Multi-color group cooperative scheduling method for printing machine based on ant colony algorithm

Abstract

The present application relates to the technical field of printing equipment control, in particular to a printing machine multi-color group cooperative scheduling method based on an ant colony algorithm.In the present application, a to-be-printed work order is disassembled into process flow transfer nodes and a node transfer path is established, combined with layout size parameters, target ink type, paper grammage, target printing quantity, hourly rated print volume and color group related timing information, an orderly scheduling diagram is constructed, and further weighted processing is performed on the plate changing time proportion, ink contamination risk proportion, paper running continuity proportion and drying waiting proportion to form a truncated limited initial weight that can participate in optimization, realizing unified quantitative representation of switching relationship, connection relationship and waiting relationship in the multi-color group printing process.The present application also continuously corrects the search direction of the virtual search individual around the total scheduling sequence total time through the cyclic optimization of the maximum and minimum ant colony algorithm, the extraction of the preferred scheduling sample and the update of the second transfer probability by the improved cross-entropy method.

Description

Technical Field

[0001] This invention relates to the field of printing equipment control technology, and in particular to a multi-color group collaborative scheduling method for printing presses based on ant colony algorithm. Background Technology

[0002] The multi-color group collaborative scheduling method for printing presses refers to the method used for task allocation, action coordination, process connection, timing arrangement and operation control of multiple printing color groups in the printing press during the operation process. It is usually used to manage the start and stop of color groups, ink and paper supply coordination, printing sequence organization, operation cycle arrangement and control command allocation in the multi-color printing process.

[0003] Traditional multi-color group collaborative scheduling methods for printing presses, while capable of unified management of task allocation, action coordination, process connection, timing arrangement, and operational control for multiple printing color groups, suffer from several drawbacks. These include: continuous switching between multiple batches of printing work orders; significant differences in plate size; frequent changes in ink type; large fluctuations in paper weight; and inconsistent preheating completion times for different color groups. These issues lead to problems such as scheduling order relying primarily on fixed experience, difficulty in quantifying the switching costs of adjacent work orders, and insufficient coordination between multiple color groups. While traditional scheduling methods can coordinate ink and paper supply, organize printing sequence, and arrange work cycle time, they also encounter issues when processes such as plate loading, impression processing, ink circulation, registration adjustment, and roll removal are performed continuously and alternately. These problems include unclear node connections, difficulty in simultaneously incorporating plate changeover and cleaning times into the scheduling criteria, insufficient utilization of color group windows, and scheduling results that tend to remain at a locally optimal state. Consequently, these issues impact total downtime, process waiting time, and the continuity of multi-color group collaboration. Summary of the Invention

[0004] To achieve the above objectives, the present invention adopts the following technical solution: a multi-color group collaborative scheduling method for printing presses based on ant colony algorithm, comprising the following steps: Obtain the printing work order data of the printing press, divide the process flow nodes and node transfer paths of the printing work order data, and combine them into an ordered scheduling diagram; Based on the changes in different print order data over time, determine the initial weights for truncating the transfer paths of nodes in the ordered scheduling graph. Based on the initial weights of the truncation limit, the virtual search individuals are set through the maximum and minimum ant colony algorithm, and the first transition probability of the virtual search individuals moving between adjacent process flow nodes is set. The process flow nodes are iteratively optimized to determine the boundary constraint guiding weights. The number of iterations of the current maximum and minimum ant colony algorithm is counted. When the number of iterations reaches the preset upper limit of the number of iterations, the preferred scheduling sample is selected, and the second transition probability derived from the specified work order data to be printed in the preferred scheduling sample is calculated. The first transition probability of the virtual search individual moving between adjacent process flow nodes is updated using the second transition probability. Combined with boundary constraints, the weights are guided to repeatedly execute node search exploration actions until the set maximum number of iterations is met, and the collaborative scheduling sequence of the printing press multi-color group is selected.

[0005] As a further aspect of the present invention, the step of acquiring the printing work order data of the printing press, dividing the process flow nodes and node transfer paths of the printing work order data, and combining them into an ordered scheduling diagram includes: Obtain the print order data for the multi-color printing press. The print order data includes the page size parameters, target color group number, target ink type, paper weight, target print quantity, and rated print volume per hour. Each work order to be printed in the data to be printed is decomposed into the printing process of the multi-color printing press into sequentially connected plate loading process nodes, printing operation process nodes, ink path circulation process nodes, registration adjustment process nodes, and roll removal and plate withdrawal process nodes. Each independent process node is arranged and combined according to the target color group number position to generate a scheduling node set. Each independent process node in the scheduling node set is mapped to a process flow node in the max-min ant colony algorithm for virtual optimization individuals to traverse and search. For combinations of different printing work orders, the roll removal and plate removal process node at the end of the preceding printing work order is connected to the plate loading process node at the beginning of the subsequent printing work order. Based on the sequential transfer order between adjacent process flow nodes, the node transfer path in the maximum and minimum ant colony algorithm is set, and the process flow nodes and node transfer paths are spliced ​​together to generate an ordered scheduling diagram.

[0006] As a further aspect of the present invention, determining the initial weight for truncating the transfer path of nodes in the ordered scheduling graph based on the changes in different print order data over time includes: The preceding and subsequent printing work orders connected by the transfer path of each node in the ordered scheduling diagram are determined. Historical statistical results matching the page size parameters of the preceding printing work order are obtained, and the empirical time consumption of the roll removal and plate withdrawal process corresponding to the preceding printing work order is determined. At the same time, historical statistical results matching the page size parameters of the subsequent printing work order are obtained, and the empirical time consumption of the printing plate loading process corresponding to the subsequent printing work order is determined. The empirical times of the two are added together to obtain the corresponding sequential plate change time. The sequential plate change time is normalized based on the preset maximum plate change time to obtain the plate change time ratio. Based on the target ink type of adjacent preceding and subsequent printing work orders, determine the ink consistency status between them, determine the ink switching time based on the ink consistency status, and perform normalization calculation on the ink switching time with the preset maximum cleaning time as the benchmark to obtain the ink contamination risk ratio. Calculate the absolute difference in paper weight between adjacent preceding and subsequent printing work orders to obtain the paper weight difference. Normalize the paper weight difference based on the preset maximum weight fluctuation to obtain the paper feed continuity ratio. Calculate the ratio of the target printing quantity to the rated printing quantity per hour to obtain the estimated printing time. Based on the target color group number and the estimated printing time, determine the departure time of the adjacent preceding printing work order from the current color group and the preheating completion time of the corresponding color group of the subsequent printing work order. Calculate the absolute difference between the two times to obtain the drying waiting time. Perform normalization calculation on the drying waiting time based on the preset maximum drying waiting time to obtain the drying waiting ratio. The corresponding weighting coefficients are configured for the plate change time ratio, the ink contamination risk ratio, the paper feed continuity ratio, and the drying waiting ratio, and a weighted summation calculation is performed to obtain the comprehensive heuristic coefficient. The ratio of the preset heuristic benchmark value to the comprehensive heuristic coefficient is calculated to obtain the original guiding value. The maximum weight boundary and the minimum weight boundary are set for the original guiding value to establish the truncation limit initial weight.

[0007] As a further aspect of the present invention, the step of determining the boundary constraint guiding weights by setting virtual search individuals and the first transition probability of virtual search individuals moving between adjacent process flow nodes through the minimax ant colony algorithm based on truncation-limited initial weights, and performing iterative optimization on process flow nodes includes: Define a set of virtual search individuals in the maximum and minimum ant colony algorithm, and the first transition probability of the virtual search individuals moving between adjacent process flow nodes. Combine the truncation limit initial weight and the first transition probability to calculate the path transfer tendency of the virtual search individuals between each process flow node. Based on the path transfer tendency, determine the color group window to which the printing work order will enter, and control the virtual search individual to traverse the process flow nodes in sequence according to the color group window to complete the path construction. After completing one path construction, generate the overall scheduling sequence, and summarize the total duration of each process flow node passed in the overall scheduling sequence to obtain the total time consumption of the overall scheduling sequence. All overall scheduling sequences are sorted in ascending order of total time consumption. The overall scheduling sequence ranked first is extracted as the optimal solution for the current round. The original guiding values ​​on the node transfer paths traversed by the optimal solution for the current round are summarized. Combined with a preset first fixed constant, the updated guiding weight is calculated. The updated guiding weight is truncated and restricted by the maximum weight boundary and the minimum weight boundary to generate boundary constraint guiding weight.

[0008] As a further aspect of the present invention, the step of counting the number of iterations of the current maximum-minimum ant colony algorithm, and when the number of iterations reaches a preset upper limit for iteration rounds, selecting a preferred scheduling sample, and calculating the second transition probability derived from the specified work order data to be printed in the preferred scheduling sample, including: Monitor the number of iterations of the current maximum and minimum ant colony algorithm. When the number of iterations reaches the preset upper limit of the number of iteration rounds, select the sequence that ranks in the top 10% from the historically retained optimal solutions of the current round based on the total time consumption of the overall scheduling sequence, and use it as the preferred scheduling sample. The probability of identical paper weight transition is obtained by statistically analyzing the ratio of the number of times that adjacent preceding and subsequent printing work orders have the same paper weight to the total number of adjacent orders in the preferred scheduling sample; the probability of same paper weight transition is obtained by the ratio of the number of times that adjacent preceding and subsequent printing work orders are in the same color group and the registration adjustment process time remains unchanged to the total number of adjacent orders; and the probability of same color group registration connection is obtained by the ratio of the number of printing work orders whose plate mounting time exceeds the preset benchmark threshold to the number of printing work orders whose plate mounting time exceeds the preset benchmark threshold to the total number of printing work orders whose plate mounting time exceeds the preset benchmark threshold. The improved cross-entropy method is applied to use the same weight transition probability, the same color group quasi-connection probability, and the mounting timeout avoidance probability as the probability distribution parameters of the current round. Combined with the preset smooth learning rate, the first product is calculated. At the same time, the learning rate difference is calculated according to the preset second fixed constant and the smooth learning rate. The product of the learning rate difference and the first transition probability retained in history is calculated to obtain the second product. The second transition probability is calculated based on the first product and the second product.

[0009] As a further aspect of the present invention, the first transition probability of the virtual search individual moving between adjacent process flow nodes is updated using the second transition probability, and the node search exploration action is repeatedly executed in conjunction with boundary constraints to guide the weights until the set maximum number of iterations is met. The selected collaborative scheduling sequence for the multi-color group of the printing press includes: The second transition probability is covered by the first transition probability of the virtual search individual moving between adjacent process flow nodes, and the transition probability distribution is updated. Based on the updated maximum and minimum ant colony algorithm, combined with the boundary constraints, the node search exploration action is repeatedly executed until the set maximum number of iterations is reached, and the exploration and optimization operation record is obtained. Based on the total time consumed by the overall scheduling sequence, the scheduling scheme with the shortest time consumption is selected from the optimal solution of each current round in the exploration and optimization operation record; Obtain the start time of each printing work order in the corresponding color group, the assigned color group number of each printing work order, and the node transfer path between adjacent preceding printing work orders and subsequent printing work orders. Then, splice the corresponding time nodes and positions to output the collaborative scheduling sequence of multiple color groups of the printing press.

[0010] Compared with the prior art, the advantages and positive effects of the present invention are as follows: In this invention, the printing work order is broken down into process flow nodes and node transfer paths are established. Combined with plate size parameters, target ink type, paper weight, target printing quantity, hourly rated printing volume, and color group-related timing information, an ordered scheduling graph is constructed. Furthermore, the proportions of plate change time, ink contamination risk, paper feed continuity, and drying waiting time are weighted to form initial weights that can participate in optimization. This achieves a unified quantitative representation of the switching, connection, and waiting relationships in the multi-color group printing process. This invention also uses a cyclical optimization algorithm based on the min-max ant colony algorithm, optimal scheduling sample extraction, and an improved cross-entropy method to update the second transition probability. This allows virtual search individuals to continuously correct their search direction around the total time consumed by the overall scheduling sequence, thereby enhancing the global coordination of the scheduling sequence, improving the tightness of color group connections, shortening downtime preparation time, reducing invalid waiting caused by plate change and cleaning, and improving the efficiency of continuous multi-color group operation on the printing press. Attached Figure Description

[0011] Figure 1 This is a schematic diagram of the main steps of the present invention. Detailed Implementation

[0012] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0013] Please see Figure 1 This invention provides a technical solution: a multi-color group collaborative scheduling method for printing presses based on ant colony algorithm, comprising the following steps: S1: Obtain the printing order data for the printing press, divide the process flow nodes and node transfer paths of the printing order data, and combine them into an ordered scheduling diagram, including: Obtain the print order data for the multi-color printing press. The print order data includes the page size parameters, target color group number, target ink type, paper weight, target print quantity, and rated print volume per hour. A multi-color printing press refers to equipment with multiple color work positions, capable of continuously completing printing operations at different color group positions; pending printing order data refers to production task data that has been registered and completed before entering the scheduling process; page size parameters refer to the long and short side dimensions of the page corresponding to the pending printing order; target color group number refers to the color group position number corresponding to the pending printing order after entering the printing press; target ink type refers to the type of ink registered for the pending printing order; paper weight refers to the mass of paper per unit area; target print quantity refers to the planned number of sheets to be printed for the pending printing order; rated hourly output refers to the standard hourly output registered for the pending printing order under current equipment conditions. During acquisition, the page size field in the prepress plate-making record, the work order field in the production scheduling table, the ink field in the ink requisition record, the paper field in the raw paper inspection record, and the rated output field in the equipment operation record are each mapped to the same work order number and then written into the same pending printing order data table. During the processing, records with empty work order numbers, missing page size parameters, missing target color group numbers, paper weight of 0, target print quantity of 0, and rated hourly print volume of 0 are deleted. Page sizes are then standardized to millimeters, paper weight to grams per square meter, target print quantity to sheets, and rated hourly print volume to sheets per hour. If the same work order number appears repeatedly in different record tables, the completeness of the fields and the order of registration are compared, retaining the record with complete fields and the later registration time. This process ensures that the fields of the same work order to be printed correspond in different data tables, avoiding errors in the subsequent breakdown of the same work order due to inconsistent units, missing fields, or duplicate records.

[0014] Each work order to be printed in the work order data is broken down into the printing process of the multi-color printing press into sequentially connected process nodes: plate loading process node, printing operation process node, ink path circulation process node, registration adjustment process node, and roll removal and plate withdrawal process node. Each independent process node is arranged and combined according to the target color group number position to generate a scheduling node set. Each independent process node in the scheduling node set is mapped to the process flow node in the max-min ant colony algorithm for virtual optimization individuals to traverse and search. The plate loading process node represents the operation stage of loading the printing plate corresponding to the current printing work order into the designated position of the color group's printing roller; the printing operation process node represents the operation stage of completing the printing of the paper through the corresponding color group; the ink circulation process node represents the operation stage of achieving stable ink delivery between the ink fountain, ink roller, and inking area; the registration adjustment process node represents the operation stage of adjusting the image and overprint positions to the registered state; the roller removal and plate removal process node represents the operation stage of releasing the printing state and removing the printing plate after the current printing work order is completed. During disassembly, the above five process nodes are written out sequentially for each printing work order, and each process node is bound to the target color group number corresponding to that printing work order; if multiple target color group numbers are registered for the same printing work order, the five process nodes corresponding to each color group are arranged in ascending order of target color group numbers; if only one target color group number is registered for the same printing work order, only the one set of process nodes corresponding to that target color group number is retained. Subsequently, each independent process node is written into the scheduling node table, which sequentially records the work order number, target color group number, process name, process sequence, and node number. Then, each independent process node in the scheduling node set is mapped to a process flow node in the minimax ant colony algorithm for virtual search individuals to traverse and search. The minimax ant colony algorithm here refers to a path search method that sets maximum and minimum boundaries for path weights, and then repeatedly constructs paths by multiple virtual search individuals while retaining the optimal sequence. A virtual search individual is a search object that moves sequentially between process flow nodes to form a complete node access order; a process flow node is a discrete node position that can be visited and connected one by one by a virtual search individual. The algorithm does not involve a training process because there is no training stage for sample labeling, parameter backpropagation, and convergence output in the current scenario. Instead, it is directly executed and applied based on the acquired work order data to be printed, the subsequently calculated path weights, and the transition probabilities. During execution, each process node is first written as a node number matrix in MATLAB, and then the relationship between nodes is written as an adjacency matrix. In subsequent steps, path search and probability update are performed on this matrix. MATLAB is chosen because it can complete table reading, matrix construction, probability update, path sorting and iterative recording in the same environment without having to repeatedly convert node data between multiple software programs.

[0015] For combinations of different printing work orders, connect the last roll removal and plate removal process node of the preceding printing work order with the first printing plate loading process node of the subsequent printing work order. Based on the sequential transfer order between adjacent process flow nodes, set the node transfer path in the maximum and minimum ant colony algorithm, splice the process flow nodes and node transfer paths, and generate an ordered scheduling diagram. When combining different printing work orders, the roll removal and plate-clearing process node at the end of the preceding printing work order is connected one by one to the plate-filling process node at the beginning of the subsequent printing work order. Here, the preceding printing work order refers to the work order that appears earlier in the current sequence, and the subsequent printing work order refers to the work order that is adjacent to it and follows it. During connection, first check if the preceding printing work order has generated a roll removal and plate-clearing process node, then check if the subsequent printing work order has generated a plate-filling process node; if both exist, establish a connection from the last node of the preceding printing work order to the first node of the subsequent printing work order. The node transfer path represents the sequential connection between two adjacent process flow nodes, recording both the fixed process sequence within the work order and the switching sequence between different work orders. The node transfer paths within a work order are kept in a fixed arrangement: the plate loading process node connects to the printing operation node, the printing operation node connects to the ink circulation process node, the ink circulation process node connects to the registration adjustment process node, and the registration adjustment process node connects to the roll removal and plate withdrawal process node. The node transfer paths between work orders are written into the path table in the manner of connecting the last node of the preceding work order to the first node of the subsequent work order. When writing to the path table, the starting node number, target node number, the number of the preceding work order to be printed, and the number of the subsequent work order to be printed are recorded for each entry. These records are then synchronously written into the adjacency matrix in MATLAB to generate an ordered scheduling graph. This process ensures that when subsequent virtual search entities move between process flow nodes, they can only access nodes in the actual existing sequential order, preventing skipping processes within a work order, entering subsequent nodes in reverse order, or creating broken connections between different work orders.

[0016] S2: Based on the changes in different print order data over time, determine the initial weights for truncating the transfer paths of nodes in the ordered scheduling graph, including: In the ordered scheduling diagram, the preceding and subsequent printing work orders connected by the transfer path of each node are determined. Historical statistical results matching the page size parameters of the preceding printing work order are obtained to determine the empirical time consumption of the roll removal and plate withdrawal process corresponding to the preceding printing work order. At the same time, historical statistical results matching the page size parameters of the subsequent printing work order are obtained to determine the empirical time consumption of the printing plate loading process corresponding to the subsequent printing work order. The empirical times of the two are added together to obtain the corresponding sequential plate change time. The sequential plate change time is normalized based on the preset maximum plate change time to obtain the plate change time ratio. For each node transfer path in the ordered scheduling diagram, determine its adjacent preceding and subsequent printing work orders, and read the historical statistical results matching the page size parameters of both. The historical statistical results come from the completed plate removal and plate mounting records in the equipment operation records. First, the plate removal records are filtered, retaining records with a long side deviation of no more than 10 mm and a short side deviation of no more than 10 mm. Then, records with missing start and end times, or start times later than end times are deleted. For each retained record, the duration of a single plate removal record is obtained by subtracting the start time from the end time. The filtered plate removal durations are sorted from smallest to largest, and the median value is taken as the empirical time consumption of the plate removal process corresponding to the preceding printing work order. The same process is applied to the plate mounting records, retaining records with a long side deviation of no more than 10 mm and a short side deviation of no more than 10 mm, and deleting records with abnormal start and end times. For each retained record, the duration of a single plate mounting record is obtained by subtracting the start time from the end time. The selected plate-mounting durations are sorted from smallest to largest, and the median value is used as the empirical time consumption for the plate-mounting process of subsequent printing work orders. The historical statistical results here refer to the statistical results obtained by sorting the records of completed work orders, selecting records with the same or similar conditions to the current node's transfer path, and then processing the duration field. The median value is used because it is in the middle of the sorted samples, reducing the impact of extremely large or small values ​​on the empirical time consumption. Next, the sequentially related plate-changing time is obtained by summing the empirical time consumption of the roll removal and plate-removal process corresponding to the previous printing work order and the empirical time consumption of the plate-mounting process corresponding to the subsequent printing work order. Then, the plate-changing time ratio is obtained by comparing the sequentially related plate-changing time with the preset maximum plate-changing time. The preset maximum plate-changing time is determined by taking the 95th percentile value after sorting the historical plate-changing time samples from smallest to largest, because this position covers most normal plate-changing scenarios while avoiding extremely large values ​​that directly raise the upper limit.

[0017] Based on the target ink types of adjacent preceding and subsequent printing work orders, the ink consistency status is determined. The ink switching time is then determined based on the ink consistency status. The ink switching time is normalized using a preset maximum cleaning time as a benchmark to obtain the ink contamination risk ratio. If the target ink types are the same, the ink consistency status is determined to be consistent; if the target ink types are different, the ink consistency status is determined to be inconsistent. The target ink types include ordinary ink, UV ink, and spot color ink. When the ink consistency status is inconsistent, the preset cleaning color group time is the ink switching time; when the ink consistency status is consistent, the ink switching time is zero. Ink consistency status refers to whether ink cleaning and switching are required when subsequent printing orders enter the same path after the previous printing order has finished its work. Then, the ink switching time is determined based on the ink consistency status. If the ink consistency status is consistent, the ink switching time is recorded as 0; if the ink consistency status is inconsistent, historical records with the same ink type in both the previous and subsequent printing orders are selected from the equipment cleaning records. Records with missing start and end cleaning times, or start times later than end times, are then deleted. For each retained record, the cleaning duration is obtained by subtracting the cleaning start time from the cleaning end time. These cleaning durations are sorted from smallest to largest, and the median value is taken as the ink switching time. The preset cleaning color group time is the ink switching time obtained from the cleaning records and written when the ink consistency status is inconsistent. Subsequently, the ink contamination risk ratio is obtained by comparing the ink switching time with the preset maximum cleaning time. The preset maximum cleaning time is determined by taking the 95th percentile value after sorting historical cleaning time samples from smallest to largest. This value covers most switching situations because the cleaning time varies greatly when switching between different ink types.

[0018] Calculate the absolute difference in paper weight between adjacent preceding and subsequent printing work orders to obtain the paper weight difference. Normalize the paper weight difference based on the preset maximum weight fluctuation to obtain the paper feed continuity ratio. The paper feed continuity ratio is calculated as the ratio of the paper weight difference to the preset maximum weight fluctuation. The preset maximum weight fluctuation is determined by taking the 95th percentile value of the weight difference samples of adjacent work orders that did not exhibit paper jams, skewness, or significant tension abnormalities in continuous operation records, sorted from smallest to largest. This value is close to the upper limit of the equipment's ability to maintain continuous paper feed. The paper feed continuity ratio here refers to the degree to which changes in the conditions of two adjacent work orders on the current node's transfer path affect the continuous paper feed status.

[0019] Calculate the ratio of the target printing quantity to the rated printing quantity per hour to obtain the estimated printing time. Based on the target color group number and the estimated printing time, determine the departure time of the adjacent preceding printing work order from the current color group and the preheating completion time of the corresponding color group of the subsequent printing work order. Calculate the absolute difference between the two times to obtain the drying waiting time. Perform normalization calculation on the drying waiting time based on the preset maximum drying waiting time to obtain the drying waiting ratio. Estimated printing time refers to the theoretical processing time required to complete the target print quantity under the standard output conditions of the equipment corresponding to the current printing work order. Then, based on the target color group number and the estimated printing time, the departure time of the adjacent preceding printing work order from the current color group and the preheating completion time of the corresponding color group for the subsequent printing work order are determined. The departure time of the preceding printing work order from the current color group is obtained by correlating the start time in the color group operation record with the estimated printing time. If the equipment record already directly records the end time, then the end time is used directly. The preheating completion time of the corresponding color group for the subsequent printing work order is read from the color group preheating record. The drying waiting time is then calculated by the absolute difference between the departure time of the preceding printing work order from the current color group and the preheating completion time of the corresponding color group for the subsequent printing work order. Here, the absolute value of the time difference is used as the drying waiting time, representing the waiting time formed during the color group handover stage between preceding and subsequent work orders. Subsequently, the drying waiting ratio is obtained by comparing the drying waiting time with the preset maximum drying waiting time. The preset maximum drying waiting time is determined by taking the 95th percentile value after sorting the historical adjacent work order handover waiting time samples from smallest to largest, because this position can cover the vast majority of color group handover waiting situations.

[0020] Weighted summation is performed on the proportions of plate change time, ink contamination risk, paper feed continuity, and drying wait time to obtain a comprehensive heuristic coefficient. The ratio of the preset heuristic benchmark value to the comprehensive heuristic coefficient is calculated to obtain the original guiding value. Maximum and minimum weight boundaries are set for the original guiding value to establish a truncation limit for the initial weight. The weight coefficient is allocated according to the degree of influence of the process switching operation corresponding to each proportion on the total downtime preparation time of the printing press. The longer the adjustment time required for the target process switching operation, the larger the corresponding weight coefficient. The weighting coefficients are assigned based on the degree of impact of each proportion of the process switching operation on the total downtime preparation time of the printing press. During processing, the time spent on plate change, cleaning, paper adjustment, and waiting are first extracted from historical work order switching records. Then, the proportion of each time spent in the total downtime preparation time is calculated. Subsequently, the average of samples with similar proportions is calculated. Finally, the four averages are uniformly corrected to a sum of 1, which are used as the weighting coefficients for plate change time, ink contamination risk, paper feed continuity, and drying waiting time, respectively. The longer the adjustment time corresponding to a certain item, the larger the average proportion of that item, and the larger the corresponding weighting coefficient. The comprehensive heuristic coefficient here refers to the comprehensive path evaluation value formed by the current node's transfer path in terms of plate change, ink switching, paper feed continuity, and waiting. The comprehensive heuristic coefficient is obtained by summing the products of the plate change time weighting coefficient and the plate change time proportion, the ink contamination risk weighting coefficient and the ink contamination risk proportion, the paper feed continuity weighting coefficient and the paper feed continuity proportion, and the drying waiting time weighting coefficient and the drying waiting proportion. Next, the initial guiding value is obtained by the ratio of the preset heuristic benchmark value to the comprehensive heuristic coefficient. The preset heuristic benchmark value is set to 1, which is used to convert the comprehensive heuristic coefficient into a guiding value that can be directly used for path comparison. Subsequently, a maximum weight boundary and a minimum weight boundary are set for the initial guiding value. The maximum weight boundary and the minimum weight boundary are determined by taking the 95th percentile and 5th percentile values ​​of the historical initial guiding value samples after sorting them from smallest to largest, respectively. If the initial guiding value is higher than the maximum weight boundary, it is written according to the maximum weight boundary; if the initial guiding value is lower than the minimum weight boundary, it is written according to the minimum weight boundary; when it is between the boundaries, the original value is maintained, thus establishing a truncation limit for the initial weight. The reason for setting a maximum weight boundary and a minimum weight boundary is that if particularly large and particularly small values ​​are not restricted, the subsequent virtual search individuals will over-concentrate on or over-ignore the selection of a few paths, which is not conducive to continuous searching among multiple paths.

[0021] S3: Based on the initial weights constrained by truncation, the virtual search individual is set using the minimax ant colony algorithm, along with the first transition probability of the virtual search individual moving between adjacent process flow nodes. The process flow nodes are then iteratively optimized to determine the boundary constraint guiding weights, including: Define a set of virtual search individuals in the max-min ant colony algorithm, and the first transition probability of the virtual search individuals moving between adjacent process flow nodes. Combine the truncation limit of the initial weight and the first transition probability to calculate the path transfer tendency of the virtual search individuals between each process flow node. Based on truncation-bound initial weights, a set of virtual search individuals in the minimax ant colony algorithm are defined in MATLAB, along with the first transition probability of each virtual search individual moving between adjacent process flow nodes. A virtual search individual represents a search object that sequentially visits process flow nodes on an ordered scheduling graph and gradually constructs a complete scheduling path. The number of virtual search individuals is set based on the total number of work orders to be printed and the total size of process flow nodes. First, the total size of process flow nodes is divided by the total number of work orders to be printed, and then the result is rounded up; if the result is less than 5, it is rounded to 5; if it is greater than 50, it is rounded to 50. This setting is to maintain sufficient search coverage when the node size is large, and to avoid excessive duplicate searches when the node size is small. The first transition probability represents the initial probability value of a virtual search individual moving from the current process flow node to an adjacent subsequent process flow node. When setting this probability, the first transition probability is evenly distributed among all feasible subsequent nodes of a given current process flow node. Then, combining the truncation-bound initial weights and the first transition probability, the path transition tendency of the virtual search individual between each process flow node is calculated. The path transfer tendency is calculated by multiplying the first transfer probability corresponding to a single adjacent path by the initial weight of the truncation constraint, and then dividing by the sum of the above products corresponding to all adjacent paths of the current node. Here, the path transfer tendency refers to the relative priority of a virtual search individual moving from the current process flow node to an adjacent subsequent process flow node. If a current process flow node has only one feasible adjacent path, the corresponding path transfer tendency is directly recorded as 1, because there are no other comparable destinations for the current node. If the denominator is 0, all adjacent paths are evenly distributed, because no differentiated tendency is formed among the paths.

[0022] Based on the path transfer tendency, determine the color group window to which the printing work order will enter. Control the virtual search individual to traverse the process flow nodes in sequence according to the color group window to complete the path construction. After completing one path construction, generate the overall scheduling sequence. Sum the duration of each process flow node passed in the overall scheduling sequence to obtain the total time consumption of the overall scheduling sequence. A color group window represents the time interval during which subsequent print orders are allowed to enter the corresponding process node for a specific target color group within a specified time period. When determining the color group window, the occupancy completion record of the current print order for each color group is read first, followed by the preheating completion record of the corresponding color group for subsequent print orders. The later of these two timestamps is taken as the starting point at which subsequent print orders can enter that color group. If the preceding process node is not yet complete, the color group cannot be entered even if it is idle. Similarly, if the target color group number does not match the color group number registered in the current print order, entry is also prohibited. After satisfying the three conditions of preceding node completion, color group availability, and matching target color group number, the node traversal order of virtual search individuals is arranged according to the path transfer tendency of adjacent paths. Each time a virtual search individual enters a process node, the duration of that node is added to the current path's cumulative time. Once all process nodes corresponding to a print order have been traversed, the virtual search individual enters the first process node of the next print order. After all accessible nodes have been traversed, a single overall scheduling sequence is formed. The overall scheduling sequence refers to the order in which all process flow nodes are accessed after a virtual search entity completes a full path construction. The total time of the overall scheduling sequence is obtained by summing the duration of each process flow node traversed in the overall scheduling sequence; that is, by adding up the durations of all process flow nodes traversed in the overall scheduling sequence. If there are no process flow nodes traversed in the overall scheduling sequence, a valid overall scheduling sequence is not formed; if a valid overall scheduling sequence is formed, it must have a first entry node, a last exit node, and a complete sequence of intermediate nodes.

[0023] All overall scheduling sequences are sorted in ascending order of total time consumption. The overall scheduling sequence ranked first is extracted as the optimal solution for the current round. The original guiding values ​​on the node transfer paths traversed by the optimal solution for the current round are summarized. Combined with the preset first fixed constant, the updated guiding weight is calculated. The updated guiding weight is truncated and restricted by the maximum weight boundary and the minimum weight boundary to generate the boundary constraint guiding weight. During sorting, only the total time of the overall scheduling sequence is compared from smallest to largest, without changing the existing node order within each overall scheduling sequence. After determining the optimal solution for the current round, the original guidance values ​​on the node transfer paths traversed by the optimal solution are summarized, and combined with a preset first fixed constant, the updated guidance weight is calculated. The first fixed constant is obtained by averaging the difference samples of the original guidance values ​​of the corresponding paths of historical better scheduling sequences and ordinary scheduling sequences, and is used to control the update magnitude in each round. The updated guidance weight is obtained by summing the original guidance value and the first fixed constant multiplied by the ratio of the current path duration to the total time of the overall scheduling sequence of the optimal solution for the current round. Here, the optimal solution for the current round refers to the one with the smallest total time among all overall scheduling sequences in the current round; the updated guidance weight refers to the path guidance value rewritten based on the original guidance value and the information of the optimal solution for the current round. Next, the update guiding weights are truncated using the maximum and minimum weight boundaries. If the update guiding weight is higher than the maximum weight boundary, it is written according to the maximum weight boundary; if the update guiding weight is lower than the minimum weight boundary, it is written according to the minimum weight boundary; if it is in between, the original value is retained, thus generating boundary constraint guiding weights. Boundary constraint guiding weights refer to the path guiding values ​​after this round of updates and again constrained by the maximum and minimum weight boundaries, used for the next round of path search.

[0024] S4: Count the number of iterations of the current maximum and minimum ant colony algorithm. When the number of iterations reaches the preset upper limit of iteration rounds, select the preferred scheduling sample and calculate the second transition probability derived from the specified work order data in the preferred scheduling sample, including: Monitor the current iteration count of the maximum and minimum ant colony algorithm. When the iteration count reaches the preset iteration round limit, select the sequence that ranks in the top 10% from the historically retained optimal solutions of the current round based on the total time consumption of the overall scheduling sequence, and use it as the preferred scheduling sample. The iteration round limit is set according to the total number of work orders to be printed and the total scale of process flow nodes. The iteration count is incremented by 1 after each round of virtual search for all feasible paths. The upper limit of the iteration count is set based on the total number of work orders to be printed and the total scale of process flow nodes. Specifically, the total number of work orders to be printed is multiplied by the total scale of process flow nodes, the product is divided by 10, and the result is rounded up. If the calculated value is less than 20, it is rounded to 20; if the calculated value is greater than 200, it is rounded to 200. This setting is because the total number of work orders to be printed and the total scale of process flow nodes together determine the possible permutation space; the larger the product, the more iteration counts are required. After reaching the upper limit of the iteration count, the historically retained optimal solutions for the current round are reordered in ascending order according to the total time consumption of the overall scheduling sequence. The sequences in the top 10% are then selected as the preferred scheduling samples. If the number corresponding to the top 10% is not an integer, it is rounded up to ensure that at least one preferred sequence is retained. Here, the preferred scheduling sample refers to the set of relatively good sequences that rank in the top 10% of the total time consumption of the overall scheduling sequence among the optimal solutions for the current round in multiple rounds. The reason why the first 10% is used as the sample retention ratio is that this interval is significantly smaller than the range of all retained sequences, which can highlight the characteristics of the better sequences, while still retaining a certain number of sequences for subsequent probability statistics, so that the statistical results will not fluctuate too much due to the small sample size.

[0025] In the statistically optimized scheduling sample, the ratio of the number of times that adjacent preceding and subsequent printing work orders have the same paper weight to the total number of adjacent orders is used to obtain the transition probability of the same paper weight; the ratio of the number of times that adjacent preceding and subsequent printing work orders are in the same color group and the registration adjustment process time remains unchanged to the total number of adjacent orders is used to obtain the registration connection probability of the same color group; and the ratio of the number of printing work orders whose plate mounting time exceeds the preset benchmark threshold are placed at the end of the sequence to the total number of printing work orders whose plate mounting time exceeds the preset benchmark threshold is used to obtain the plate mounting timeout avoidance probability. The probability of registration connection within the same color group is calculated as the ratio of the number of adjacent work orders in the preferred scheduling sample that are in the same color group and whose registration adjustment process time remains unchanged to the total number of adjacent work orders in the preferred scheduling sample. The probability of plate-setting timeout avoidance is calculated as the ratio of the number of work orders in the preferred scheduling sample whose plate-setting time exceeds a preset baseline threshold and are at the end of the sequence to the total number of work orders whose plate-setting time exceeds the preset baseline threshold. The preset baseline threshold is determined by taking the 75th percentile value after sorting historical plate-setting time samples from smallest to largest. This position is higher than most conventional plate-setting times and can be used to screen out relatively long plate-setting work orders. If the plate-setting time is less than the preset baseline threshold, it is recorded as not exceeding the threshold; if the plate-setting time is equal to the preset baseline threshold, it is recorded as critical; if the plate-setting time is greater than the preset baseline threshold, it is recorded as exceeding the threshold. Only work orders exceeding the threshold are included in the statistics of plate-setting timeout avoidance probability. The probability of identical grammage transition refers to the degree to which adjacent printing orders in the preferred scheduling sample maintain consistency in paper conditions; the probability of same color group registration connection refers to the degree to which adjacent printing orders in the preferred scheduling sample maintain continuous stability in color group position and registration adjustment time; the probability of plate mounting timeout avoidance refers to the degree to which printing orders with longer plate mounting time in the preferred scheduling sample are arranged at the end of the sequence. If the above probability value is 0, it means that the corresponding feature does not appear in the preferred scheduling sample; if the probability value is closer to 1, it means that the corresponding feature appears more frequently in the preferred scheduling sample.

[0026] The improved cross-entropy method is applied to use the same weight transition probability, same color group quasi-connection probability, and mounting timeout avoidance probability as the probability distribution parameters of the current round. Combined with the preset smooth learning rate, the first product is calculated. At the same time, the learning rate difference is calculated according to the preset second fixed constant and smooth learning rate. The product of the learning rate difference and the first transition probability retained in history is calculated to obtain the second product. The second transition probability is calculated based on the first product and the second product. In MATLAB, the cross-entropy method is applied to the optimized scheduling samples. The probability distribution parameters of the same weight transition, the probability of quasi-connection of same color group sets, and the probability of timeout avoidance in plate mounting are organized into probability distribution parameters of the transfer relationship between adjacent process nodes in the current round. These parameters are then combined with the historically retained first transfer probability to calculate the second transfer probability. The update formula for the basic second transfer probability adopts the discrete probability distribution smooth update form in the cross-entropy method, written as follows: ;in, Indicates the first The basic second transition probability of a round represents the update probability value of the transition relationship of a certain adjacent process flow node in the next round; Indicates the first The first transfer probability retained in the round is derived from the probability value of the transfer relationship of the adjacent process flow node being retained after the previous round of search; Indicates the first The smoothing learning rate for each round ranges from 0 to 1, and is set based on the average difference between the first transition probabilities of two adjacent rounds. A higher average learning rate generally results in a higher learning rate. The closer the value is to 0.2, the smaller the average value. The closer the value is to 0.4; Indicates the first The probability distribution parameters represent the transfer relationships between adjacent process nodes on a single cycle. The probability distribution parameters are written as... ;in, Indicates the first The transition probability of the same paper weight for each batch is obtained by the ratio of the number of times that the paper weight of the adjacent preceding work order to be printed is the same as that of the subsequent work order to be printed in the preferred scheduling sample to the total number of adjacent batches. Indicates the first The probability of registration connection of the same color group of the wheel is obtained by the ratio of the number of times that adjacent preceding and subsequent printing work orders are in the same color group and the registration adjustment process takes no longer in the preferred scheduling sample to the total number of adjacent times. Indicates the first The probability of avoiding a wheel due to plate loading timeout is obtained by the ratio of the number of times a work order to be printed with plate loading time exceeding a preset baseline threshold is placed at the end of the sequence to the total number of work orders to be printed with plate loading time exceeding the preset baseline threshold. This represents the response coefficient for paper weight equality, ranging from 0 to 1. It is set based on the correlation between paper weight equality and the overall reduction in total scheduling time. A higher correlation coefficient results in a higher response coefficient. The larger the value; This represents the stable response coefficient for color group registration, ranging from 0 to 1. It is set based on the correlation between the overall scheduling sequence's total time reduction and the fact that the registration adjustment process takes no longer for the same color group. The higher the correlation, the better. The larger the value; This represents the response coefficient for avoiding plate mounting timeouts, ranging from 0 to 1. It is set based on the correlation between placing printing orders whose plate mounting time exceeds a preset baseline threshold at the end of the sequence and the overall reduction in the total time of the scheduling sequence. A higher correlation coefficient results in a greater reduction in the overall time. The larger the value; Thus The probability value is kept between 0 and 1. To increase the proportion of the optimal scheduling sample with the smaller total time of the overall scheduling sequence in the single probability distribution parameter, the optimal scheduling sample is weighted and rearranged before calculating the probability distribution parameter, and written as follows: ;in, Indicates the first The parameters of a single probability distribution after round weighting; , , Let represent the weighted probability of identical weight transition, the weighted probability of same-color group lamination connection, and the weighted probability of mounting timeout avoidance, respectively. Any weighted probability is written as... ;in, This indicates the probability type index, with values ​​of 1, 2, and 3. Indicates the first The number of sequences in the round-robin scheduling sample; Indicates the first Round The weighted values ​​of the preferred scheduling sample sequences; Indicates the first Round The preferred scheduling sample sequence for the first The observation results of the class features are as follows: during the observation, the adjacent preceding and subsequent work orders to be printed are read one by one according to the order of the work orders in the preferred scheduling sample sequence, and then the first... Feature-by-feature verification: When When comparing the paper weight values ​​of the adjacent preceding and subsequent printing work orders, if they are completely consistent, the corresponding observation result is set to 1; otherwise, the corresponding observation result is set to 0. First, compare whether the target color group number of the adjacent preceding print order is consistent with that of the subsequent print order. Then, compare whether the registration adjustment process time of the two is the same. If both conditions are met, the corresponding observation result is set to 1; if either condition is not met, the corresponding observation result is set to 0. First, determine whether the plate mounting time of the work order to be printed in the sequence exceeds the preset benchmark threshold. Then, check whether the work order to be printed is at the end of the preferred scheduling sample sequence. If both conditions are met, the corresponding observation result is set to 1. If either condition is not met, the corresponding observation result is set to 0. This represents the preferred scheduling sample sequence number. The weighted value is written as... ;in, This represents the time-sensitivity coefficient, ranging from 0.5 to 5. It is set based on the dispersion of the total time of the overall scheduling sequence in the optimal scheduling sample. The greater the difference in the total time of the overall scheduling sequence, the higher the sensitivity coefficient. The larger the value, the smaller the difference in the total time consumption of the overall scheduling sequence. The closer the value is to 0.5; Indicates the first Round The total scheduling time of the entire optimized scheduling sample sequence; Indicates the first The minimum total time taken for the overall scheduling sequence among all the optimal scheduling sample sequences in the round; Indicates the first The maximum total time taken for the entire scheduling sequence in all the preferred scheduling sample sequences. If Then all Take 1. Substituting the weighted individual probability distribution parameters into the update formula for the basic second transition probability, we obtain the improved update formula for the second transition probability, which is written as follows: ;in, Indicates the first The second transition probability of the wheel is used to cover the first transition probability corresponding to the transition relationship of a certain adjacent process flow node; Corresponding to the first group of products; The difference between the second fixed constant and the smoothed learning rate; This corresponds to the second set of products. Compared to the basic second transition probability update formula, the improved second transition probability update formula first organizes the three types of probability features into individual probability distribution parameters corresponding to the transfer relationship of a single adjacent process flow node, and then updates them together with the historically retained first transition probability. This allows the obtained second transition probability to directly cover the first transition probability at the corresponding position and remain consistent with subsequent transition probability distribution updates.

[0027] S5: The first transition probability of the virtual search individual moving between adjacent process flow nodes is updated using the second transition probability. Combined with boundary constraints, the weights are guided to repeatedly execute the node search exploration action until the set maximum number of iterations is met. The selected collaborative scheduling sequence for the multi-color group of the printing press includes: The second transition probability is covered by the first transition probability of the virtual search individual moving between adjacent process flow nodes, and the transition probability distribution is updated. Based on the updated maximum and minimum ant colony algorithm, combined with boundary constraints, the weights are guided to repeatedly execute the node search exploration action until the set maximum number of iterations is reached, and the exploration and optimization operation record is obtained. During overlay, the original first transition probability is replaced with the corresponding second transition probability according to the correspondence between the current process flow node number and the adjacent subsequent process flow node numbers. After replacement, the node search exploration action is repeatedly executed based on the updated max-min ant colony algorithm and the boundary constraint guiding weights, until the set maximum number of iterations is reached. The repeated execution of the node search exploration action here refers to recalculating the path transition tendency, re-determining the color group window, re-traversing the process flow nodes, and reconstructing the overall scheduling sequence according to the updated first transition probability and boundary constraint guiding weights. After each round of exploration, the node order, node transition path, total time of the overall scheduling sequence, and corresponding probability value corresponding to the optimal solution of the current round are written into the exploration record table, thus obtaining the exploration optimization operation record. The exploration optimization operation record here refers to the set of path construction results retained after each round of node search exploration action within the maximum number of iterations. If a valid overall scheduling sequence is not formed in the current round, it is not written into the exploration record table; if a valid overall scheduling sequence is formed, the node order, node transition path, and total time of the overall scheduling sequence must be written.

[0028] Based on the total time consumed by the overall scheduling sequence, the scheduling scheme with the shortest time consumption is selected from the optimal solution of each current round in the exploration and optimization operation record; the scheduling scheme includes each process flow node in the overall scheduling sequence with the shortest time consumption and its corresponding node transfer path; During processing, all current-round optimal solutions retained in the exploration and optimization operation record are reordered in ascending order based on the total time consumed by the overall scheduling sequence. The scheme corresponding to the current-round optimal solution ranked first is the scheduling scheme with the shortest time. The scheduling scheme retains each process flow node and its corresponding node transfer path in the overall scheduling sequence with the shortest time, that is, it retains the order of consecutive process nodes within the work order and the adjacent transfer relationships between work orders. Here, the scheduling scheme refers to the overall scheduling sequence with the shortest total time among all retained current-round optimal solutions and the set of node transfer paths corresponding to that sequence. If multiple current-round optimal solutions have the same total time, the one that first enters the first position in the sorting is retained; if there is only one current-round optimal solution, that one is directly used as the scheduling scheme.

[0029] Obtain the start time of each printing work order in the corresponding color group, the assigned color group number of each printing work order, and the node transfer path between adjacent preceding printing work orders and subsequent printing work orders. Then, splice the corresponding time nodes and positions to output the collaborative scheduling sequence of multiple color groups of the printing press. The start time here refers to the moment when the first plate-filling process node of the print order enters the corresponding color group; the assigned color group number refers to the target color group position of the print order in the final scheduling plan; the node transfer path refers to the connection relationship between adjacent preceding print orders and subsequent print orders, from the last node of the previous order to the first node of the next order. During splicing, according to the overall scheduling sequence determined in the scheduling plan, the start time, assigned color group number, and corresponding node transfer path of each print order are sequentially written into the scheduling result table to form a complete scheduling output unfolded in chronological and color group position order. The collaborative scheduling sequence here refers to the operation sequence formed by the entry order, start time, and adjacent switching relationship of each print order at different color group positions under the current scheduling result for the multi-color group printing press. If a printing work order does not have a corresponding start time in the final scheduling scheme, it means that the work order has not entered the final overall scheduling sequence and will not be written into the scheduling result table. If a printing work order enters the final overall scheduling sequence, its start time, assigned color group number, and corresponding node transfer path must be written simultaneously. The overall process can be understood as follows: first, the printing work orders are broken down into process nodes that can be queued step by step; then, the plate change, ink change, paper change, and waiting status between work orders are converted into path weights; then, a virtual search individual repeatedly tries and compares the arrangements, retaining the arrangement with the shorter total time; and then, the transfer probability of the next round is further adjusted based on the optimal scheduling sample until the final order is obtained. After being output to the printing press, the final output can make the entry order of each printing work order among multiple color groups smoother, reduce the downtime caused by plate change and cleaning, shorten the total downtime preparation time, and make the color group switching more continuous.

[0030] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A multi-color group collaborative scheduling method for printing presses based on ant colony algorithm, characterized in that, Includes the following steps: Obtain the printing work order data of the printing press, divide the process flow nodes and node transfer paths of the printing work order data, and combine them into an ordered scheduling diagram; Based on the changes in different print order data over time, determine the initial weights for truncating the transfer paths of nodes in the ordered scheduling graph. Based on the initial weights of the truncation limit, the virtual search individuals are set through the maximum and minimum ant colony algorithm, and the first transition probability of the virtual search individuals moving between adjacent process flow nodes is set. The process flow nodes are iteratively optimized to determine the boundary constraint guiding weights. The number of iterations of the current maximum and minimum ant colony algorithm is counted. When the number of iterations reaches the preset upper limit of the number of iterations, the preferred scheduling sample is selected, and the second transition probability derived from the specified work order data to be printed in the preferred scheduling sample is calculated. The first transition probability of the virtual search individual moving between adjacent process flow nodes is updated using the second transition probability. Combined with boundary constraints, the weights are guided to repeatedly execute node search exploration actions until the set maximum number of iterations is met, and the collaborative scheduling sequence of the printing press multi-color group is selected.

2. The multi-color group collaborative scheduling method for printing presses based on ant colony algorithm according to claim 1, characterized in that, The process of acquiring the printing work order data of the printing press, dividing the process flow nodes and node transfer paths of the printing work order data, and combining them into an ordered scheduling diagram includes: Obtain the print order data for the multi-color printing press. The print order data includes the page size parameters, target color group number, target ink type, paper weight, target print quantity, and rated print volume per hour. Each print order in the print order data is decomposed into sequentially connected plate loading process nodes, printing operation process nodes, ink path circulation process nodes, registration adjustment process nodes, and roll removal and plate withdrawal process nodes according to the printing process of the multi-color printing press. Each independent process node is arranged and combined according to the target color group number position to generate a scheduling node set. Each independent process node in the scheduling node set is mapped to a process flow node in the max-min ant colony algorithm for virtual optimization individuals to traverse and search. For combinations of different printing work orders, the roll removal and plate removal process node at the end of the preceding printing work order is connected to the plate loading process node at the beginning of the subsequent printing work order. Based on the sequential transfer order between adjacent process flow nodes, the node transfer path in the maximum and minimum ant colony algorithm is set, and the process flow nodes and node transfer paths are spliced ​​together to generate an ordered scheduling diagram.

3. The multi-color group collaborative scheduling method for printing presses based on ant colony algorithm according to claim 1, characterized in that, The determination of the initial weights for truncating the transfer paths of nodes in the ordered scheduling graph based on the changes in different print order data over time includes: The preceding and subsequent printing work orders connected by the transfer path of each node in the ordered scheduling diagram are determined. Historical statistical results matching the page size parameters of the preceding printing work order are obtained, and the empirical time consumption of the roll removal and plate withdrawal process corresponding to the preceding printing work order is determined. At the same time, historical statistical results matching the page size parameters of the subsequent printing work order are obtained, and the empirical time consumption of the printing plate loading process corresponding to the subsequent printing work order is determined. The empirical times of the two are added together to obtain the corresponding sequential plate change time. The sequential plate change time is normalized based on the preset maximum plate change time to obtain the plate change time ratio. Based on the target ink type of adjacent preceding and subsequent printing work orders, determine the ink consistency status between them, determine the ink switching time based on the ink consistency status, and perform normalization calculation on the ink switching time with the preset maximum cleaning time as the benchmark to obtain the ink contamination risk ratio. Calculate the absolute difference in paper weight between adjacent preceding and subsequent printing work orders to obtain the paper weight difference. Normalize the paper weight difference based on the preset maximum weight fluctuation to obtain the paper feed continuity ratio. Calculate the ratio of the target printing quantity to the rated printing quantity per hour to obtain the estimated printing time. Based on the target color group number and the estimated printing time, determine the departure time of the adjacent preceding printing work order from the current color group and the preheating completion time of the corresponding color group of the subsequent printing work order. Calculate the absolute difference between the two times to obtain the drying waiting time. Perform normalization calculation on the drying waiting time based on the preset maximum drying waiting time to obtain the drying waiting ratio. The corresponding weighting coefficients are configured for the plate change time ratio, the ink contamination risk ratio, the paper feed continuity ratio, and the drying waiting ratio, and a weighted summation calculation is performed to obtain the comprehensive heuristic coefficient. The ratio of the preset heuristic benchmark value to the comprehensive heuristic coefficient is calculated to obtain the original guiding value. The maximum weight boundary and the minimum weight boundary are set for the original guiding value to establish the truncation limit initial weight.

4. The multi-color group collaborative scheduling method for printing presses based on ant colony algorithm according to claim 1, characterized in that, The process involves determining the boundary constraint guiding weights by setting virtual search individuals and their first transition probability when moving between adjacent process flow nodes using the minimax ant colony algorithm, based on truncation-limited initial weights. This is achieved through iterative optimization of the process flow nodes. Define a set of virtual search individuals in the maximum and minimum ant colony algorithm, and the first transition probability of the virtual search individuals moving between adjacent process flow nodes. Combine the truncation limit initial weight and the first transition probability to calculate the path transfer tendency of the virtual search individuals between each process flow node. Based on the path transfer tendency, determine the color group window to which the printing work order will enter, and control the virtual search individual to traverse the process flow nodes in sequence according to the color group window to complete the path construction. After completing one path construction, generate the overall scheduling sequence, and summarize the total duration of each process flow node passed in the overall scheduling sequence to obtain the total time consumption of the overall scheduling sequence. All overall scheduling sequences are sorted in ascending order of total time consumption. The overall scheduling sequence ranked first is extracted as the optimal solution for the current round. The original guiding values ​​on the node transfer paths traversed by the optimal solution for the current round are summarized. Combined with a preset first fixed constant, the updated guiding weight is calculated. The updated guiding weight is truncated and restricted by the maximum weight boundary and the minimum weight boundary to generate boundary constraint guiding weight.

5. The multi-color group collaborative scheduling method for printing presses based on ant colony algorithm according to claim 1, characterized in that, The algorithm counts the number of iterations of the current maximum and minimum ant colony algorithm. When the number of iterations reaches the preset upper limit of iteration rounds, a preferred scheduling sample is selected. The second transition probability derived from the specified work order data in the preferred scheduling sample includes: Monitor the number of iterations of the current maximum and minimum ant colony algorithm. When the number of iterations reaches the preset upper limit of the number of iteration rounds, select the sequence that ranks in the top 10% from the historically retained optimal solutions of the current round based on the total time consumption of the overall scheduling sequence, and use it as the preferred scheduling sample. The probability of identical paper weight transition is obtained by statistically analyzing the ratio of the number of times that adjacent preceding and subsequent printing work orders have the same paper weight to the total number of adjacent orders in the preferred scheduling sample; the probability of same paper weight transition is obtained by the ratio of the number of times that adjacent preceding and subsequent printing work orders are in the same color group and the registration adjustment process time remains unchanged to the total number of adjacent orders; and the probability of same color group registration connection is obtained by the ratio of the number of printing work orders whose plate mounting time exceeds the preset benchmark threshold to the number of printing work orders whose plate mounting time exceeds the preset benchmark threshold to the number of printing work orders whose plate mounting time exceeds the preset benchmark threshold. The improved cross-entropy method is applied to use the same weight transition probability, the same color group quasi-connection probability, and the mounting timeout avoidance probability as the probability distribution parameters of the current round. Combined with the preset smooth learning rate, the first product is calculated. At the same time, the learning rate difference is calculated according to the preset second fixed constant and the smooth learning rate. The product of the learning rate difference and the first transition probability retained in history is calculated to obtain the second product. The second transition probability is calculated based on the first product and the second product.

6. The multi-color group collaborative scheduling method for printing presses based on ant colony algorithm according to claim 1, characterized in that, The process involves updating the first transition probability of the virtual search individual's movement between adjacent process flow nodes using the second transition probability, and combining this with boundary constraints to guide the repeated execution of node search exploration actions until the set maximum number of iterations is met. The selected collaborative scheduling sequence for the multi-color group of the printing press includes: The second transition probability is covered by the first transition probability of the virtual search individual moving between adjacent process flow nodes, and the transition probability distribution is updated. Based on the updated maximum and minimum ant colony algorithm, combined with the boundary constraints, the node search exploration action is repeatedly executed until the set maximum number of iterations is reached, and the exploration and optimization operation record is obtained. Based on the total time consumed by the overall scheduling sequence, the scheduling scheme with the shortest time consumption is selected from the optimal solution of each current round in the exploration and optimization operation record; Obtain the start time of each printing work order in the corresponding color group, the assigned color group number of each printing work order, and the node transfer path between adjacent preceding printing work orders and subsequent printing work orders. Then, splice the corresponding time nodes and positions to output the collaborative scheduling sequence of multiple color groups of the printing press.

7. The multi-color group collaborative scheduling method for printing presses based on ant colony algorithm according to claim 3, characterized in that, If the target ink types are the same, the ink consistency status is determined to be consistent; if the target ink types are different, the ink consistency status is determined to be inconsistent. The target ink types include ordinary ink, UV ink, and spot color ink. When the ink consistency status is inconsistent, the preset cleaning color group time is the ink switching time. When the ink consistency status is consistent, the ink switching time is zero.

8. The multi-color group collaborative scheduling method for printing presses based on ant colony algorithm according to claim 3, characterized in that, The weighting coefficients are allocated based on the degree of influence of each process switching operation on the total downtime preparation time of the printing press. The longer the adjustment time required for the target process switching operation, the larger the corresponding weighting coefficient.

9. The multi-color group collaborative scheduling method for printing presses based on ant colony algorithm according to claim 5, characterized in that, The upper limit of the iteration rounds is set based on the total number of work orders to be printed and the total scale of process flow nodes.

10. The multi-color group collaborative scheduling method for printing presses based on ant colony algorithm according to claim 6, characterized in that, The scheduling scheme includes each process flow node in the overall scheduling sequence with the shortest time consumption and its corresponding node transfer path.

Images