbox environment for format analysis, if valid print instruction fields are extracted after the analysis, a standardized print file is regenerated according to the instruction fields and returned to a hot folder, otherwise the file is marked as a format exception file and deleted, obtaining a first interception strategy; If the byte size of the current write file exceeds the single file capacity threshold, the write file is split into multiple sub-files according to a preset size and a sharding print order identifier is attached, a write delay time window of the split sub-files is calculated according to the number of current unprocessed tasks in the hot folder, and the sub-files are injected into the hot folder in batches according to the delay time window, obtaining a second interception strategy; When it is detected that the write frequency of the same hash fingerprint within a preset time window exceeds a write frequency threshold or exists in a hash blacklist, the write process of the write file is frozen and a repeated task alert is generated, the risk level of the file is evaluated based on the task rollback record of the hash fingerprint in the historical interception log, if the risk level exceeds a preset threshold, the hash fingerprint is permanently added to the hash blacklist, otherwise the write frequency of the hash fingerprint is temporarily limited and a temporary interception lock is generated, obtaining a third interception strategy.

4. The POD print monitoring method based on load prediction according to claim 1, characterized in that, The load of the POD printer is predicted according to the print file data, and a load feature map is obtained, specifically: Real-time monitoring of print file data, determining a print file queue according to the upload timestamp of the print file, obtaining file attribute features of the print file queue, the file attribute features including file page number, color mode, medium type and print quality parameters; obtaining historical print log data of the POD printer in a preset time period, extracting historical file attribute features and task time consumption data of the historical print log data, performing correlation analysis on the historical file attribute features and the task time consumption data, determining the influence of different file attribute features on the POD printing time consumption, and obtaining influence data; obtaining historical print file queue change data in a second preset time period, analyzing the historical print file queue change data according to the influence data, determining task time consumption change data of the POD printer in the second preset time period, and determining POD printer load change data in the second preset time period according to the task time consumption change data; inputting the POD printer load change data into an LSTM-based load prediction model for training, the LSTM using a gating mechanism to capture the time sequence dependence of the printer load change, and extracting load fluctuation features of different time scales through stacked bidirectional LSTM layers; obtaining print file queue change data in a current preset time period, determining task time consumption change of the print file queue change data in the current preset time period according to the influence data, and determining current load change data; inputting the current load change data into the trained load prediction model to predict the load in a future preset time period, and obtaining a load change prediction result. According to the prediction result, the load intensity is clustered and analyzed to identify peak load mode, stable load mode and idle load mode, the file attribute features and time distribution features corresponding to different load modes are graphically modeled to generate a load feature graph containing a thermal map of different time load intensity.

5. The POD print monitoring method based on load prediction according to claim 1, characterized in that, According to the load feature graph, the print file data in the hot folder is printed to construct a print strategy of the POD printer under different loads, specifically: Obtain the print failure occurrence data of the POD printer under different print durations in each load mode, determine the failure probability distribution data under different print durations in each load mode according to the failure occurrence data, and establish an association model of print duration and failure probability under each load mode according to the failure probability distribution data; Extract the print task queue corresponding to the current load intensity from the load feature graph, obtain the task attributes and estimated print time of each print file in the print task queue, and determine the continuous print duration of the POD printer for the print task queue according to the estimated print time; A predetermined print failure safety threshold is imported into the association model for matching to determine the safe continuous print duration of the POD; If the continuous print duration exceeds the safe continuous print duration, the print files in the print task queue are divided into multiple print groups within the safe continuous print duration according to the estimated print time of each print file in the print task queue; When printing of a print group is completed, obtain the working temperature data and the optimal working temperature interval of the POD printer, calculate the required duration for the working temperature of the printer to recover to the optimal working temperature interval according to the working temperature data, and obtain the print interval duration of the print group; The print group and print interval duration are used to construct a print strategy for a future predetermined time period.

6. The POD print monitoring method based on load prediction according to claim 1, wherein, The print strategy is executed, the print strategy is monitored, the fault tolerance capability of the POD printer is determined, the fault tolerance compensation scheme of the POD printer is determined according to the fault tolerance capability, and specifically: The print strategy is executed, the print state data of the POD printer is collected in real time, including the print head temperature fluctuation value, the paper transmission rate deviation value and the abnormal increment of ink consumption, the print state data is input into a predetermined fault prediction model, the real-time failure probability of the current print task within the remaining print duration is calculated; When it is detected that the real-time failure probability exceeds a predetermined failure warning threshold, the remaining number of pages of the current print task and the hash check value of the printed pages are obtained, a redundant print channel matching the current print task is constructed according to the file fragment features of the unexecuted tasks in the print task queue, and the remaining number of pages and the unexecuted task fragments are synchronously distributed to the redundant print channel; In the redundant printing channel, a virtual printing queue parallel to the main printing channel is deployed, a page generation log of the main printing channel and the redundant printing channel is compared in real time, when the number of continuous inconsistency between the page hash check value of the main printing channel and the page hash check value of the redundant printing channel exceeds a preset fault tolerance threshold, the current task of the main printing channel is frozen and the printing control right is switched to the redundant printing channel, and a compensation printing task is regenerated according to the undamaged fragments extracted from the virtual printing queue according to the fragment identifier corresponding to the fault page of the main printing channel, so that a fault-tolerant printing scheme is obtained.

7. A POD print monitoring system based on load prediction, characterized by, The POD printing monitoring system based on load prediction comprises a storage and a processor, the storage comprises a POD printing monitoring method program based on load prediction, and the following steps are realized when the POD printing monitoring method program based on load prediction is executed by the processor: A hot folder of a POD printer is constructed, a data monitoring operation is performed on the hot folder, and printing file data of the hot folder is acquired in real time; Load prediction is performed on the POD printer according to the printing file data, and a load characteristic map is obtained; The printing file data in the hot folder is processed according to the load characteristic map, and a printing strategy of the POD printer under different loads is constructed; The printing strategy is executed, the printing strategy is monitored, fault analysis is performed on the POD printer, the fault tolerance capability of the POD printer is determined, and a fault tolerance compensation scheme of the POD printer is determined according to the fault tolerance capability.

Images