A Multi-Objective Scheduling Optimization Method and System for Green Printing Workshops Based on Big Data

By generating a degradation-coupled emission source strength correction vector and a degradation trend prediction model, the coupling relationship between equipment degradation and VOC emission source strength in the printing workshop was solved, achieving accuracy in concentration prediction and stability in scrap rate estimation, and ensuring the robustness and consistency of the scheduling scheme.

CN123070540APending Publication Date: 2026-11-10SUZHOU GUOAO PRINTING TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611152267.1
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-07-31
Publication Date
2026-11-10

AI Technical Summary

Technical Problem

Existing printing workshop scheduling methods fail to effectively address the coupling relationship between equipment degradation and VOC emission source intensity, leading to biased concentration predictions and inaccurate scrap rate estimates. Furthermore, scheduling schemes are difficult to maintain stability and consistency in dynamic environments.

Method used

By acquiring equipment degradation observation vectors and concentration observation vectors, a degradation coupled emission source strength correction vector is generated, the concentration response coefficient matrix is ​​corrected, and the degradation trend prediction model is updated and corrected online using Bayesian methods. Combined with three-dimensional weight adjustment, a multi-objective scheduling scheme is generated to ensure the robustness of concentration and scrap rate constraints and the stability of the scheduling scheme.

Benefits of technology

It improved the accuracy of concentration prediction, reduced the inaccuracy rate of scrap rate estimation, maintained the stability and execution consistency of the scheduling scheme in dynamic environments, and reduced disturbances in the production execution process.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123070540A_ABST
    Figure CN123070540A_ABST
Patent Text Reader

Abstract

This invention relates to the field of production scheduling technology in printing workshops, and discloses a multi-objective scheduling optimization method for green printing workshops based on big data. The method includes: synchronously acquiring operating condition parameter records and degradation observation vectors from a big data acquisition platform; generating a degradation-coupled emission source strength correction vector based on the operating condition parameter records and degradation observation vectors; online correcting the concentration response coefficient matrix based on the degradation-coupled emission source strength correction vector; performing Bayesian online updates to the initial degradation trend prediction model based on degradation residuals and operating condition adjustments; performing joint condition prediction on candidate production sequences based on the corrected degradation trend prediction model; generating candidate multi-objective scheduling schemes based on three-dimensional fusion priority scores; and generating a rolling closed-loop multi-objective optimized scheduling scheme based on scheduling oscillation feedback. The scheduling optimization method of this invention maintains the execution continuity of confirmed process arrangements during dynamic updates, reducing the disturbance to production execution caused by frequent changes.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The embodiments disclosed herein relate to the field of production scheduling technology in printing workshops, and in particular to a multi-objective scheduling optimization method for green printing workshops based on big data. Background Technology

[0002] With the continuous advancement of green manufacturing policies and the gradual tightening of VOCs emission control standards, printing workshops face the pressure of balancing multiple objectives—delivery efficiency, print quality, and environmental compliance—in terms of production scheduling. The widespread adoption of big data collection and industrial IoT technologies has made it possible to aggregate multi-source heterogeneous data such as equipment operating status and environmental concentration distribution in real time, providing a data foundation for intelligent scheduling oriented towards green constraints.

[0003] However, with the extension of printing equipment service life and the accumulation of production batches, equipment degradation increasingly interferes with scheduling decisions. On the one hand, the wear of ink rollers and the ink supply compensation effect caused by equipment degradation lead to deviations of actual VOC emission source strength from nominal values, causing systematic prediction biases in concentration response models based on fixed emission coefficients. On the other hand, the accumulation of degradation directly affects the scrap rate of printed products, and the distribution of the scrap rate varies with different production sequences, resulting in a strong coupling relationship between production scheduling decisions and quality risks. Existing scheduling methods typically treat equipment degradation and VOC concentration constraints as independent constraints, lacking explicit modeling of the coupling relationship between the two. This leads to varying degrees of decline in the accuracy of concentration prediction and scrap rate estimation in scenarios with high degradation levels or frequent environmental constraints triggering operational adjustments.

[0004] Furthermore, under the rolling scheduling framework, the arrangement differences caused by degradation prediction updates and concentration constraint fluctuations between adjacent time domains can lead to frequent changes in the scheduling scheme, thereby disturbing production execution processes such as material preparation and equipment switchover. Existing schemes lack quantitative assessment and proactive adjustment methods for scheduling stability, making it difficult to maintain the consistency of scheduling scheme execution in dynamic environments.

[0005] Current printing workshop scheduling methods fail to consider the coupling effect between equipment degradation and VOC emission intensity, as well as the reverse impact of operating condition adjustments on degradation rate prediction. This leads to a cumulative inconsistency between concentration prediction and degradation prediction in multi-constraint rolling scheduling scenarios, making it difficult to simultaneously guarantee the environmental compliance and quality robustness of the scheduling scheme. Therefore, a big data-based multi-objective scheduling optimization method for green printing workshops is urgently needed to address the aforementioned technical problems. Summary of the Invention

[0006] The summary section of this application is intended to provide a brief overview of the concepts, which will be described in detail in the detailed description section below. This summary section is not intended to identify key or essential features of the claimed technical solutions, nor is it intended to limit the scope of the claimed technical solutions.

[0007] To address the problems and shortcomings of existing technologies, the present invention aims to provide a multi-objective scheduling optimization method and system for green printing workshops based on big data. This method can be used to solve the technical problem that the coupling relationship between equipment degradation status and VOC emission concentration in printing workshops is not incorporated into scheduling decisions, leading to both the risk of emission constraint violation and inaccurate scrap rate estimation.

[0008] To achieve the above objectives, the present invention provides the following technical solution: As a first aspect of this application, the present invention discloses a multi-objective scheduling optimization method for green printing workshops based on big data, comprising the following steps: Step S1: Obtain the degradation observation vector of each device, the concentration observation vector of multiple monitoring points, and the operating condition parameter records; Step S2: Based on the recorded operating parameters and the degradation observation vector, generate a degradation coupled emission source strength correction vector using the regression relationship between the degradation increment and the ink supply compensation amount; Step S3: After superimposing the degradation coupled emission source strength correction vector onto the nominal emission source strength, calculate the nominal predicted concentration vector, and based on the residual between the concentration observation vector and the nominal predicted concentration vector, generate a pure environmental drift correction response coefficient matrix by online correction of the concentration response coefficient matrix using weighted least squares. Step S4: The operating condition adjustment amount is incorporated into the Bayesian prior distribution as a comprehensive operating condition correction factor for the degradation rate, and the initial degradation trend prediction model is updated online using Bayesian methods to generate a corrected degradation trend prediction model. Step S5: Based on the corrected degradation trend prediction model and the pure environmental drift correction response coefficient matrix, perform joint condition prediction on the candidate production scheduling sequence to extract the concentration prediction peak and robust scrap rate estimate; Step S6: Combine the urgency score, emission impact score and degradation risk score of each process into a comprehensive scheduling priority score according to the three-dimensional adjustable weights, insert the processes into the time axis according to priority order, and simultaneously verify the concentration constraint and scrap rate constraint to generate candidate multi-objective scheduling schemes; Step S7: Calculate the scheduling oscillation of the candidate multi-objective scheduling scheme. If the scheduling oscillation exceeds the stability threshold, adjust the three-dimensional adjustable weights and re-execute the process arrangement. After iterative convergence, lock the current window front-end decision and output the rolling closed-loop multi-objective optimized scheduling scheme.

[0009] Preferred, the step of generating a degradation-coupled emission source strength correction vector based on the recorded operating parameters and the degradation observation vector, using the regression relationship between degradation increment and ink supply compensation, includes: Extract the actual ink supply from the operating condition parameter records, obtain the standard ink supply under standard operating conditions, and calculate the ink supply deviation. Obtain the degradation increment of a certain device in the previous batch cycle from the degradation observation vector; The amount of ink supply compensation caused by degradation is calculated using pre-fitted regression coefficients; The ink supply compensation amount is mapped to the emission source strength correction increment, and then the degradation coupled emission source strength correction vector is generated after being executed on all devices.

[0010] Preferably, the step of superimposing the degradation-coupled emission source strength correction vector onto the nominal emission source strength to calculate the nominal predicted concentration vector, and performing weighted least squares online correction on the concentration response coefficient matrix based on the residual between the concentration observation vector and the nominal predicted concentration vector, includes: Calculate the corrected emission source strength vector based on the aforementioned degraded coupled emission source strength correction vector; Calculate the nominal predicted concentration vector using the current concentration response coefficient matrix; Calculate the concentration residual vector for the modified emission source strength vector and the nominal predicted concentration vector; Using the concentration residual vector and the corrected emission source strength vector as inputs, the concentration response coefficient matrix is ​​corrected online using the weighted least squares method, and the pure environmental drift corrected response coefficient matrix is ​​output. The concentration response coefficient matrix is ​​a diagonal matrix, and its diagonal elements are determined based on the sensor measurement accuracy and historical residual variance of each monitoring point.

[0011] Preferably, the step of incorporating the operating condition adjustment amount as a comprehensive operating condition correction factor for the degradation rate into the Bayesian prior distribution and performing Bayesian online updates to generate a corrected degradation trend prediction model includes: The degradation rate offset is calculated based on the degradation increment and the degradation rate prediction value of the initial degradation trend prediction model. Extract the actual operating speed adjustment ratio and ink supply adjustment ratio due to VOCs constraints from the scheduling execution record of the previous time domain, and calculate the comprehensive operating condition correction factor. The Bayesian prior distribution mean parameter of the degradation rate is corrected by combining the comprehensive working condition correction factor; The modified prior distribution and the degradation rate offset are combined to perform a Bayesian online update, which outputs the posterior estimate of the degradation rate and its posterior confidence interval. The corrected degradation trend prediction model is generated based on the posterior estimate of the degradation rate and the current degradation observation.

[0012] Preferred, the joint conditional prediction of candidate production scheduling sequences based on the corrected degradation trend prediction model and the pure environmental drift corrected response coefficient matrix includes: For an order at a certain position in the candidate production sequence, calculate its expected degradation contribution increment on a certain device; The degradation contribution increment of each order is gradually accumulated according to the position order of the candidate production sequence to generate a conditional degradation prediction curve; The predicted values ​​of degradation-sensing emission source intensity at each time point are calculated based on the conditional degradation prediction curve. Substitute the degradation sensing emission source strength of all equipment at each time point into the pure environmental drift correction response coefficient matrix, calculate the degradation sensing concentration prediction value of each monitoring point and extract the concentration prediction peak value, and at the same time calculate the robust scrap rate estimate based on the upper bound of the confidence interval of the conditional degradation prediction curve.

[0013] Priority is given to the fusion of urgency scores, emission impact scores, and degradation risk scores of each process into a comprehensive scheduling priority score using three-dimensional adjustable weights, including: The overall scheduling priority score for each process is calculated and expressed as follows: ;in, 、 and Adjustable weights for urgency score, emission impact score, and degradation risk score, respectively, and satisfying the following conditions: Urgency rating The emission impact score is determined based on the remaining time between the delivery deadline of the order and the current time. The concentration increment is determined by calculating the proportion of the VOCs concentration emission limit based on the predicted value of the degradation-sensed emission source intensity corresponding to the execution time of the process, using the pure environmental drift correction response coefficient matrix; degradation risk score. The score is determined based on the ratio between the upper bound of the conditional degradation prediction confidence interval corresponding to the execution time of the process and the equipment degradation threshold; all three scores are normalized and then taken as values ​​from... Interval.

[0014] Prior to this, the scheduling oscillation degree of the candidate multi-objective scheduling scheme is calculated, expressed as: ; in, This represents the number of overlapping processes between the current candidate multi-objective scheduling scheme and the previous time-domain output scheme. For the first The planned execution times of overlapping processes in the current scheme. This refers to the planned execution time of this process in the previous time-domain scheme. For the first The weight coefficients of each overlapping process and satisfying The weighting coefficient is determined based on the degree of confirmation of the process in the previous time domain. Processes that have issued pre-notification to the material preparation stage are assigned a higher weight than processes that are in the pre-scheduling stage.

[0015] Prior to this, after the iteration converges, the decision at the beginning of the current window is locked, and a rolling closed-loop multi-objective optimization scheduling scheme is output, including: After iterating until the scheduling oscillation does not exceed the stability threshold or reaches the preset maximum number of iterations, the process arrangement within the most recent batch interval in the current window is locked as a deterministic execution decision, and the subsequent processes are retained as pre-scheduling states; The actual degradation monitoring data, VOCs concentration data and operating parameters continuously collected during the batch execution of the deterministic execution decision are fed back to the generation step of the degradation coupled emission source strength correction vector and the Bayesian online update step at the start of the next rolling scheduling time domain to form a rolling closed loop.

[0016] As a second aspect of this application, the present invention also discloses an electronic device, comprising: At least one processor, and a memory communicatively connected to said at least one processor; The memory stores instructions that can be executed by the at least one processor, which are then executed by the at least one processor to enable the at least one processor to perform the steps of the above-described multi-objective scheduling optimization method for a green printing workshop based on big data.

[0017] As a third aspect of this application, the present invention also discloses a computer storage medium storing a computer program thereon, characterized in that the computer program, when executed by a processor, implements the steps of the above-described multi-objective scheduling optimization method in a green printing workshop based on big data.

[0018] Compared with the prior art, the beneficial effects of the present invention are as follows: This invention aims to provide a multi-objective scheduling optimization method for green printing workshops based on big data. By independently estimating the ink supply compensation caused by degradation from the degradation increment and mapping it to the emission source strength correction increment, a degradation-coupled emission source strength correction vector is generated. This ensures that the concentration residual on which the online correction of the concentration response coefficient matrix depends does not include the emission source strength offset component caused by equipment degradation. Therefore, the corrected concentration response coefficient matrix can more accurately reflect pure environmental drift factors such as changes in workshop ventilation conditions, reducing the systematic bias in concentration prediction. By incorporating the operating condition adjustment triggered by VOCs constraints in the previous time domain into the Bayesian prior distribution mean parameter of the degradation rate as a comprehensive operating condition correction factor, the degradation trend prediction model can perceive the actual impact of environmental constraints on equipment operating conditions. In scenarios where VOCs constraints frequently trigger operating condition adjustments, the cumulative bias in degradation prediction is suppressed, ensuring consistency between the degradation prediction channel and the concentration prediction channel in rolling iterations. By combining the predicted emission source strength of degradation perception with the pure environmental drift correction response coefficient matrix for concentration prediction of candidate production scheduling sequences, and calculating the robust scrap rate estimate based on the upper bound of the conditional degradation prediction confidence interval, the scheduling scheme has a certain robust margin for both concentration constraints and scrap rate constraints under degradation uncertainty. By introducing a scheduling oscillation index to quantitatively assess the scheduling differences between adjacent time domains, and by adjusting the proportion of urgency scores in the three-dimensional weights to re-execute the process scheduling when the oscillation exceeds the stability threshold, the scheduling scheme maintains the execution continuity of confirmed process arrangements during dynamic updates, reducing the disturbance of frequent changes to the production execution process. Attached Figure Description

[0019] The accompanying drawings, which form part of this application, are used to provide a further understanding of the application and to make other features, objects, and advantages of the application more apparent. The illustrative embodiments and descriptions of this application are used to explain the application and do not constitute an undue limitation of the application. In the drawings: Figure 1 This is a flowchart of a multi-objective scheduling optimization method for a green printing workshop based on big data, according to an embodiment of the present invention. Figure 2 This is a schematic diagram comparing the degradation observation values ​​of each device with the emission source strength correction increment in an embodiment of the present invention; Figure 3 This is a schematic diagram comparing the observed and predicted VOCs concentrations at each monitoring point according to an embodiment of the present invention. Figure 4 This is a schematic diagram comparing the prior predicted values ​​and Bayesian posterior estimated values ​​of the degradation rate of each device in an embodiment of the present invention. Figure 5 This is a schematic diagram showing the trend of each degree of degradation prediction value and concentration peak value in the candidate production sequence according to an embodiment of the present invention; Figure 6This is a schematic diagram of the composition and sorting results of the three-dimensional fusion priority scores of each process in an embodiment of the present invention; Figure 7 This is a schematic diagram of the overall data flow architecture of the rolling closed-loop multi-objective optimization scheduling system according to an embodiment of the present invention; Figure 8 This is a schematic diagram of the time deviation and weight distribution of each overlapping process in the scheduling oscillation calculation of an embodiment of the present invention; Figure 9 This is a schematic diagram comparing the nominal emission source strength of each device with the corrected emission source strength in an embodiment of the present invention. Detailed Implementation

[0020] Embodiments of this disclosure will now be described in more detail with reference to the accompanying drawings. While some embodiments of this disclosure are shown in the drawings, it should be understood that this disclosure can be implemented in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of this disclosure. It should be understood that the accompanying drawings and embodiments of this disclosure are for illustrative purposes only and are not intended to limit the scope of protection of this disclosure.

[0021] It should also be noted that, for ease of description, only the parts relevant to the invention are shown in the accompanying drawings. Unless otherwise specified, the embodiments and features described in this disclosure can be combined with each other. Example

[0022] This invention discloses a multi-objective scheduling optimization method for green printing workshops based on big data. The following will describe this disclosure in detail with reference to the accompanying drawings and embodiments. (Refer to...) Figures 1 to 9 As shown, the specific steps include: Step S1: Obtain the degradation observation vector of each device, the concentration observation vector of multiple monitoring points, and the operating condition parameter records; Step S2: Based on the recorded operating parameters and the degradation observation vector, generate a degradation coupled emission source strength correction vector using the regression relationship between degradation increment and ink supply compensation amount; Step S3: After superimposing the degradation coupled emission source strength correction vector onto the nominal emission source strength, calculate the nominal predicted concentration vector. Based on the residual between the concentration observation vector and the nominal predicted concentration vector, generate the pure environmental drift correction response coefficient matrix by online correction of the concentration response coefficient matrix through weighted least squares. Step S4: The operating condition adjustment amount is incorporated into the Bayesian prior distribution as a comprehensive operating condition correction factor for the degradation rate, and the initial degradation trend prediction model is updated online using Bayesian methods to generate a corrected degradation trend prediction model. Step S5: Based on the corrected degradation trend prediction model and the pure environmental drift correction response coefficient matrix, perform joint conditional prediction on the candidate production scheduling sequence to extract the concentration prediction peak and robust scrap rate estimate. Step S6: The urgency score, emission impact score and degradation risk score of each process are integrated into a comprehensive scheduling priority score according to the three-dimensional adjustable weights. The processes are inserted into the time axis in priority order and the concentration constraints and scrap rate constraints are verified simultaneously to generate candidate multi-objective scheduling schemes. Step S7: Calculate the scheduling oscillation of the candidate multi-objective scheduling scheme. If the scheduling oscillation exceeds the stability threshold, adjust the three-dimensional adjustable weights and re-execute the process arrangement. After iterative convergence, lock the current window front-end decision and output the rolling closed-loop multi-objective optimized scheduling scheme.

[0023] Regarding step S1, the degradation observation vectors of each device, concentration observation vectors of multiple monitoring points, and operating parameter records are obtained. Specifically, at the beginning of each rolling scheduling time domain, degradation monitoring data of each device and real-time VOCs sensor concentration readings of each monitoring point are synchronously acquired from the big data acquisition platform. Simultaneously, the actual operating parameter records of each device from the previous batch are acquired, generating degradation observation vectors, concentration observation vectors, and operating parameter records. For example, a green printing workshop includes... Taiwanese printing equipment and VOCs concentration monitoring points, among which This represents the total number of printing machines. This represents the total number of VOCs concentration monitoring points. Degradation monitoring data includes registration deviation, ink density shift, and ink roller wear for each device, which are then normalized using Z-scores to generate a dimensionless dataset. Degenerate observation vector , of which Each component Characterizing the first The overall degree of degradation of the equipment at the current moment. The real-time concentration readings of VOCs sensors at each monitoring point are generated in the following dimensions: Concentration observation vector , of which Each component Characterizing the first The current VOCs concentration values ​​at each monitoring point The operating parameters recorded include the actual operating speed, actual ink supply, and actual drying temperature of each piece of equipment in the previous batch. The big data acquisition platform directly reads the equipment operating parameters and degradation index sensor data from the control systems of each printing device via industrial Ethernet. It also obtains VOCs concentration readings and temperature and humidity data at each monitoring point through an independently deployed environmental monitoring subsystem. The above data is synchronously written into a unified time-series database at the start of the rolling scheduling time domain for subsequent steps to read.

[0024] Regarding step S2, based on the operating condition parameter records and degradation observation vector, a degradation-coupled emission source strength correction vector is generated using the regression relationship between degradation increment and ink supply compensation. Specifically, based on the operating condition parameter records, the deviation between the actual ink supply and the standard ink supply for each device is extracted. The degradation-induced ink supply compensation is calculated using the regression relationship between the ink supply deviation and the degradation increment. This ink supply compensation is then mapped to the degradation correction increment of the emission source strength for each device, generating the degradation-coupled emission source strength correction vector.

[0025] For example, for the first The device extracts the actual ink supply volume of the previous batch from the operating parameter records. The standard ink supply volume under standard operating conditions is obtained based on the type of substrate and ink type corresponding to the equipment. Calculate ink supply deviation ,in For the first The actual ink supply volume of the previous batch on the machine. For the first Standard ink supply volume of the equipment under standard operating conditions. This represents the deviation between the actual ink supply and the standard ink supply. Simultaneously, the first [unclear] is obtained from the degradation observation vector. The degradation increment of the equipment in the previous batch cycle The degradation increment is the degradation observation value at the current moment. Degenerate observations at the start of the previous rolling time domain The difference between them, i.e. .in, For the first Degraded observations of the instrument at the current moment , For the first The degraded observations of the instrument at the start of the previous rolling time domain. and All values ​​have been standardized to dimensionless values ​​using Z-score in step S1, therefore It is also a dimensionless degenerate increment.

[0026] Using regression coefficients pre-fitted based on historical degradation trajectories and ink supply compensation records The amount of ink supply compensation caused by degradation is calculated and expressed as: ;in For the first The linear regression coefficient between the degradation increment of the equipment and the ink supply compensation amount is measured in units of ink supply (ml / sheet) to the dimensionless degradation increment, and is determined by fitting historical data. For the first The dimensionless degradation increment of the equipment in the previous batch cycle; For degradation increment The corresponding expected ink supply compensation amount, in milliliters per sheet. Ink supply deviation. It also includes degradation-induced compensation components and non-degradation components such as operator manual adjustments, and the regression coefficients Its function is to independently estimate the amount of ink supply compensation caused by degradation from the dimension of degradation increment, rather than directly using the full value of ink supply deviation, thereby avoiding the introduction of non-degradation factors into emission source strength correction.

[0027] Furthermore, the amount of ink supply compensation caused by degradation The mapping to the emission source strength correction increment is expressed as: ; in For the first The VOCs emission coefficient per unit ink supply of the ink used in the equipment is expressed in grams of VOCs per milliliter of ink supply and is determined by the calibration of the physicochemical parameters of the ink. The amount of ink supply compensation due to degradation is expressed in milliliters per sheet. For the first The incremental increase in emission source strength due to degradation of the equipment, expressed in grams of VOCs per sheet. After each device performs the above calculations, the generated dimension is: The degradation-coupled emission source strength correction vector is expressed as: ; in, 、 、 The first Taiwan, No. Taiwan, No. The incremental emission source strength correction caused by the degradation of the equipment in Taiwan. This indicates transpose. The degradation-coupled emission source strength correction vector characterizes the offset of VOCs emission source strength of each device caused by the ink supply compensation effect due to device degradation, and is used to separate the interference of degradation factors from the concentration residual in step S3.

[0028] Regarding step S3, the nominal predicted concentration vector is calculated by superimposing the degradation-coupled emission source strength correction vector onto the nominal emission source strength. Based on the residual between the observed concentration vector and the nominal predicted concentration vector, the concentration response coefficient matrix is ​​corrected online using weighted least squares to generate a pure environmental drift correction response coefficient matrix. Specifically, the nominal predicted concentration vector is recalculated after superimposing the degradation-coupled emission source strength correction vector onto the nominal emission source strength. The residual between the observed concentration vector and the corrected nominal predicted concentration vector is calculated, and the concentration response coefficient matrix is ​​corrected online using the weighted least squares method to generate a pure environmental drift correction response coefficient matrix after removing the interference of degradation factors.

[0029] For example, nominal emission source strength vector For dimension The vector, where the first... Each component Indicates the first The VOCs emission source strength of the equipment under standard operating conditions and without degradation offset is expressed in grams of VOCs per sheet. The degradation-coupled emission source strength correction vector generated in step S2 is then used. Superimposed on the nominal emission source strength, the corrected emission source strength vector is obtained, expressed as: ; in, For dimension The nominal emission source strength vector. For dimension The degradation coupling emission source strength correction vector. This is the corrected emission source strength vector of dimension , in grams of VOCs per sheet. The response coefficient matrix is ​​used in this process. (dimension is) , of which Line number Column elements Characterizing the first Unit emission source intensity of equipment in the first The concentration response values ​​caused by each monitoring point are expressed in units of concentration (gVOCs / sheet). The corrected nominal predicted concentration vector is calculated and represented as follows: ; in, For dimension The concentration response coefficient matrix, For dimension The corrected emission source strength vector For dimension The nominal predicted concentration vector, with units equal to the concentration observation vector. Consistent. The concentration observation vector... With the nominal predicted concentration vector The residual is expressed as: ; in, For dimension Concentration observation vector, For dimension The nominal predicted concentration vector, For dimension The concentration residual vector. Since the emission source strength drift caused by degradation has been incorporated into the nominal predicted emission source strength calculation through the degradation coupled emission source strength correction vector in step S2, the concentration deviation reflected by the concentration residual vector only comes from the real changes in environmental factors such as changes in workshop ventilation conditions and fluctuations in meteorological diffusion parameters, and does not include the false concentration deviation introduced by the emission source strength shift caused by equipment degradation.

[0030] Using concentration residual vector As the basis for correcting the concentration response coefficient matrix, the weighted least squares method is used to modify the concentration response coefficient matrix. Online correction is performed, where the input of the weighted least squares method is the concentration residual vector. and corrected emission source strength vector The output is the corrected concentration response coefficient matrix. The weighting matrix is The diagonal matrix is ​​determined by the sensor measurement accuracy and historical residual variance of each monitoring point. Monitoring points with higher measurement accuracy and smaller historical residual variance are assigned larger weights. The corrected concentration response coefficient matrix. This is the pure environmental drift correction response coefficient matrix after removing the interference of degradation factors.

[0031] Regarding step S4, the operating condition adjustment amount is incorporated as a comprehensive operating condition correction factor for the degradation rate into the Bayesian prior distribution, and the initial degradation trend prediction model is updated online using Bayesian methods to generate a corrected degradation trend prediction model. Specifically, the degradation rate offset is extracted by calculating the residual between the incremental degradation observation data and the predicted values ​​of the initial degradation trend prediction model in the previous batch. Simultaneously, the actual operating condition adjustment amount implemented in the previous time domain due to VOCs constraints is incorporated as an operating condition correction factor for the degradation rate into the Bayesian prior distribution, and an online Bayesian update is performed to generate a corrected degradation trend prediction model that incorporates the influence of operating conditions. The online Bayesian update is a conventional algorithm, with the corrected prior distribution parameters and degradation rate offset as inputs, and the posterior estimate of the degradation rate and its posterior confidence interval as outputs.

[0032] For example, for the first For each device, the degradation rate offset is calculated as follows, and expressed as: ; in, The first one obtained in step S2 The dimensionless degradation increment of the equipment in the previous batch cycle; The time interval between two adjacent rolling time domains, in hours; For the initial degradation trend prediction model, at the start time of the previous time domain, the first... The predicted rate of degradation of the equipment is given in dimensionless degradation per hour. This represents the offset between the actual and predicted degradation rates, expressed as dimensionless degradation amount per hour. The operating condition adjustments triggered by VOCs concentration constraints in the previous time domain are incorporated into the Bayesian prior distribution of the degradation rate. Specifically, the [missing information] is extracted from the scheduling execution records of the previous time domain. The actual operating speed adjustment ratio of the equipment due to VOCs constraints Ink supply adjustment ratio Calculate the comprehensive operating condition correction factor: ; in, The sensitivity coefficient for the effect of operating speed variation on degradation rate. The sensitivity coefficient for the effect of ink supply variation on degradation rate. For the first The actual operating speed adjustment ratio of the equipment due to VOCs constraints. For the first The actual ink supply adjustment ratio implemented by the equipment due to VOCs constraints. This is the comprehensive operating condition correction factor. , All are dimensionless proportional quantities. and It is also a dimensionless sensitivity coefficient, determined by the correspondence between historical operating condition changes and degradation rate changes, thereby enabling... Maintain dimensionless values. The mean parameter of the Bayesian prior distribution of the degradation rate is obtained from the original predicted values. Revised to This allows the prior distribution to reflect the actual impact of environmental constraints on the degradation rate in the previous time domain.

[0033] When the first in the previous time domain The equipment was reduced in speed because the VOCs concentration was close to the emission limit. When the combined operating condition correction factor is reached, A negative value corresponds to a downward adjustment of the prior distribution mean, reflecting the inhibitory effect of reduced speed operation on the degradation rate; in another example, when the frequency of processing highly abrasive substrates increases due to production sequence adjustments ( Close to zero but When the combined operating condition correction factor is reached, For positive values, the mean of the prior distribution is adjusted upwards accordingly. This is reflected in the corrected prior distribution and the degradation rate offset. Under the combined effect of these mechanisms, Bayesian online updates are performed, outputting the corrected posterior estimate of the degradation rate. and its posterior confidence interval. Based on the posterior estimate of the degradation rate. and current degradation observations , generate the first The corrected degradation trend prediction curve for the device within the current scrolling window describes the predicted degradation level and its confidence bandwidth for the device at future time steps. After each device performs the above-mentioned Bayesian online update, a corrected degradation trend prediction model is generated.

[0034] It should be noted that the comprehensive operating condition correction factor is used. The model can perceive the actual impact of environmental constraints on equipment operating conditions in the previous time domain, avoiding cumulative bias in degradation predictions due to neglecting changes in operating conditions in scenarios where VOCs constraints frequently trigger operating condition adjustments. This correction, together with the correction of the degradation to the concentration response coefficient matrix in steps S2 to S3, forms a two-way coupled correction structure, ensuring that the two prediction channels maintain prediction consistency in multiple iterations of rolling scheduling.

[0035] Regarding step S5, based on the corrected degradation trend prediction model and the pure environmental drift corrected response coefficient matrix, joint conditional prediction is performed on the candidate production scheduling sequences to extract the concentration prediction peak and robust scrap rate estimate. Specifically, the order queue to be scheduled and the substrate characteristic parameters and ink type parameters of each order in the current rolling window are obtained. Based on the corrected degradation trend prediction model, the expected contribution increment of each order to equipment degradation is calculated for different candidate production scheduling sequences, combined with the pure environmental drift corrected response coefficient matrix. The conditional degradation prediction curve and degradation sensing concentration prediction distribution are jointly calculated for each candidate production sequence. The concentration prediction peak value of each monitoring point and the robust scrap rate estimate of each order are extracted.

[0036] For example, the current scrolling window contains There are 10 orders awaiting production, among which This represents the total number of orders awaiting production. The substrate characteristics parameters for each order include substrate type, surface roughness, and ink absorption coefficient; the ink parameters include ink type, standard ink film thickness, and VOCs volatility coefficient. For a candidate production sequence (in Indicates ranking in Order number for each production position. 、 、 They were ranked in the 1st place respectively , No. , No. (Order number of each production position), and calculate the expected incremental contribution of each order to the degradation of each piece of equipment in the order of the candidate production scheduling sequence. Single-digit orders In the The expected degradation contribution increment on the equipment Calculate as follows: ;in, The output of step S4 The a posteriori estimate of the degradation rate of the equipment, in dimensionless degradation per sheet; For orders The number of prints, in sheets; For orders The substrate ink absorption coefficient is a dimensionless correction coefficient that reflects the relative impact of different substrates on the equipment degradation rate. For orders In the The expected incremental contribution to the degradation of the equipment when executed on the machine is a dimensionless degradation amount. The degradation contribution increments of each order are then sorted according to the candidate production scheduling sequence. The order of positions is gradually accumulated onto the correction degradation trend prediction curve to generate candidate production scheduling sequences. Next Conditional degradation prediction curves of the equipment ,in This represents the expected execution time for each position in the candidate production schedule.

[0037] It should be noted that the posterior estimate of the degradation rate in step S4 The unit is dimensionless degradation per hour, while in the above formula... With the number of printed sheets Multiplication requirements The unit is dimensionless degradation per sheet. To ensure dimensional consistency, the posterior estimate of the degradation rate is used. Before substituting into the above formula, we need to use the first... The equipment is in the candidate production sequence The nominal operating speed below (Unit: sheets per hour) Convert this to a degradation rate based on sheets. ,in It is determined by the operating parameters of the candidate production sequence.

[0038] Furthermore, based on the conditional degradation prediction curve For each corresponding time point, the predicted degradation level is used to correct the regression coefficients for the degradation trend prediction using the mean parameter of the Bayesian prior distribution incorporated in step S2. and emission coefficient Calculate the corresponding time for each digit. Degradation sensing emission source intensity prediction values ​​for the equipment: ; in, For the first The nominal emission source strength of the equipment is expressed in grams of VOCs per sheet. This is the reference degradation level corresponding to the nominal emission source strength of the equipment, and it is a dimensionless degradation amount. Candidate production sequence Next The equipment is at all times The conditional degradation prediction value is a dimensionless degradation quantity; This is the dimensionless offset of the current degradation level relative to the reference degradation level; The combined dimension is grams of VOCs per sheet (and) (with consistent dimensions) The unit remains grams of VOCs per sheet; For the first The equipment is at all times The predicted values ​​of emission source intensity at each time point. Substituting the degradation sensing emission source strength vector of the equipment into the pure environmental drift correction response coefficient matrix Calculate the predicted degradation sensing concentration values ​​for each monitoring point at each corresponding time: ; in, This represents the total number of printing machines. Corrected response coefficient matrix for pure environmental drift The Middle Line number Column elements, For the first The equipment is at all times The predicted value of emission source intensity for degradation perception. Candidate production sequence Next Each monitoring point at time The predicted concentration of degradation sensing. For each monitoring point in the candidate production sequence. The maximum value of the predicted concentration at all time points is taken to obtain the predicted peak concentration at that monitoring point. .

[0039] Meanwhile, based on the conditional degradation prediction curve The degradation level corresponding to the upper bound of the confidence interval is used to calculate the robust scrap rate estimate for each order under the conservative degradation estimate, taking into account the substrate characteristics and print quality standards for each order. Specifically, regarding orders Take its execution time The corresponding upper bound of the conditional degradation prediction confidence interval ,in For orders The expected execution time, Candidate production sequence Next The equipment is at all times The upper bound of the conditional degradation prediction confidence interval is used. Based on the pre-calibrated relationship between the degree of equipment degradation and the rate of non-compliance in print quality, the scrap rate estimate at this conservative degradation level is obtained by looking up a table or interpolation. This scrap rate estimate is the robust scrap rate estimate. This is reflected in orders under the upper bound of degradation uncertainty. The expected percentage of print quality defects due to equipment degradation. It should be noted that this is a robust scrap rate estimate. The calculation depends on the order. The assigned execution device number Candidate production sequence The correspondence between each order and the equipment has been determined. Therefore, for the same order under different candidate production schedules, the upper bound of the conditional degradation prediction confidence interval varies depending on the equipment allocated. The differences also mean that the robust scrap rate estimate varies with the candidate production sequence.

[0040] Regarding step S6, the urgency score, emission impact score, and degradation risk score of each process are fused into a comprehensive scheduling priority score using three-dimensional adjustable weights. Processes are then inserted into the timeline in priority order, and concentration constraints and scrap rate constraints are verified simultaneously to generate candidate multi-objective scheduling schemes. Specifically, the comprehensive scheduling priority score of each process is calculated. The urgency score, emission impact score based on degradation-perceived emission source strength, and degradation risk score based on the upper bound of the conditional degradation prediction confidence interval are fused using three-dimensional adjustable weights. Processes are inserted into the timeline in descending order of priority. After each insertion, concentration constraints and scrap rate constraints are verified simultaneously. After a constraint violation delay, the process proceeds to the nearest feasible time slot, and candidate multi-objective scheduling schemes are output.

[0041] For example, for the first Each process, comprehensive scheduling priority score Calculate as follows: in, For the first The urgency rating of each process step For the first Emission impact score for each process step For the first Degradation risk score for each process; 、 and Adjustable weights for urgency, emissions impact, and degradation risk, respectively, and satisfying the following conditions: , For the first The overall scheduling priority score for each process. Urgency score. Emissions Impact Score and degradation risk score All three scores were normalized and then assigned values. Intervals are used to eliminate the impact of different physical quantity dimensions on the weighted summation degradation risk score. All three scores were normalized and then assigned values. The interval is used to eliminate the influence of different physical quantity dimensions on the weighted summation operation. The calculation method for each score is as follows: Urgency Score Based on the delivery deadline of the order to which this process belongs. With the current moment The remaining time between them is determined, specifically, ,in For the first Delivery deadline for orders belonging to each process. The current time represents the time domain length of the current scrolling window; the shorter the remaining time, the higher the urgency score. The higher the score, the greater the emissions impact rating. Based on the predicted value of degradation sensing emission source intensity at the time of execution of this process calculated in step S5. The relationship between the concentration margin at each monitoring point and the concentration margin at each monitoring point was determined, specifically, ,in This represents the total number of printing machines. The first element in the pure environmental drift corrected response coefficient matrix Line number Column elements, For the first The equipment is at all times The predicted value of emission source intensity for degradation perception. For VOCs emission limits, the higher the proportion of the expected increase in concentration relative to the emission limit, the higher the emission impact score. The higher the score, the greater the risk of degradation. Based on the upper bound of the conditional degradation prediction confidence interval corresponding to the execution time of this process in step S5. With equipment degradation threshold The distance between them is determined, specifically, ,in For the first The equipment is at all times The upper bound of the confidence interval for conditional degradation prediction. For the first The degradation threshold of the equipment is a dimensionless degradation quantity, pre-calibrated and determined by the equipment maintenance procedures; the closer to the degradation threshold, the higher the degradation risk score. The higher the score, the better. The normalization of the above three scores ensures the overall scheduling priority score is achieved. The weighted summation operation is consistent in dimensions.

[0042] Based on comprehensive scheduling priority score Insert each process into the timeline in descending order of priority. After each process insertion, simultaneously verify the following two constraints: Concentration constraint verification is based on the superposition of degradation-sensing emission source strengths of all processes inserted on the current timeline and the pure environmental drift-corrected response coefficient matrix. Calculate the predicted concentration values ​​at each monitoring point and verify whether the predicted concentration values ​​at all monitoring points are lower than the VOCs emission limits. The scrap rate constraint verification is based on the upper bound of the conditional degradation prediction confidence interval corresponding to the cumulative degradation contribution of the inserted process. The robust scrap rate estimate of the order to which the current process to be inserted belongs is calculated, and the robust scrap rate estimate is verified to be lower than the preset scrap rate threshold.

[0043] If either the concentration constraint or the scrap rate constraint is violated, the current process is removed from the time slot and postponed to the nearest feasible time slot that satisfies all constraints. After all processes have been inserted, candidate multi-objective scheduling schemes are output.

[0044] Regarding step S7, the scheduling oscillation of the candidate multi-objective scheduling scheme is calculated. If the scheduling oscillation exceeds the stability threshold, the three-dimensional adjustable weights are adjusted and the process arrangement is re-executed. After iterative convergence, the current window's front-end decision is locked, and a rolling closed-loop multi-objective optimized scheduling scheme is output. Specifically, the scheduling oscillation of the candidate multi-objective scheduling scheme is calculated. The difference between the planned execution time of each process in the candidate multi-objective scheduling scheme and the planned execution time of the corresponding process in the previous time-domain output scheme is calculated, and the weighted root mean square is calculated. If the scheduling oscillation exceeds the stability threshold, the weight ratio of the urgency score in the three-dimensional weights is adjusted, and the process arrangement is re-executed. After iteration until the scheduling oscillation converges, the current window's front-end decision is locked. The actual degradation data, concentration data, and operating parameters during the execution period are fed back to the dual-channel correction input, and a rolling closed-loop multi-objective optimized scheduling scheme is output.

[0045] For example, scheduling oscillations Calculated as follows, expressed as: ;in, This represents the number of overlapping processes between the current candidate multi-objective scheduling scheme and the previous time-domain output scheme. For the first The planned execution times of overlapping processes in the current candidate multi-objective scheduling scheme; This refers to the planned execution time of this process in the previous time domain output scheme; For the first The weighting coefficients of the overlapping processes satisfy the following conditions: to make Maintain and 、 The same time units, where ; To regulate oscillations, its unit is... and The time units are consistent. Weighting coefficients. The weighting is determined based on the degree of confirmation of the process in the previous time domain: processes that have already issued pre-notifications to the material preparation stage are assigned higher weights, while processes still in the pre-scheduling stage are assigned lower weights. This makes the scheduling oscillation index more sensitive to process changes that have a significant impact on actual execution. If the scheduling oscillation... Exceeding the preset stability threshold Then increase the weight of the urgency score in the three-dimensional weighting. The proportion of emissions impact is reduced proportionally. and degradation risk weight Based on the adjusted weights, the process arrangement in step S6 is re-executed to generate a new candidate multi-objective scheduling scheme, and the scheduling oscillation is recalculated.

[0046] It should be noted that increasing the weighting of urgency makes the priority ranking of each process closer to an order primarily based on delivery deadlines. This order exhibits smaller variations between adjacent rolling time domains (delivery deadlines are exogenous determinants), thus reducing the scheduling discrepancies caused by prediction fluctuations in environmental and degradation constraints between adjacent time domains. By iteratively adjusting the weights and repeatedly calculating the scheduling oscillations, until... Alternatively, after reaching the preset maximum number of iterations, the process arrangement in the first part of the current window (i.e., within the most recent batch interval) is locked as a deterministic execution decision, while the subsequent processes are retained as pre-scheduled processes to be re-evaluated in the next rolling time domain.

[0047] After the deterministic execution decision is locked, during the execution of this batch, actual degradation monitoring data of each device, VOCs concentration data of each monitoring point, and operating parameters such as actual operating speed and ink supply of each device are continuously collected. At the beginning of the next rolling scheduling time domain, the above data is fed back to the degradation coupled emission source strength correction entry in step S2 to update the degradation coupled emission source strength correction vector, and is also fed back to the Bayesian online update entry in step S4 to provide new degradation residual observations and operating condition adjustment records, forming a rolling closed loop. Steps S1 to S7 are repeated in each rolling time domain, continuously outputting a rolling closed-loop multi-objective optimization scheduling scheme.

[0048] like Figures 2 to 9 As shown, in the rolling scheduling scenario of a green printing workshop, a printing production base (hereinafter referred to as "Plant A") is equipped with... Taiwanese printing equipment (equipment 1 to equipment 4) and At the start of the rolling scheduling time domain of batch A in 20XX, the big data acquisition platform synchronously reads the registration deviation, ink density offset and ink roller wear of each device through industrial Ethernet at each VOCs concentration monitoring point (monitoring point 1 to monitoring point 3), and obtains the real-time concentration readings of the VOCs sensors at each monitoring point from the environmental monitoring subsystem. At the same time, it extracts the actual operating speed, actual ink supply and actual drying temperature of each device in the previous batch and writes them into a unified time series database.

[0049] The original degradation monitoring indicators of each device were standardized using Z-scores to generate a degradation observation vector with a dimension of 4. The VOCs sensor readings from the three monitoring points generate a concentration observation vector with a dimension of 3. The operating parameters record includes the actual operating speed, actual ink supply, and actual drying temperature of each device. As shown in Tables 1 and 2 below, Table 1 shows the degradation monitoring data and concentration monitoring data at the beginning of batch A, and Table 2 shows the operating parameters record of the previous batch.

[0050]

[0051] Table 1

[0052] Table 2 At the start of batch A, the difference between the actual ink supply and the standard ink supply for each device is extracted from the time-series database. Simultaneously, the degradation increment of each device compared to the start of the previous rolling time domain is obtained from the degradation observation vector. Using pre-calibrated regression coefficients Calculate the amount of ink supply compensation caused by degradation Then through the ink emission coefficient Mapped to the emission source strength correction increment of each device Finally, a 4-dimensional correction vector for the degradation coupled emission source strength is generated. .

[0053] Taking device 3 as an example, its previous batch actual ink supply volume ml / sheet, standard ink supply Milliliters per sheet, ink supply deviation ml / sheet; degraded observation at the start of the previous rolling time domain Current degradation observations Degradation increment Using regression coefficients Milliliters per sheet, calculate the ink supply compensation amount due to degradation. Using ink emission coefficients gram VOCs / ml ink supply, calculate emission source strength correction increment Table 3 shows the calculation process for the degradation-coupled emission source strength correction vector.

[0054]

[0055] Table 3 This generates a correction vector for the intensity of the degraded coupled emission sources. gram of VOCs per sheet of printing.

[0056] The modified emission source strength vector is obtained by superimposing the modified emission source strength vector onto the nominal emission source strength vector. Using the current concentration response coefficient matrix (dimension is) Calculate the nominal predicted concentration vector , and concentration observation vector The concentration residual vector is obtained by calculating the residual. , and then and As input, the pure environment drift corrected response coefficient matrix is ​​output using the weighted least squares method. The diagonal elements of the weighting matrix are determined based on the sensor measurement accuracy and historical residual variance of each monitoring point. The nominal emission source strength of each device is... VOCs / printed sheet, corrected emission source strength is Taking monitoring point 2 as an example, its predicted concentration value is... Calculated mg / Measured concentration mg / residual mg / The residual only reflects environmental factors such as changes in ventilation conditions; emission offsets caused by degradation have been excluded. The weighted least squares method, with weights of 0.85, 0.90, and 0.80 for the three monitoring points (calibrated according to measurement accuracy), was applied to... Perform online correction output Table 4 shows a comparison between the concentration residuals and the predicted concentrations before and after correction.

[0057]

[0058] Table 4 At the start of batch A, the degradation increment of each device is extracted from the time-series database. Adjacent rolling time domain time interval (In this example) (hours) and the degradation rate prediction model's predicted values ​​for each device in the previous time domain. Calculate the degradation rate offset Simultaneously, extract the operating speed adjustment ratio of each device triggered by VOCs constraints from the scheduling execution record of the previous time domain. Ink supply adjustment ratio Using sensitivity coefficient 、 Calculate the comprehensive operating condition correction factor The prior distribution mean is changed from Revised to Perform Bayesian online updates and output posterior estimates of the degradation rate for each device. and its posterior confidence interval.

[0059] Taking device 3 as an example, its degradation rate offset is In the previous time domain, device 3 was slowed down due to VOCs concentrations approaching emission limits. , The comprehensive operating condition correction factor is The prior distribution mean is corrected to a dimensionless degradation rate per hour, and the overall degradation rate shift is calculated. After performing the Bayesian online update, the posterior estimate is output. Dimensionless degradation rate per hour. Table 5 shows a summary of the inputs and outputs for the Bayesian online update.

[0060]

[0061] Table 5 The current scrolling window contains There are three pending orders (Order A, Order B, and Order C), and the ink absorption coefficient of the substrate for each order. and print quantity All are known. For the candidate production scheduling sequence... The degradation contribution increment of each order is progressively added to the correction degradation trend prediction curve according to its position, thus generating the conditional degradation prediction curve for each piece of equipment. Using regression coefficients Emission coefficient and nominal emission source strength Calculate the predicted values ​​of degradation-sensing emission source intensity at each time point. Substitute the pure environmental drift correction response coefficient matrix Calculate the predicted degradation sensing concentration at each monitoring point and extract the peak concentration prediction; simultaneously, calculate the robust scrap rate estimate for each order based on the upper bound of the conditional degradation prediction confidence interval. .

[0062] Taking equipment 3 executing order A (rank 1) as an example, its nominal operating speed Printings per hour, converting the posterior estimate of the degradation rate into a degradation rate based on printings. Dimensionless degradation per sheet, the number of sheets printed for order A. Printing sheet, ink absorption coefficient of the printing substrate The incremental contribution of degradation is Predicted value of conditional degradation at corresponding time point Reference degradation level The predicted value of the emission source intensity for degradation sensing is: Table 6 shows the incremental contribution of order degradation and key prediction results for each candidate production scheduling sequence.

[0063]

[0064] Table 6 Based on the degradation sensing concentration prediction and robust scrap rate estimate obtained in step S5, the comprehensive scheduling priority score is calculated for each process within the current rolling window. The weight configuration is as follows , The urgency level of each process is satisfied. Emissions Impact Score and degradation risk score All values ​​are normalized and then taken as follows: Interval.

[0065] Taking the process corresponding to order B (process B) as an example, the difference between its delivery deadline and the current time is 1.2 hours, and the time domain length is... Hours, urgency rating From step S5, it can be seen that the predicted peak concentration at the time of execution of this process accounts for the VOCs emission limit ( mg / The proportion is Emissions impact score; upper bound of conditional degradation prediction confidence interval; equipment 3 degradation threshold; degradation risk score. The overall priority score is .according to Insert each process step into the timeline sequentially from highest to lowest, and verify the concentration constraint simultaneously after each insertion (the predicted concentration value at each monitoring point must be below 300 mg / L). The process is constrained by both a scrap rate constraint (the robust scrap rate must be lower than a preset threshold of 3.0%) and a defect rate constraint. All processes meet the constraints and do not need to be delayed to a feasible time slot. Candidate multi-objective scheduling schemes are then output. Table 7 shows the comprehensive scheduling priority score and constraint verification results for each process.

[0066]

[0067] Table 7 The scheduling oscillation is obtained by calculating the difference between the planned execution time of each process in the candidate multi-objective scheduling scheme and the planned execution time of the corresponding process in the previous time-domain output scheme, and then calculating the weighted root mean square. In this example, the current candidate scheme differs from the previous time-domain scheme. For overlapping processes, the process weight coefficient is determined based on the degree of confirmation: Process B has already issued a pre-notification to the material preparation stage, and its weight... Processes A and C are in the pre-production scheduling stage, with weights... ,satisfy Dispatch oscillation Preset stability threshold Hourly, current scheduling oscillation Hourly satisfaction Without adjusting the weights, the process arrangement in the first half of the current window (within the most recent batch interval) is directly locked as a deterministic execution decision, while the subsequent processes are retained as pre-production scheduling.

[0068] During batch execution, the big data acquisition platform continuously collects actual degradation monitoring data of each device, VOCs concentration data of each monitoring point, and operating parameters such as actual operating speed and ink supply of each device. At the start of the next rolling scheduling time domain (Batch B), the above data is synchronously fed back to the degradation coupled emission source strength correction entry in step S2 and the Bayesian online update entry in step S4, updating the degradation coupled emission source strength correction vector and correcting the degradation trend prediction model, forming a rolling closed loop, and continuously outputting the rolling closed loop multi-objective optimization scheduling scheme. As shown in Table 8, the scheduling oscillation calculation and closed loop feedback data are summarized.

[0069]

[0070] Table 8 To implement the above embodiments, this application also discloses an electronic device. The electronic device may include a processing unit (e.g., a central processing unit, a graphics processing unit, etc.) that can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) or a program loaded from a storage device into a random access memory (RAM). Various programs and data required for the operation of the electronic device are also stored in the RAM. The processing unit, ROM, and RAM are interconnected via a bus. An input / output (I / O) interface is also connected to the bus. Typically, the following devices can be connected to the I / O interface: input devices including, for example, a touchscreen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; output devices including, for example, a liquid crystal display (LCD), speaker, vibrator, etc.; storage devices including, for example, magnetic tape, hard disk, etc.; and communication devices. The communication device allows the electronic device to communicate wirelessly or wiredly with other devices to exchange data. Although electronic devices with various devices are shown, it should be understood that it is not required to implement or possess all of the shown devices. More or fewer devices may be implemented or possessed alternatively.

[0071] In particular, according to some embodiments of this disclosure, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, some embodiments of this disclosure include a computer program product comprising a computer program carried on a computer storage medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from a storage device, or installed from a ROM. When the computer program is executed by a processing device, it performs the functions defined above in the methods of some embodiments of this disclosure.

[0072] It should be noted that the computer storage medium in some embodiments of this disclosure may be a computer-readable signal medium or a computer-readable storage medium, or any combination thereof. A computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of a computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof.

[0073] In some embodiments of this disclosure, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device. In some embodiments of this disclosure, a computer-readable signal medium can include a data signal propagated in baseband or as part of a carrier wave, carrying computer-readable program code. Such propagated data signals can take various forms, including but not limited to electromagnetic signals, optical signals, or any suitable combination thereof. A computer-readable signal medium can also be any computer storage medium other than a computer-readable storage medium, which can send, propagate, or transmit a program for use by or in connection with an instruction execution system, apparatus, or device. The program code contained on the computer storage medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (radio frequency), etc., or any suitable combination thereof.

[0074] In other implementations, clients and servers may communicate using any currently known or future-developed network protocol such as HTTP (Hypertext Transfer Protocol), and may interconnect with digital data communication (e.g., communication networks) of any form or medium. Examples of communication networks include local area networks (“LANs”), wide area networks (“WANs”), the Internet (e.g., the Internet of Things), and end-to-end networks (e.g., ad hoc end-to-end networks), as well as any currently known or future-developed networks.

[0075] The aforementioned computer storage medium may be included within the aforementioned electronic device, or it may exist independently without being assembled into the electronic device. The aforementioned computer storage medium carries one or more programs, which, when executed by the electronic device, enable the electronic device to implement a multi-objective scheduling optimization method for green printing workshops based on big data.

[0076] Computer program code for performing operations of some embodiments of this disclosure can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0077] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present disclosure. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. Each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, may be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions. Units described in some embodiments of the present disclosure may be implemented in software or hardware. The described units may also be located in a processor, and the names of these units do not necessarily constitute a limitation on the unit itself.

[0078] The functions described above in this document can be performed, at least in part, by one or more hardware logic components. For example, exemplary types of hardware logic components that can be used, without limitation, include: Field Programmable Gate Arrays (FPGAs), Application-Specific Integrated Circuits (ASICs), Application Standard Products (ASSPs), System-on-Chip (SoCs), Complex Programmable Logic Devices (CPLDs), and so on.

[0079] All technologies not described in detail in this invention are existing technologies. The above descriptions are merely some preferred embodiments of this disclosure and explanations of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in the embodiments of this disclosure is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalent features without departing from the above-described inventive concept. For example, technical solutions formed by substituting the above-described features with (but not limited to) technical features with similar functions disclosed in the embodiments of this disclosure.

Claims

1. A multi-objective scheduling optimization method for green printing workshops based on big data, characterized in that, Includes the following steps: Step S1: Obtain the degradation observation vector of each device, the concentration observation vector of multiple monitoring points, and the operating condition parameter records; Step S2: Based on the recorded operating parameters and the degradation observation vector, generate a degradation coupled emission source strength correction vector using the regression relationship between the degradation increment and the ink supply compensation amount; Step S3: After superimposing the degradation coupled emission source strength correction vector onto the nominal emission source strength, calculate the nominal predicted concentration vector, and based on the residual between the concentration observation vector and the nominal predicted concentration vector, generate a pure environmental drift correction response coefficient matrix by online correction of the concentration response coefficient matrix using weighted least squares. Step S4: The operating condition adjustment amount is incorporated into the Bayesian prior distribution as a comprehensive operating condition correction factor for the degradation rate, and the initial degradation trend prediction model is updated online using Bayesian methods to generate a corrected degradation trend prediction model. Step S5: Based on the corrected degradation trend prediction model and the pure environmental drift correction response coefficient matrix, perform joint conditional prediction on the candidate production scheduling sequence to extract the peak concentration prediction and robust scrap rate estimate. Step S6: The urgency score, emission impact score and degradation risk score of each process are integrated into a comprehensive scheduling priority score according to the three-dimensional adjustable weights. The processes are inserted into the time axis in priority order and the concentration constraints and scrap rate constraints are verified simultaneously to generate candidate multi-objective scheduling schemes. Step S7: Calculate the scheduling oscillation of the candidate multi-objective scheduling scheme. If the scheduling oscillation exceeds the stability threshold, adjust the three-dimensional adjustable weights and re-execute the process arrangement. After iterative convergence, lock the current window front-end decision and output the rolling closed-loop multi-objective optimized scheduling scheme.

2. The multi-objective scheduling optimization method for green printing workshops based on big data according to claim 1, characterized in that, The step of generating a degradation-coupled emission source strength correction vector based on the recorded operating parameters and the degradation observation vector, using the regression relationship between degradation increment and ink supply compensation, includes: Extract the actual ink supply from the operating condition parameter records, obtain the standard ink supply under standard operating conditions, and calculate the ink supply deviation. Obtain the degradation increment of a certain device in the previous batch cycle from the degradation observation vector; The amount of ink supply compensation caused by degradation is calculated using pre-fitted regression coefficients; The ink supply compensation amount is mapped to the emission source strength correction increment, and then the degradation coupled emission source strength correction vector is generated after being executed on all devices.

3. The multi-objective scheduling optimization method for green printing workshops based on big data according to claim 2, characterized in that, The step of superimposing the degradation-coupled emission source strength correction vector onto the nominal emission source strength to calculate the nominal predicted concentration vector, and performing weighted least squares online correction on the concentration response coefficient matrix based on the residual between the concentration observation vector and the nominal predicted concentration vector, includes: Calculate the corrected emission source strength vector based on the aforementioned degraded coupled emission source strength correction vector; Calculate the nominal predicted concentration vector using the current concentration response coefficient matrix; Calculate the concentration residual vector for the modified emission source strength vector and the nominal predicted concentration vector; Using the concentration residual vector and the corrected emission source strength vector as inputs, the concentration response coefficient matrix is ​​corrected online using the weighted least squares method, and the pure environmental drift corrected response coefficient matrix is ​​output. The concentration response coefficient matrix is ​​a diagonal matrix, and its diagonal elements are determined based on the sensor measurement accuracy and historical residual variance of each monitoring point.

4. The multi-objective scheduling optimization method for green printing workshops based on big data according to claim 3, characterized in that, The step of incorporating the operating condition adjustment amount as a comprehensive operating condition correction factor for the degradation rate into a Bayesian prior distribution and performing Bayesian online updates to generate a corrected degradation trend prediction model includes: The degradation rate offset is calculated based on the degradation increment and the degradation rate prediction value of the initial degradation trend prediction model. Extract the actual operating speed adjustment ratio and ink supply adjustment ratio due to VOCs constraints from the scheduling execution record of the previous time domain, and calculate the comprehensive operating condition correction factor. The Bayesian prior distribution mean parameter of the degradation rate is corrected by combining the comprehensive working condition correction factor; The modified prior distribution and the degradation rate offset are combined to perform a Bayesian online update, which outputs the posterior estimate of the degradation rate and its posterior confidence interval. The corrected degradation trend prediction model is generated based on the posterior estimate of the degradation rate and the current degradation observation.

5. The multi-objective scheduling optimization method for green printing workshops based on big data according to claim 4, characterized in that: The joint conditional prediction of candidate production scheduling sequences based on the corrected degradation trend prediction model and the pure environmental drift corrected response coefficient matrix includes: For an order at a certain position in the candidate production sequence, calculate its expected degradation contribution increment on a certain device; The degradation contribution increment of each order is gradually accumulated according to the position order of the candidate production sequence to generate a conditional degradation prediction curve; The predicted values ​​of degradation-sensing emission source intensity at each time point are calculated based on the conditional degradation prediction curve. Substitute the degradation sensing emission source strength of all equipment at each time point into the pure environmental drift correction response coefficient matrix, calculate the degradation sensing concentration prediction value of each monitoring point and extract the concentration prediction peak value, and at the same time calculate the robust scrap rate estimate based on the upper bound of the confidence interval of the conditional degradation prediction curve.

6. The multi-objective scheduling optimization method for green printing workshops based on big data according to claim 5, characterized in that, The process of integrating the urgency score, emission impact score, and degradation risk score of each process into a comprehensive scheduling priority score using three-dimensional adjustable weights includes: The overall scheduling priority score for each process is calculated and expressed as follows: ;in, 、 and Adjustable weights for urgency score, emission impact score, and degradation risk score, respectively, and satisfying the following conditions: Urgency rating The emission impact score is determined based on the remaining time between the delivery deadline of the order and the current time. The concentration increment is determined by calculating the proportion of the VOCs concentration emission limit based on the predicted value of the degradation-sensed emission source intensity corresponding to the execution time of the process, using the pure environmental drift correction response coefficient matrix; degradation risk score. The score is determined based on the ratio between the upper bound of the conditional degradation prediction confidence interval corresponding to the execution time of the process and the equipment degradation threshold; all three scores are normalized and then taken as values ​​from... Interval.

7. The multi-objective scheduling optimization method for green printing workshops based on big data according to claim 6, characterized in that, The calculation of scheduling oscillation for the candidate multi-objective scheduling scheme is expressed as follows: ; in, This represents the number of overlapping processes between the current candidate multi-objective scheduling scheme and the previous time-domain output scheme. For the first The planned execution times of overlapping processes in the current scheme. This refers to the planned execution time of this process in the previous time-domain scheme. For the first The weight coefficients of each overlapping process and satisfying The weighting coefficient is determined based on the degree of confirmation of the process in the previous time domain. Processes that have issued pre-notification to the material preparation stage are assigned a higher weight than processes that are in the pre-scheduling stage.

8. The multi-objective scheduling optimization method for green printing workshops based on big data according to claim 7, characterized in that, After the iteration converges, the decision for the first segment of the current window is locked, and a rolling closed-loop multi-objective optimization scheduling scheme is output, including: After iterating until the scheduling oscillation does not exceed the stability threshold or reaches the preset maximum number of iterations, the process arrangement within the most recent batch interval in the current window is locked as a deterministic execution decision, and the subsequent processes are retained as pre-scheduling states; The actual degradation monitoring data, VOCs concentration data and operating parameters continuously collected during the batch execution of the deterministic execution decision are fed back to the generation step of the degradation coupled emission source strength correction vector and the Bayesian online update step at the start of the next rolling scheduling time domain to form a rolling closed loop.

9. An electronic device, characterized in that, include: At least one memory for storing computer-executable instructions; At least one processor, configured to run the computer-executable instructions, The computer-executable instructions are executed by the processor to implement the method according to any one of claims 1 to 8.

10. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions that, when executed by at least one processor, implement the method according to any one of claims 1 to 8.