Synchronization signal fault-tolerant method and system for multi-head printing

Abstract

This invention discloses a synchronization signal fault-tolerant method and system for multi-head printing, relating to the field of multi-head printing technology. The method includes: acquiring raw printing decision data and the cumulative number of anomaly handling based on received abnormal printing detection results; obtaining the abnormal printing detection results based on coded synchronization signal detection; generating the coded synchronization signal from the raw photocell signal through parity-even sequence differentiation and differential pulse width modulation; performing discrete level classification on the raw printing decision data to obtain decision factor levels, and determining the decision mode based on the cumulative number of anomaly handling; making intelligent decisions based on the decision mode and the decision factor levels to obtain the target fault-tolerant action; and controlling the printhead control board corresponding to the abnormal printing detection results to execute the target fault-tolerant action. This invention addresses the technical problem of insufficient printing operation stability caused by the susceptibility of synchronization signals to interference and the limited fault-tolerant handling methods after anomalies during multi-head printing.

Description

Technical Field

[0001] This invention relates to the field of multi-head printing technology, and more particularly to a synchronous signal fault-tolerant method and system for multi-head printing. Background Technology

[0002] To meet the demands of large printing formats and multi-color production, industrial inkjet printers generally employ a multi-printer architecture for collaborative operation. Existing printing equipment relies on photoelectric sensors to collect trigger signals, which are then distributed to the individual printhead control boards via a pulse distribution board. This synchronization pulse enables collaborative printing across multiple printheads.

[0003] The electromagnetic environment at industrial production sites is complex, and the synchronization pulses of traditional synchronous transmission schemes are easily affected by electromagnetic interference. Abnormal synchronization signals can easily lead to printing misalignments, production interruptions, and in severe cases, the scrapping of large quantities of consumables. Traditional synchronous fault-tolerance solutions are singular and fixed, with poor flexibility in their fault-tolerance strategies, making it difficult to adapt to changing on-site interference conditions. Summary of the Invention

[0004] The purpose of this invention is to provide a fault-tolerant method and system for synchronization signals in multi-head printing, which solves the technical problem of insufficient printing stability caused by the fact that synchronization signals are easily interfered with during multi-head printing and the fault-tolerant handling method after anomalies is singular.

[0005] To achieve the above objectives, the present invention provides the following technical solution: In a first aspect, the present invention provides a fault-tolerant method for synchronization signals in multi-printer printing, the method comprising: Based on the received abnormal printing detection results, the original data for printing decisions and the cumulative number of abnormal processing are obtained; the abnormal printing detection results are obtained based on the detection of the encoded synchronization signal; the encoded synchronization signal is generated by the original photoelectric sensor signal through parity number differentiation and differential pulse width modulation; Discretely classify the raw data for printing decisions to obtain the decision factor levels, and determine the decision mode based on the cumulative number of anomaly handling. Based on the decision-making model and combined with the level of decision-making factors, intelligent decision-making is carried out to obtain the target fault-tolerant action; The printhead control board corresponding to the abnormal printing detection results executes the target fault-tolerant action.

[0006] Optionally, the raw data for printing decisions includes substrate value parameters, number of lost pulses, printing progress data, and signal error rate; The raw data for printing decisions is discretely categorized into levels to obtain the corresponding decision factor levels, specifically including: Based on the preset mapping relationship, the substrate value parameters, the number of lost pulses, the printing progress data, and the signal bit error rate are mapped to different levels to obtain the substrate value level, the lost pulse level, the printing stage level, and the interference intensity level.

[0007] Optionally, the decision-making model is determined based on the cumulative number of exceptions handled, specifically including: When the cumulative number of exceptions handled is less than the first number, a preset rule decision tree is used for decision-making. When the cumulative number of exceptions handled is greater than or equal to the first number and less than the second number, a weighted scoring decision is adopted. When the cumulative number of exceptions handled is greater than or equal to the second number, the AI ​​agent is used to query the pre-stored state action value query table to make a decision.

[0008] Optionally, based on the decision-making model and combined with the level of decision-making factors, intelligent decision-making is performed to obtain the target fault-tolerant action, specifically including: When using a pre-defined rule decision tree, decisions are made according to the pre-defined priority. If the base material value level is high value, the target fault-tolerant action is determined to be shutdown. If the base material value level is medium or low and the lost pulse level is 1, the target fault tolerance action is determined to be resend. If the base material value level is medium or low, the lost pulse level is greater than 1, and the printing stage level is the final stage, the target fault-tolerant action is determined to be reissue. If the substrate value level is medium or low, the lost pulse level is greater than 1, the printing stage level is non-final stage, and the interference intensity level is severe interference, the target fault-tolerant action is determined to be shutdown. If any of the aforementioned conditions are not met, the target fault-tolerant action is determined to be a page jump.

[0009] Optionally, based on the decision-making model and combined with the level of decision-making factors, intelligent decision-making is performed to obtain the target fault-tolerant action, specifically including: When using a weighted scoring system for decision-making, the individual scores for each fault-tolerant action corresponding to the base material value, lost pulse, printing stage, and interference intensity are retrieved from the four preset scoring tables and substituted into the comprehensive scoring formula: ; Calculate the combined scores for shutdown, resend, and page skipping respectively, and select the fault-tolerant action with the highest score as the target fault-tolerant action; among them, Characterizes fault-tolerant actions; Characterizes the value of the substrate; Characterizes lost pulses; Characterizing the printing stage; Characterizes the intensity of interference; The individual score corresponding to the error-tolerant action; , , , These are the preset weighting coefficients.

[0010] Optionally, based on the decision-making model and combined with the level of decision-making factors, intelligent decision-making is performed to obtain the target fault-tolerant action, specifically including: When using an AI agent to query a pre-stored state-action value lookup table for decision-making, the formula is: ; The substrate value level, lost pulse level, printing stage level, and interference intensity level are encoded as unique status codes; among them... For state coding; , , , The numerical indices are, in order, the base material value level, the lost pulse level, the printing stage level, and the interference intensity level. The selectable values ​​for all four types of numerical indices are 0, 1, and 2. Based on the status number, query the status action value query table, and read the expected revenue value corresponding to the three fault-tolerant actions of shutdown, reissue, and page jump respectively. Select the fault-tolerant action with the largest expected revenue value as the initial fault-tolerant action. Based on expected returns and the formula: ; Calculate the confidence level of the initially selected fault-tolerant actions; where, Confidence level; The maximum expected return among the expected return values; It is the second largest expected return value; The confidence level is compared with a preset confidence threshold. If the confidence level is greater than or equal to the preset confidence threshold, the initially selected fault-tolerant action is determined as the target fault-tolerant action. If the confidence level is less than the preset confidence threshold, the process is backtracked and a rule-based decision tree is used to re-determine the target fault-tolerant action.

[0011] Optionally, perform the target fault-tolerant action, specifically including: When the target fault-tolerant action is to stop, a stop command is sent to the corresponding printhead control board. The printhead control board stops printing ignition and retains the current printing position data. When the target fault-tolerant action is to reissue, a reissue command is sent to the printhead control board. The printhead control board sends a reissue request to the pulse distribution board based on the reissue command. The pulse distribution board reissues the corresponding sequence pulse based on the reissue request and sends it to the corresponding printhead control board to resume printing. When the target error-tolerant action is page skipping, a page skipping command is sent to the printhead control board. The printhead control board skips the specified number of pages based on the page skipping command and then continues printing.

[0012] Optionally, the method further includes: After completing the target fault-tolerant action, record the substrate value level, lost pulse level, printing stage level, interference intensity level, selected target fault-tolerant action, and action execution result for this printing. The cumulative number of exceptions handled is incremented by one and persisted. If the decision-making mode corresponding to the AI ​​agent is used for this printing, the reward value for a single interaction is calculated according to the preset reward rules, and the state, action, reward and subsequent state are stored in the experience pool.

[0013] Optionally, the preset reward rules include: assigning a positive reward for successful printing, assigning different negative rewards for error shutdown and substrate scrap, and setting corresponding reward values ​​for successful reprint and normal continuation of printing after page skipping.

[0014] Compared with existing technologies, the synchronous signal fault-tolerant method for multi-head printing provided by this invention distinguishes between odd and even numbers and modulates the pulse width difference of the original photoelectric signal to generate an coded synchronous signal. Anomaly detection is completed based on this highly anti-interference coded synchronous signal, effectively solving the problems of traditional synchronous signals being susceptible to interference and anomaly detection distortion. At the same time, it collects multi-dimensional printing decision raw data and completes discrete level quantization. Based on the cumulative number of anomaly handling, it adaptively matches and adapts to the decision-making mode of different production stages. Combined with the multi-dimensional decision factor levels, it intelligently outputs adaptive target fault-tolerant actions, abandoning the traditional single and fixed fault-tolerant handling method, realizing differentiated intelligent fault tolerance under all working conditions. It effectively solves the defects of rigid and poor adaptability of multi-head printing fault tolerance strategies, greatly improves the operational stability of multi-head printing, and improves the accuracy of printing anomaly fault tolerance and production efficiency.

[0015] In a second aspect, the present invention also provides a synchronous signal fault-tolerant system for multi-printer printing, comprising: a host computer, a pulse distribution board, and multiple printer control boards; The host computer is used to obtain the original data for printing decisions and the cumulative number of abnormalities processed based on the abnormal printing detection results fed back by the printhead control board; the abnormal printing detection results are obtained based on the detection of the coded synchronization signal; the coded synchronization signal is generated by the pulse distribution board by distinguishing the original photoelectric sensor signal by parity sequence number and differential pulse width modulation; Discretely classify the raw data for printing decisions to obtain the decision factor levels, and determine the decision mode based on the cumulative number of anomaly handling. Based on the decision-making model and combined with the level of decision-making factors, intelligent decision-making is carried out to obtain the target fault-tolerant action; The printhead control board corresponding to the abnormal printing detection results executes the target fault-tolerant action.

[0016] Compared with the prior art, the beneficial effects of the second aspect of the present invention are the same as those of the method provided in the first aspect above, and will not be repeated here. Attached Figure Description

[0017] The accompanying drawings, which are included to provide a further understanding of the invention and form part of this invention, illustrate exemplary embodiments of the invention and are used to explain the invention, but do not constitute an undue limitation of the invention. In the drawings: Figure 1 A flowchart illustrating a synchronization signal fault-tolerant method for multi-head printing provided in one embodiment of the present invention; Figure 2 A schematic diagram of a fault-tolerant synchronization signal system for multi-head printing provided as an embodiment of the present invention; Figure 3 This is a schematic diagram of pulse modulation of a synchronization signal provided in one embodiment of the present invention. Detailed Implementation

[0018] To facilitate a clear description of the technical solutions in the embodiments of the present invention, the terms "first" and "second" are used to distinguish identical or similar items with essentially the same function and effect. For example, the first threshold and the second threshold are merely used to distinguish different thresholds and do not limit their order. Those skilled in the art will understand that the terms "first" and "second" do not limit the quantity or execution order, and that the terms "first" and "second" are not necessarily different.

[0019] It should be noted that in this invention, the terms "exemplary" or "for example" are used to indicate examples, illustrations, or descriptions. Any embodiment or design described as "exemplary" or "for example" in this invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of terms such as "exemplary" or "for example" is intended to present the relevant concepts in a concrete manner.

[0020] In this invention, "at least one" means one or more, and "more than one" means two or more. "And / or" describes the relationship between the associated objects, indicating that three relationships can exist.

[0021] like Figure 1 As shown, this embodiment of the invention provides a fault-tolerant method for synchronization signals in multi-head printing, the method including: Step S1: Based on the received abnormal printing detection results, the host computer obtains the original data for printing decisions and the cumulative number of abnormal processing; the abnormal printing detection results are obtained based on the detection of the encoded synchronization signal; the encoded synchronization signal is generated by the original photoelectric sensor signal through parity number differentiation and differential pulse width modulation; It should be noted that the synchronization signal fault-tolerant method for multi-head printing provided in this embodiment of the invention is implemented using a synchronization signal fault-tolerant system for multi-head printing. For example, the system is as follows: Figure 2 As shown, the system includes a host computer, photoelectric sensors, a pulse distribution board, and multiple printhead control boards arranged in an array. The host computer establishes bidirectional communication connections with the pulse distribution board and each printhead control board. The pulse distribution board sends coded synchronization signals to the pre-set printhead control boards. These coded synchronization signals are transmitted sequentially from the set printhead control board along the set printhead direction (e.g., the column direction of the array arrangement) to the next printhead until they are transmitted to the last printhead control board among the multiple printhead control boards. In this way, the multiple printhead control boards arranged in an array work together to complete the multi-printhead synchronous printing operation.

[0022] The photoelectric sensor (photocell) is installed on the side of the printing platform's motion axis to detect the substrate's movement position or the change in the grating scale in real time. Each time a valid printing trigger position is detected, a raw trigger pulse is output. The raw trigger pulse is connected to the pulse distribution board, and after board-level signal shaping and filtering, a stable and recognizable raw photocell signal is formed for subsequent encoding and modulation.

[0023] The pulse distribution board integrates a modulation and coding logic unit and a network communication unit, with the two units working in a separate manner.

[0024] The modulation and coding logic unit is used to perform sequence number statistics and pulse width reconstruction on the original electro-eye signal. Specifically, the parity sequence number identification and differential pulse width modulation process is performed as follows: The pulse distribution board has a built-in incremental trigger counter. This counter updates its count value each time it receives a raw photocell signal and assigns an even or odd number to the current trigger event based on the count value; combined with... Figure 3 To illustrate, for example, when the pulse distribution board performs pulse modulation on a pre-set nozzle control board (e.g., the first nozzle control board), if the sequence number of the current triggering event is odd, the modulation encoding logic unit generates a pulse with a first preset pulse width (e.g., Figure 3 The modulation pulse has an odd-numbered pulse width L; if the sequence number of the current triggering event is even, the modulation coding logic unit generates a modulation pulse with a second preset pulse width (e.g., the pulse width L of the photoelectric sensor). Figure 3 The modulation pulse of the even-numbered photoelectric pulse width (2L) in the photoelectric sensor enables the odd-even trigger pulse to have distinguishable pulse width characteristics.

[0025] The original photoelectric sensor signal is modulated by the aforementioned odd-even pulse width modulation to generate a main modulation pulse. The modulation coding logic unit can also add a coded frame immediately after the main modulation pulse. The main modulation pulse and the additional coded frame together constitute the aforementioned coded synchronization signal. The additional coded frame can carry targeted printhead control instructions for fine-tuning the printing action of a specified number of printheads. For example, a printhead origin offset adjustment instruction can be written into the additional coded frame for subsequent printhead control board parsing and execution, realizing online dynamic compensation of the printing position.

[0026] The network communication unit is used to realize network data interaction between the pulse distribution board, the host computer, and each nozzle control board. It is responsible for sending out coded synchronization signals, receiving equipment status data and abnormal feedback data returned by the nozzle control board, and ensuring stable communication transmission between the pulse distribution board, the host computer, and the nozzle control board.

[0027] Multiple printhead control boards are arranged in an array, and each printhead control board integrates a protocol decoder and status monitoring logic. The protocol decoder is used to receive the coded synchronization signal sent by the pulse distribution board, restore the parity trigger sequence number by identifying the pulse width characteristics, synchronously verify the continuity and timing legality of pulse transmission, determine whether there are abnormalities such as lost pulses, signal disorder, and signal interference, and finally generate abnormal printing detection results and feed them back to the host computer.

[0028] The status monitoring logic is a built-in operation monitoring program logic of the printhead control board. Relying on the printhead control board's built-in hardware monitoring pins and signal acquisition circuit, it collects the current printhead's printing operation status, printing progress, and signal error rate in real time, and provides real-time feedback on the equipment's operating status to the host computer. Simultaneously, it can accurately receive fault-tolerant control commands issued by the host computer, driving the corresponding physical printhead to perform normal printing operations or corresponding fault-tolerant actions. Furthermore, after successfully decoding the main modulation pulse signal through the protocol decoder, the printhead control board can continue to parse the additional coded frame attached to the main pulse. If the additional coded frame carries a fine-tuning control command adapted to the current printhead control board or its managed printhead row, the status monitoring logic will execute the corresponding fine-tuning action, modifying the relative ignition offset address of the corresponding printhead row for this printing in real time, achieving dynamic calibration compensation of the printing position.

[0029] Step S2: Perform discrete level classification on the printed decision data to obtain the decision factor levels, and determine the decision mode based on the cumulative number of anomaly handling. Step S3: Based on the decision-making model and the level of decision factors, make intelligent decisions to obtain the target fault-tolerant action; Step S4: Control the printhead control board corresponding to the abnormal printing detection result to perform the target fault-tolerant action.

[0030] For example, fault-tolerant actions may include resending, page skipping, and stopping. The target fault-tolerant action is one of these three: resending, page skipping, and stopping.

[0031] The beneficial effects of this embodiment are as follows: 1) By relying on parity-differentiated pulse width encoding in conjunction with protocol decoding, the defect of synchronization signals being susceptible to electromagnetic interference is improved. The photoelectric sensor outputs the original trigger pulse, which is sent to the pulse distribution board for shaping and filtering to form the original photoelectric sensor signal. The modulation and encoding logic unit in the board relies on the built-in incremental trigger counter to count the trigger sequence and divide the pulse parity sequence number. Odd pulses are configured with a first preset pulse width, and even pulses are configured with a second preset pulse width to generate a main modulation pulse with differentiated characteristics. An additional encoded frame is appended after the main modulation pulse. The main modulation pulse and the additional encoded frame together constitute the encoded synchronization signal. Unlike conventional unmarked single pulse signals, this embodiment uses differentiated pulse widths to form exclusive protocol characteristics. Randomly generated electromagnetic interference spikes on site cannot match the predetermined pulse width specifications and can be naturally filtered out by hardware, optimizing anti-interference capabilities from the signal source. The nozzle control board is equipped with a protocol decoder. After receiving the encoded synchronization signal, it uses a hardware timer to measure the pulse width and restore the pulse parity attribute. Then, it relies on the internal sequence number state machine to verify the pulse alternation pattern. Once the same type of pulse appears continuously or there is no signal after timeout, it can be determined that the signal is disturbed or the pulse is lost. By using protocol decoding to distinguish between valid signals and interference noise, signal loss can be quickly located, overcoming the shortcomings of traditional solutions that cannot distinguish noise and are prone to misjudging faults. The anti-interference performance of the photoelectric sensor signal is improved by relying on the whole encoding and decoding architecture, and the fault of lost pulse can be detected in a timely manner; 2) After the host computer obtains relevant data, it divides the parameter levels and selects the decision mode. Based on the decision result, the corresponding fault-tolerant action is output. Compared with the traditional fixed and single fault handling method, it can select the appropriate fault-tolerant operation for abnormal working conditions, and overcome the defect of the one-size-fits-all fault-tolerant strategy.

[0032] 3) The additional encoded frame in the encoding synchronization signal can carry printhead fine-tuning control instructions. After the printhead control board parses the instructions, it dynamically adjusts the printhead ignition offset parameters, compensates for printing position deviations online, reduces printing defects caused by printhead offset, and further helps to improve the overall printing stability.

[0033] In an exemplary embodiment, the printing decision factors may include four factors: substrate value, lost pulses, printing stage, and interference intensity. Based on these four factors, the corresponding raw data for printing decisions includes substrate value parameters, number of lost pulses, printing progress data, and signal error rate. The raw data for printing decisions is then discretely categorized to obtain corresponding decision factor levels, which may specifically include: Based on the preset mapping relationship, the substrate value parameters, the number of lost pulses, the printing progress data, and the signal bit error rate are mapped to different levels to obtain the substrate value level, the lost pulse level, the printing stage level, and the interference intensity level.

[0034] The preset mapping relationship refers to the hierarchical determination rules for the four types of raw data for printing decisions. For example, the raw data for each printing decision are mapped to a different level according to the following rules: Regarding the substrate value parameter: Based on the thresholds preset by the system, the substrate value is divided into three levels: high, medium and low. A single substrate with a value exceeding the high threshold is classified as high value, a value between the high and low thresholds is classified as medium value, and a value below the low threshold is classified as low value. Regarding the number of lost pulses: 1 lost pulse corresponds to Level 1, 2 to 5 lost pulses correspond to Level 2, and more than 5 consecutive lost pulses correspond to Level 3. Regarding printing progress data: This printing progress data is generated by calculating the percentage of printing completion from the number of printed lines and the total number of printed lines. Based on this percentage, the printing stage is divided into levels: <5% is the starting stage, >95% is the ending stage, and the rest is the intermediate stage. Regarding the signal bit error rate: based on the percentage of bit errors obtained from nearly 100 pulse cycles, the interference intensity level is divided into levels, with values ​​<1% being slight interference, 1% to 5% being moderate interference, and >5% being severe interference.

[0035] The beneficial effects of this embodiment are as follows: This embodiment uses four decision factors—substrate value, lost pulse, printing stage, and interference intensity—as the evaluation criteria. Unlike existing technologies that rely on only a single condition to determine faults, this embodiment comprehensively collects operating condition information from four dimensions: consumable attributes, fault severity, production progress, and on-site interference environment, achieving a three-dimensional and quantitative assessment of the fault status. By comprehensively analyzing the actual impact of the fault through multi-dimensional information, it avoids the evaluation bias caused by a single indicator, ensuring that the fault determination results closely match the actual on-site operating conditions. This provides accurate data support for subsequent fault-tolerant processing, optimizes the rationality of handling from the source of decision-making, and ultimately improves the stability of the entire printing machine's operation.

[0036] In an exemplary embodiment, determining the decision-making mode based on the cumulative number of exceptions handled specifically includes: When the cumulative number of exceptions handled is less than the first number, a preset rule decision tree is used for decision-making. When the cumulative number of exceptions handled is greater than or equal to the first number and less than the second number, a weighted scoring decision is adopted. When the cumulative number of exceptions handled is greater than or equal to the second number, the AI ​​agent (Artificial Intelligence Agent) is used to query the pre-stored state action value query table to make a decision; the second number is greater than the first number.

[0037] Specifically, the cumulative number of exceptions handled is the total number of exceptions handled by the system since its first run.

[0038] For example, the system continuously counts each anomaly and accumulates the cumulative number of anomaly handling. Based on the real-time statistical values, it switches the corresponding decision mode: when the system has just been put into production and the cumulative number of anomaly handling is less than 200, the decision tree is activated; when the number of anomaly handling accumulates to the range of 200 to 499, it switches to weighted scoring; when the number of anomaly samples accumulates to 500 or more, the AI ​​agent is activated to make decisions.

[0039] The beneficial effects of this embodiment are as follows: 1) When the number of samples is small during the cold start phase, the decision tree with fixed rules can stably output the handling plan, solving the problem that the intelligent algorithm cannot be implemented due to insufficient data, and ensuring that the equipment can be normally fault-tolerant as soon as it is put into operation; 2) When the cumulative number of anomalies is in the middle range, weighted scoring is adopted, which refines the evaluation of various factors compared with fixed rules, steadily improves the decision accuracy, and continuously accumulates sample data to build up the training of intelligent models; 3) After the cumulative number of anomalies is sufficient, AI intelligent agent decision-making is enabled, which autonomously optimizes the judgment logic based on massive historical working conditions, and the decision adaptability is stronger; the multi-segmentation mechanism takes into account the decision reliability under different data volumes throughout the entire life cycle of the equipment, and continuously optimizes the printing fault tolerance effect.

[0040] In an exemplary embodiment, step S3: Based on the decision-making model and combined with the level of decision factors, intelligent decision-making is performed to obtain the target fault-tolerant action, specifically including: When using a rule-based decision tree, the decision tree is a deterministic if-else rule preset during the system deployment phase. During equipment operation, the parameters and judgment logic remain fixed and are judged sequentially according to the preset priority. If the base material value level is high value, the target fault-tolerant action is determined to be shutdown. If the base material value level is medium or low and the lost pulse level is 1, the target fault tolerance action is determined to be resend. If the base material value level is medium or low, the lost pulse level is greater than 1, and the printing stage level is the final stage, the target fault-tolerant action is determined to be reissue. If the substrate value level is medium or low, the lost pulse level is greater than 1, the printing stage level is non-final stage, and the interference intensity level is severe interference, the target fault-tolerant action is determined to be shutdown. If any of the aforementioned conditions are not met, the target fault-tolerant action is determined to be a page jump.

[0041] Specifically, the three types of fault-tolerant actions in this embodiment of the invention are defined as follows: Stopping refers to the host computer issuing a stop command to the corresponding printhead control board, which stops printing ignition and retains the current printing position data. Re-issuance refers to sending a re-issuance command to the printhead control board, which then sends a re-issuance request to the pulse distribution board based on the re-issuance command. The pulse distribution board re-issus the corresponding sequence pulse and sends it to the printhead control board, thereby resuming printing. Page skipping refers to issuing a page skipping command to the printhead control board, which skips a specified number of pages and continues printing. Therefore, executing the target fault-tolerant action may specifically include the following steps: (1) When the target fault-tolerant action is to stop, a stop command is sent to the corresponding printhead control board, the printhead control board stops printing ignition and retains the current printing position data; (2) When the target fault-tolerant action is to reissue, a reissue command is sent to the printhead control board. The printhead control board sends a reissue request to the pulse distribution board based on the reissue command. The pulse distribution board reissues the corresponding sequence number pulse based on the reissue request and sends it to the corresponding printhead control board to restore printing. (3) When the target error-tolerant action is to skip pages, a page skip command is sent to the printhead control board. The printhead control board skips the specified number of pages based on the page skip command and then continues printing.

[0042] The beneficial effects of this embodiment are as follows: 1) The decision tree is a fixed if-else logic that is deployed and finalized. The rules cannot be changed during operation. In the stages of system cold start, lack of historical abnormal samples, and inability to carry out weighted scoring and AI intelligent agent calculation, the equipment can complete the fault judgment based on the established conditions without the need for pre-training the model. This solves the problem that intelligent algorithms cannot be implemented in the early stage and ensures that fault-tolerant management can be achieved as soon as the equipment goes online; 2) The various judgment conditions of the decision tree are precisely matched with the applicable scenarios of the three types of fault-tolerant actions. Based on the linkage of four parameters, namely the value of the substrate, lost pulse, printing stage, and interference intensity, the handling plan is screened. High-value faults are directly shut down to protect the substrate. Occasional lost pulses are replaced to preserve the finished product. When there are multiple pulse abnormalities and non-serious interference, page skipping is used to maintain production capacity; 3) Setting page skipping as a fallback action fully covers all abnormal working conditions, eliminates the situation where there is no handling plan for faults, breaks the drawback of the traditional one-size-fits-all fault-tolerant strategy, takes into account material costs and production efficiency, effectively reduces problems such as substrate scrapping, unwarranted shutdown, and batch printing misalignment, and improves the overall printing stability of multi-head system.

[0043] In an exemplary embodiment, intelligent decision-making is performed based on a decision-making pattern and the level of decision factors to obtain a target fault-tolerant action, specifically including: When using a weighted scoring system for decision-making, the individual scores for each fault-tolerant action corresponding to the base material value, lost pulse, printing stage, and interference intensity are retrieved from the four preset scoring tables and substituted into the comprehensive scoring formula: (1) Calculate the combined scores for shutdown, resend, and page skipping respectively, and select the fault-tolerant action with the highest score as the target fault-tolerant action; in formula (1), Characterizes fault-tolerant actions; Characterizes the value of the substrate; Characterizes lost pulses; Characterizing the printing stage; Characterizes the intensity of interference; The individual score corresponding to the error-tolerant action; , , , These are the preset weighting coefficients.

[0044] For example, the system pre-sets scoring tables and fixed weights for each dimension, as follows: The fixed weights are preset as follows: , , , .

[0045] The scoring tables for each dimension are as follows: Table 1: Substrate Value Scoring

[0046]

[0047] Table 2: Lost Pulse Scoring

[0048]

[0049] Table 3: Scoring during the printing stage

[0050]

[0051] Table 4: Interference Intensity Scoring

[0052]

[0053] The beneficial effects of this embodiment are: 1) It replaces the fixed if-else hard judgment with weighted quantitative scoring. Relying on the scores and weights of different decision dimensions, it quantifies the degree of fault impact, no longer relying on rigid condition jumps. It achieves refined quantitative assessment of anomalies between boundary conditions, and the fault tolerance judgment results are more in line with the subtle differences in the working conditions, which can make up for the defects of decision trees that are black and white and rigid boundary condition judgments; 2) The weights and scoring tables are preset and fixed when the equipment is deployed at the factory and cannot be modified during operation. It retains the advantages of fixed parameters and stable system operation, and achieves more flexible compromise decisions than rule decision trees by relying on multi-dimensional score quantification. It is suitable for the gradual accumulation of samples during the transition period when the data is not yet available. 3) The impact weight of four factors—substrate loss, fault size, production progress, and environmental interference—is intuitively quantified through quantitative scores. For example, priority is given to the two key indicators of substrate value and pulse fault, with secondary reference to the printing stage and interference situation. This anchors the core focus of production from a numerical perspective, and the fault tolerance trade-offs are more in line with the actual demands of industrial production to reduce losses and maintain capacity. 4) The calculation formula is unified to standardize the operation logic of all working conditions. The judgment process is reproducible and the logic is transparent, which facilitates later debugging and parameter optimization. The decision tree and AI agent are smoothly transferred in the stage of transitioning from a small number of samples to a sufficient number of samples, so as to achieve a smooth transition of three decision modes.

[0054] In an exemplary embodiment, intelligent decision-making is performed based on a decision-making pattern and the level of decision factors to obtain a target fault-tolerant action, specifically including: (1) When using an AI agent to query a pre-stored state-action value query table for decision-making, the formula is: (2) The substrate value level, lost pulse level, printing stage level, and interference intensity level are encoded as unique status codes; among them... For state coding; , , , The numerical indices are, in order, the base material value level, the lost pulse level, the printing stage level, and the interference intensity level. The selectable values ​​for all four types of numerical indices are 0, 1, and 2, respectively. , , , , .

[0055] (2) Query the status action value query table according to the status number, and read the expected revenue value corresponding to the three fault-tolerant actions of shutdown, reissue and page jump respectively. Select the fault-tolerant action with the largest expected revenue value as the initial fault-tolerant action. (3) Based on expected return and formula: (3) Calculate the confidence level of the initially selected fault-tolerant actions; where, Confidence level; This is the maximum expected return value among the expected return values, which is also the expected return value corresponding to the initial selection error-tolerant action; It is the second largest expected return value among the expected return values, and is the return value corresponding to other fault-tolerant actions; (4) Compare the confidence level with the preset confidence threshold (e.g., 0.15). If the confidence level is greater than or equal to the preset confidence threshold, the initial fault-tolerant action is determined as the target fault-tolerant action. If the confidence level is less than the preset confidence threshold, the action is rolled back and the rule decision tree is used to re-determine the target fault-tolerant action.

[0056] In one specific implementation, the above formula (2) can generate 81 unique state codes ranging from 0 to 80. The AI ​​agent has an internally built-in 81-row × 3-column state action value query table (Q table). Each row of the Q table corresponds to a state code from 0 to 80, and the three columns correspond to the expected benefit values ​​of three fault-tolerant actions: shutdown, reissue, and page jump. The expected benefit value can also be called the Q value. The Q table is not stored externally but is embedded in the AI ​​agent. During the first stage (rule decision tree) and the second stage (weighted scoring) of the device operation, the system stores the working condition experience into the experience pool after each abnormality is handled. When the experience pool accumulates to 500 data points, batch offline training is started to iteratively optimize and update the Q table parameters built into the agent.

[0057] The AI ​​agent retrieves three sets of action Q values ​​from the corresponding row of the Q-table based on the state codes obtained in real time, and selects the maximum value among the three values. The corresponding action is selected as the initial tolerance action, and the larger of the remaining two Q values ​​is the second-best Q value. The confidence level of the initial fault-tolerant action is calculated according to formula (3). The system presets the confidence judgment threshold to 0.15: when the calculated confidence level is ≥0.15, it means that the AI ​​selection is sufficiently certain and the initial action is directly output; when the confidence level is <0.15, it means that the benefits of the best action and the second best action are close and the agent's decision is ambiguous. The system automatically reverts to the rule decision tree to obtain the final fault-tolerant action.

[0058] In another specific implementation, an exemplary numerical illustration is provided in conjunction with the Q table: Table 5: Q Table 0 12.5 78.2 11.3 1 -8.3 85.6 22.1 ... … … … 79 92.1 -45.2 18.5 80 -12.4 25.7 79.9 When the state code is calculated When =0, query the data in the Q table for that row: Shutdown Q=12.5, Reissue Q=78.2, Page Skip Q=11.3; where (Reissued), the second largest value among the remaining values ​​is (Shutdown); If the confidence level is met, the AI ​​agent directly outputs a resend as the target fault-tolerant action.

[0059] The beneficial effects of this embodiment are as follows: 1) The Q-table is embedded in the AI ​​agent, and local table lookup does not require external data interaction. The industrial control hardware reads the table quickly, meeting the timing requirements of the printing system to issue fault-tolerant instructions in real time. The Q-table is trained offline based on historical samples accumulated during the early cold start and transition period, avoiding the problem of no training data and inability to use reinforcement learning in the early stage of equipment production. 2) A confidence judgment and decision tree fallback mechanism based on dual Q-values ​​is added. When the AI ​​agent faces the critical situation where the benefits of two actions are similar and the choice is shaky, it automatically switches to mature fixed rules to make up for the defects of decision-making errors in boundary scenarios of reinforcement learning models, and takes into account both intelligent self-optimization and system operation stability. 3) The four-dimensional parameter compression encoding is 81 groups of finite state codes, which greatly compresses the data volume of the Q-table, reduces the storage overhead and computing power consumption of the agent, and adapts to the resource limitations of embedded hardware in industrial inkjet equipment.

[0060] In an exemplary embodiment, the method further includes: After completing the target fault-tolerant action, record the substrate value level, lost pulse level, printing stage level, interference intensity level, selected target fault-tolerant action, and action execution result for this printing. The cumulative number of exceptions handled is incremented by one and persisted. If the decision-making mode corresponding to the AI ​​agent is used for this printing, the reward value for a single interaction is calculated according to the preset reward rules, and the state, action, reward and subsequent state are stored in the experience pool.

[0061] The preset reward rules include: a positive reward for successful printing, different negative rewards for error shutdown and scrapped substrate, and corresponding reward values ​​for successful reprint and normal continuation of printing after page skipping.

[0062] In practical implementation, the reward value is... Quantitatively: Successfully completed the printing task Error-induced shutdown (please confirm that calls can resume after shutdown). Substrate becomes unusable The reissue was successfully executed. The device resumed printing normally after the page skipped. The states stored in the experience pool as recorded here uniformly refer to the state codes generated through encoding. .

[0063] In one specific implementation, the scenario parameter is: the substrate value level is medium ( ), the level of lost pulse is level 2 ( The printing stage level is intermediate. The interference intensity level is slight. ).

[0064] First, calculate the current state code ID: The system is in Phase 3. After looking up the table, the AI ​​agent selects the fault-tolerant action: reissue.

[0065] Next, a re-send command was issued, the pulse distribution board filled in the missing pulses, and finally the re-send was successful, and printing continued smoothly.

[0066] Next, data recording will be performed: Data will be stored in the historical database. , , , Minor, action = reissue, execution result = success.

[0067] Update the count: cumulative number of anomalies The new values ​​are persistently stored in memory and used as the basis for subsequent determination of the three stages of cold start, transition, and optimization.

[0068] Reward Calculation: According to the rules, this reward... Collect the operating parameters of the equipment at the next moment and convert them into the next state code. .

[0069] Experience storage: storing a set of experience data ( Write the data into the Q-Learning exclusive experience pool; once the experience pool has accumulated 500 data entries, the system will start batch offline training and iteratively optimize the 81×3 Q-table built into the AI ​​agent.

[0070] In another specific implementation, same-state coding The AI ​​agent selected a shutdown action. After shutdown, a re-inspection revealed that the equipment had no hardware faults and could have continued printing, which was an incorrect shutdown. Execution result flag: Error and shutdown; Reward value: ; Pack( The data is stored in the experience pool for subsequent Q-table iterative training.

[0071] The beneficial effects of this embodiment are as follows: 1) The system retains the working conditions, actions, and results data for each abnormal closed-loop process, and the accumulated statistical data accurately updates the cumulative number of abnormalities D, ensuring that the system can accurately distinguish between the three decision-making stages of cold start, transition, and optimization, and realize automatic switching of decision-making modes; 2) Quadruple experience data is collected only during the AI ​​operation stage and stored in the experience pool. Combined with hierarchical quantitative reward and punishment rules, positive actions are rewarded with higher returns and erroneous actions are rewarded with negative returns, guiding Q-Learning to gradually select more economical and fault-tolerant solutions in subsequent training; 4) After the experience pool is full, batch offline training is performed, and real-time training is not performed during the device printing operation, avoiding online training from occupying the computing power of embedded devices and interfering with the normal printing sequence, thus balancing model iteration optimization and device operation stability.

[0072] like Figure 2 As shown, embodiments of the present invention also provide a synchronous signal fault-tolerant system for multi-head printing, used to implement the synchronous signal fault-tolerant method for multi-head printing in any of the above embodiments. The system may include: a host computer, a pulse distribution board, and multiple printhead control boards; the host computer establishes bidirectional communication connections with the pulse distribution board and each printhead control board respectively; the pulse distribution board communicates with a pre-set printhead control board; the multiple printhead control boards cooperate to form a multi-head control array, collaboratively completing multi-head synchronous printing operations; The host computer is used to obtain the original data for printing decisions and the cumulative number of abnormalities processed based on the abnormal printing detection results fed back by the printhead control board; the abnormal printing detection results are obtained based on the detection of the coded synchronization signal; the coded synchronization signal is generated by the pulse distribution board by distinguishing the original photoelectric sensor signal by parity sequence number and differential pulse width modulation; Discretely classify the raw data for printing decisions to obtain the decision factor levels, and determine the decision mode based on the cumulative number of anomaly handling. Based on the decision-making model and combined with the level of decision-making factors, intelligent decision-making is carried out to obtain the target fault-tolerant action; The printhead control board corresponding to the abnormal printing detection results executes the target fault-tolerant action.

[0073] Although the invention has been described herein in conjunction with various embodiments, those skilled in the art will understand and implement other variations of the disclosed embodiments by reviewing the accompanying drawings, the disclosure, and the appended claims in carrying out the claimed invention. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude a plurality. Although the invention has been described in conjunction with specific features and embodiments, it is apparent that various modifications and combinations can be made thereto without departing from the spirit and scope of the invention. Accordingly, this specification and the accompanying drawings are merely exemplary descriptions of the invention as defined by the appended claims and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the invention. Obviously, those skilled in the art can make various alterations and modifications to the invention without departing from its spirit and scope. Thus, if such modifications and modifications of the invention fall within the scope of the claims and their equivalents, the invention is also intended to include such modifications and modifications.

Claims

1. A fault-tolerant method for synchronization signals in multi-head printing, characterized in that, The methods include: Based on the received abnormal printing detection results, obtain the original data for printing decisions and the cumulative number of abnormalities handled; The abnormal printing detection results are obtained based on the detection of the encoded synchronization signal; The encoded synchronization signal is generated from the original photoelectric eye signal by parity number differentiation and differential pulse width modulation. The original data for printing decisions is discretely classified to obtain the decision factor levels, and the decision mode is determined based on the cumulative number of anomaly handling. Based on the decision-making model, and combined with the decision factor levels, intelligent decision-making is performed to obtain the target fault-tolerant action; The printhead control board corresponding to the abnormal printing detection result is controlled to perform the target fault-tolerant action.

2. The synchronous signal fault-tolerant method for multi-head printing according to claim 1, characterized in that, The raw data for printing decisions includes substrate value parameters, number of lost pulses, printing progress data, and signal error rate; The raw data for printing decisions is discretely classified into levels to obtain the corresponding decision factor levels, specifically including: Based on a preset mapping relationship, the substrate value parameter, the number of lost pulses, the printing progress data, and the signal bit error rate are mapped to levels to obtain the substrate value level, lost pulse level, printing stage level, and interference intensity level.

3. The synchronous signal fault-tolerant method for multi-head printing according to claim 2, characterized in that, The decision-making model is determined based on the cumulative number of anomalies handled, specifically including: When the cumulative number of exceptions handled is less than the first number, a preset rule decision tree is used for decision-making. When the cumulative number of exceptions handled is greater than or equal to the first number and less than the second number, a weighted scoring decision is adopted. When the cumulative number of exceptions handled is greater than or equal to the second number, the AI ​​agent is used to query the pre-stored state action value query table to make a decision.

4. The synchronous signal fault-tolerant method for multi-head printing according to claim 3, characterized in that, Based on the aforementioned decision-making model, and combined with the level of decision-making factors, intelligent decision-making is performed to obtain the target fault-tolerant action, specifically including: When using a preset rule decision tree, the decision is made according to the preset priority. If the value level of the base material is high, the target fault-tolerant action is determined to be shutdown. If the value level of the substrate is medium or low, and the level of the lost pulse is 1, the target fault-tolerant action is determined to be resend. If the substrate value level is medium or low, the lost pulse level is greater than 1, and the printing stage level is the final stage, the target fault-tolerant action is determined to be reissue. If the substrate value level is medium or low, the lost pulse level is greater than 1, the printing stage level is non-final stage, and the interference intensity level is severe interference, the target fault-tolerant action is determined to be shutdown. If any of the aforementioned conditions are not met, the target fault-tolerant action is determined to be a page jump.

5. The synchronous signal fault-tolerant method for multi-head printing according to claim 3, characterized in that, Based on the aforementioned decision-making model, and combined with the level of decision-making factors, intelligent decision-making is performed to obtain the target fault-tolerant action, specifically including: When using a weighted scoring system for decision-making, the individual scores for each fault-tolerant action corresponding to the base material value, lost pulse, printing stage, and interference intensity are retrieved from the four preset scoring tables and substituted into the comprehensive scoring formula: ； Calculate the combined scores for shutdown, resend, and page skipping respectively, and select the fault-tolerant action with the highest score as the target fault-tolerant action; among them, Characterizes fault-tolerant actions; Characterizes the value of the substrate; Characterizes lost pulses; Characterizing the printing stage; Characterizes the intensity of interference; The individual score corresponding to the error-tolerant action; , , , These are the preset weighting coefficients.

6. The synchronous signal fault-tolerant method for multi-head printing according to claim 3, characterized in that, Based on the aforementioned decision-making model, and combined with the level of decision-making factors, intelligent decision-making is performed to obtain the target fault-tolerant action, specifically including: When using an AI agent to query a pre-stored state-action value lookup table for decision-making, the formula is: ； The substrate value level, lost pulse level, printing stage level, and interference intensity level are encoded as unique status codes; among them... For state coding; , , , The numerical indices are, in order, the base material value level, the lost pulse level, the printing stage level, and the interference intensity level. The selectable values ​​for all four types of numerical indices are 0, 1, and 2. Based on the status number, query the status action value query table, and read the expected revenue value corresponding to the three fault-tolerant actions of shutdown, reissue, and page jump respectively. Select the fault-tolerant action with the largest expected revenue value as the initial fault-tolerant action. Based on expected returns and the formula: ； Calculate the confidence level of the initially selected fault-tolerant actions; where, Confidence level; The maximum expected return among the expected return values; It is the second largest expected return value; The confidence level is compared with a preset confidence threshold. If the confidence level is greater than or equal to the preset confidence threshold, the initial fault-tolerant action is determined as the target fault-tolerant action. If the confidence level is less than the preset confidence threshold, the process reverts to using a rule-based decision tree to re-determine the target fault-tolerant action.

7. The synchronous signal fault-tolerant method for multi-head printing according to claim 4, characterized in that, Executing the target fault-tolerant action specifically includes: When the target fault-tolerant action is to stop, a stop command is sent to the corresponding printhead control board, and the printhead control board stops printing ignition and retains the current printing position data. When the target fault-tolerant action is to reissue, a reissue command is sent to the printhead control board. The printhead control board sends a reissue request to the pulse distribution board based on the reissue command. The pulse distribution board reissues the corresponding sequence number pulse based on the reissue request and sends it to the corresponding printhead control board to resume printing. When the target error-tolerant action is page skipping, a page skipping command is sent to the printhead control board, and the printhead control board continues printing after skipping a specified number of pages based on the page skipping command.

8. The synchronous signal fault-tolerant method for multi-head printing according to claim 4, characterized in that, The method further includes: After completing the target fault-tolerant action, record the substrate value level, lost pulse level, printing stage level, interference intensity level, selected target fault-tolerant action, and action execution result for this printing. The cumulative number of exceptions handled is incremented by one and persisted. If the decision-making mode corresponding to the AI ​​agent is used for this printing, the reward value for a single interaction is calculated according to the preset reward rules, and the state, action, reward and subsequent state are stored in the experience pool.

9. The synchronous signal fault-tolerant method for multi-head printing according to claim 8, characterized in that, The preset reward rules include: a positive reward for successful printing, different negative rewards for error shutdown and substrate scrap, and corresponding reward values ​​for successful reprint and normal continuation of printing after page skipping.

10. A fault-tolerant synchronization signal system for multi-head printing, characterized in that, The system includes: Host computer, pulse distribution board, and multiple nozzle control boards; The host computer is used to obtain the original data for printing decisions and the cumulative number of abnormalities processed based on the abnormal printing detection results fed back by the printhead control board. The abnormal printing detection results are obtained based on the detection of the encoded synchronization signal; The encoded synchronization signal is generated by the pulse distribution board by distinguishing the original photoelectric eye signal by odd and even numbers and modulating the pulse width difference. Discretely classify the raw data for printing decisions to obtain the decision factor levels, and determine the decision mode based on the cumulative number of anomaly handling. Based on the aforementioned decision-making model, intelligent decision-making is performed by combining the levels of decision factors to obtain the target fault-tolerant action; The printhead control board corresponding to the abnormal printing detection result is controlled to perform the target fault-tolerant action.

Images