Control method and system for carton production

Abstract

The invention relates to the technical field of production line control, and relates to a control method and system for carton production, and the method comprises the steps: obtaining a carton production task comprising N processes, obtaining an ant colony, and initializing the ant colony into an exploration sub-population and a mining sub-population; according to the pheromones and the composite heuristic information, determining the probability that the ants select to-be-processed procedures; after all ants of the sub-population are explored and all ants of the sub-population are explored to complete path construction, updating a global optimal path and updating pheromones; and performing iteration until a preset termination condition is met, and determining a global optimal path obtained by the last iteration as a production control scheme. According to the method, the exploration sub-population and the exploitation sub-population which work in parallel are set, and different search strategies are configured, so that the global and local capabilities of ants are balanced, the search process is prevented from falling into local optimum too early, and the overall efficiency of carton production is improved.

Description

Technical Field

[0001] This invention relates to the field of production line control technology. More specifically, this invention relates to a control method and system for paper box production. Background Technology

[0002] The paper box production process typically involves multiple steps such as printing, lamination, die-cutting, and gluing, requiring processing on multiple machines according to specific technological paths. The core objective of production scheduling is to rationally arrange the processing sequence and equipment allocation of each step while meeting various technological constraints and resource limitations, in order to achieve optimization goals such as the shortest production cycle, the highest equipment utilization rate, and the lowest production cost.

[0003] As the number of workpieces and the scale of processes increase, the solution space grows exponentially, making it difficult to find an exact optimal solution in polynomial time. Production scheduling methods often rely on experience or simple priority rules, such as first-come, first-served or shortest processing time priority. While these methods are simple and easy to implement, they often fail to achieve satisfactory scheduling results in complex production environments with multiple objectives and constraints, easily leading to problems such as idle equipment, bottleneck processes, and delivery delays.

[0004] Ant colony optimization (ACO) transforms the scheduling problem into a combinatorial optimization problem of finding the optimal path by simulating the foraging behavior of real ants using pheromones for communication. However, in the early stages of the search, the convergence speed is slow due to the relatively uniform distribution of pheromones; in the later stages, it is prone to getting stuck in local optima due to excessive concentration of pheromones on a few better paths, leading to premature convergence and search stagnation, making it difficult to balance global exploration and local exploitation capabilities. Especially for the max-min ant colony system, although setting upper and lower limits for pheromones can prevent algorithm stagnation, the fixed limit setting weakens the algorithm's adaptability. In addition, the heuristic information relied upon by ACO in path selection is often relatively simple and fails to fully consider the dynamic characteristics of the production site, such as equipment load and the time and cost consumption caused by process switching, resulting in a deviation between the generated scheduling scheme and actual production needs. Summary of the Invention

[0005] To address the technical problem that the scheduling scheme determined by the ant colony algorithm may deviate from the actual production needs, the present invention provides solutions in the following aspects.

[0006] In a first aspect, a control method for paper box production includes: obtaining a package containing... N The paper box production task for each process involves obtaining an ant colony and initializing it into an exploration sub-species and a mining sub-species, then determining the operating parameters for these sub-species. NThe integer is positive; it enables ants exploring the subpopulation and ants mining the subpopulation to construct scheduling paths in parallel, and determines the processing steps to be performed by the ants based on pheromones and composite heuristic information. j The probability of composite heuristic information and candidate processes; where composite heuristic information and candidate processes are related. j The bottleneck equipment's expected waiting time, remaining total processing time, and process changeover costs are related; the selection of processing processes by ant subpopulations is explored. j The probability is also related to the perturbation factor, which is inversely proportional to the structural overlap of the current global optimal path. Ants in the mining sub-species make decisions based on probability. After all ants in the exploring sub-species and all ants in the mining sub-species have completed path construction, the global optimal path and pheromone are updated. In this process, the corresponding process segment in the optimal path of the exploring sub-species is used to replace the bottleneck segment in the current global optimal path to update the global optimal path. The path segments on the updated global optimal path are enhanced with pheromone, the exploring path with the greatest difference from the global optimal path is rewarded, and the worst path in the current iteration is penalized with evaporation. The pheromone lower limit for the next iteration is updated based on the coefficient of variation and kurtosis of the pheromone values ​​on each path segment in the global optimal path. The iteration continues until the preset termination condition is met, and then the global optimal path obtained in the last iteration is output as the production control scheme.

[0007] Preferably, calculating composite heuristic information includes: targeting the current process i Transfer to any candidate process j Obtain the execution procedure j Required equipment M j ; Calculation process j Expected device wait time EWT ij The waiting time value is the device M j The time to complete all assigned tasks and the current process i The difference in completion times, where if the waiting time value is negative, the waiting time is determined to be 0; calculate the difference including candidate processes. j Total remaining processing time for all unscheduled processes, including RPT j ; Determine from the current process i Switch to candidate process j Process switching costs ST j ; Calculate the waiting time for the device EWT ij The remaining total processing time RPT j and the process switching cost ST jThe reciprocal of the weighted sum is used to obtain the composite heuristic information, wherein the composite heuristic information is calculated. or ij The formula is: ,in, β 1. β 2 and β 3 is a preset weighting coefficient, and β 1+ β 2+ β 3 = 1.

[0008] Preferably, the method by which the exploratory subpopulation introduces a perturbation factor inversely proportional to the structural overlap of the current globally optimal path during selection is as follows: for ants in the exploratory subpopulation k In selecting candidate processes j Previously, the calculation of the constructed partial path was performed. P k Compared with the current globally optimal path L gb The number of identical process pairs is denoted as the structural overlap. SO ( k ); Calculate the ant's perturbation factor based on the structural overlap. D ( k The calculation formula is as follows: ; Calculate ants k by process i Select process j probability The calculation formula is as follows: .

[0009] l For all optional next steps, α , β and c To preset weights, or ij From the process i Transfer to process j Heuristic information values, t ij ( t ) is in the t In the next iteration, the path segment ( i , j The value of pheromones accumulated on the surface, or il From the process i Transfer to process l Heuristic information values, t il ( t ) is in the t In the next iteration, the path segment (i , l The value of pheromones accumulated on the surface.

[0010] Preferably, the attempt to replace the bottleneck segment in the current globally optimal path with the corresponding process segment in the optimal path of the exploratory subpopulation includes: obtaining the bottleneck device with the longest total load time in the current globally optimal path. M b Extract the bottleneck device from the current globally optimal path. M b The sequence of processes is denoted as S gb Extract the bottleneck device allocated to the optimal path in the current exploration subpopulation. M b The sequence of processes is denoted as S es Construct a candidate scheduling path, and identify the bottleneck devices within the candidate scheduling path. M b The process sequence above adopts S es The process sequence on other equipment remains consistent with the current global optimal path. Based on process dependencies and equipment occupancy constraints, the start and finish times of all affected processes in the candidate scheduling path are recalculated to obtain a complete and feasible scheduling scheme. If the total completion time of the candidate scheduling path is less than the total completion time of the current global optimal path, the global optimal path is updated to the candidate scheduling path.

[0011] Preferably, pheromone updates include: updating all path segments ( i , j Perform global pheromone evaporation, where in the (t+1)th iteration, the updated pheromone is... The formula for calculating the updated pheromone is: , r The global volatility coefficient is set to a preset value. The pheromone level is as follows: The globally optimal path is enhanced, and exploration paths are rewarded. The update formula is: Among them, the increment In the path segment ( i , j Located on the globally optimal path L gb When the value is positive and it is close to the globally optimal path L gb The total completion time is inversely proportional to the increment. In the path segment ( i , j Located on the globally optimal path L gb The exploration path with the greatest differenceL ed When the value is positive, the increment is... Exploration Path L ed The total completion time is inversely proportional to the total completion time; for the worst path in the current iteration L w Path segment on ( i , j A penalty evaporation is applied to obtain the pheromone in the current iteration; where the pheromone obtained in the (t+1)th iteration is obtained. The formula is: ,in r p This is the penalty evaporation coefficient; for the (t+1)th iteration, for paths not belonging to the worst path in the current iteration... L w The path segment in which the pheromone value remains at 0. .

[0012] Preferably, updating the pheromone lower bound for the next iteration includes: extracting the pheromone values ​​of all path segments on the current globally optimal path to form a set; and calculating the coefficient of variation of the set. CV with kurtosis Kurt According to the preset target coefficient of variation CV 0 and target kurtosis Kurt 0 Update pheromone lower limit, including the update of the first t The lower bound of pheromones in the next iteration is obtained to obtain the first... t The formula for the lower limit of pheromones after +1 iterations is: ,in l 1 and l 2 is for adjusting the step size coefficient. t min ( t ) is the first t The lower bound of pheromones in the next iteration. t min ( t +1) is the first t +1 iterations of pheromone lower bound t It is a positive integer.

[0013] Preferably, determining the operating parameters of the exploratory subpopulation and the mining subpopulation includes: determining the pheromone importance factor and the composite heuristic information importance factor of the exploratory subpopulation, denoted as the operating parameters of the exploratory subpopulation; determining the pheromone importance factor and the composite heuristic information importance factor of the mining subpopulation, denoted as the operating parameters of the mining subpopulation; wherein, the pheromone importance factor of the mining subpopulation is greater than the pheromone importance factor of the exploratory subpopulation, and the composite heuristic information importance factor of the exploratory subpopulation is greater than the composite heuristic information importance factor of the mining subpopulation.

[0014] In a second aspect, a control system for paper box production includes: a parameter setting module for acquiring parameters including... N For each process of cardboard box production, the ant colony is initialized into an exploration sub-swarm and a mining sub-swarm, with operating parameters set for each. A path construction module is used to construct scheduling paths in parallel for the two sub-swarms, determining the probability of an ant selecting the next process j based on pheromones and composite heuristic information. The composite heuristic information is determined based on the expected waiting time of the bottleneck equipment in the candidate process j, the remaining total processing time, and the process switching cost. The exploration sub-swarm introduces a perturbation factor inversely proportional to the structural overlap of the current globally optimal path during selection, while the mining sub-swarm makes decisions based on probability. A path and pheromone update module is used to update the globally optimal path after all ants have completed path construction. New pheromone: Attempt to replace the bottleneck segment in the current global optimal path with the corresponding process segment in the optimal path of the exploration subpopulation to generate a new global optimal path with better evaluation indicators; perform pheromone update, including pheromone enhancement only on the path segments of the updated global optimal path, rewarding the exploration path with the greatest difference from the global optimal path, and imposing a penalty on the worst path in the current iteration; update the pheromone lower limit for the next iteration based on the coefficient of variation and kurtosis of the pheromone values ​​of each path segment of the global optimal path; scheme output module, used to repeatedly execute the path construction and update steps until the preset termination condition is met, and output the final global optimal path as the production control scheme.

[0015] Preferably, the composite heuristic information is based on the selected process. j The expected waiting time for bottleneck equipment, the remaining total processing time, and the process changeover cost are determined.

[0016] Preferably, the exploratory subpopulation is selected by introducing a perturbation factor that is inversely proportional to the structural overlap of the current globally optimal path.

[0017] The beneficial effects of this invention are as follows: This invention balances the global and local capabilities of ants by establishing parallel exploratory and mining subpopulations and configuring different search strategies, thus preventing the search process from prematurely falling into local optima. During the pheromone update phase, the quality of understanding and population diversity are improved by replacing bottleneck segments of the globally optimal path, rewarding differentiated paths, and penalizing the worst path. Simultaneously, adjusting the search lower bound based on the pheromone distribution further improves the algorithm's convergence performance, resulting in a production control scheme with shorter production cycles and more efficient resource utilization, thereby improving the overall efficiency of cardboard box production. Attached Figure Description

[0018] Figure 1This is a schematic flowchart illustrating the steps of a control method for paper box production according to an embodiment of the present invention. Detailed Implementation

[0019] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are some embodiments of the present invention, but not all embodiments.

[0020] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0021] Figure 1 This is a schematic flowchart illustrating the steps of a control method for paper box production according to an embodiment of the present invention.

[0022] like Figure 1 As shown, a control method for paper box production includes steps S1 to S4.

[0023] Step S1: Obtain a cardboard box production task containing N processes, obtain an ant colony and initialize the ant colony into an exploration sub-population and a mining sub-population, and determine the operating parameters of the exploration sub-population and the mining sub-population.

[0024] It should be noted that cardboard box production involves processing flat paper into packaging containers with specific shapes, structures, and functions through a series of processes such as printing, die-cutting, and gluing. Task data is retrieved from the production management system, including the number of workpieces, the number of processes within each workpiece, the processing time for each process, and available equipment information. The total number of ants in the ant colony is set, for example, 100 ants. The ants are then divided into two sub-species, each with 50 ants: an exploration sub-species and a mining sub-species. A lower pheromone importance factor and a higher heuristic importance factor are assigned to the exploration sub-species to encourage them to explore unknown paths. A higher pheromone importance factor and a lower heuristic importance factor are assigned to the mining sub-species to encourage them to utilize existing experience.

[0025] In one embodiment, determining the operating parameters of the exploratory subpopulation and the mining subpopulation includes: determining the pheromone importance factor and the composite heuristic information importance factor of the exploratory subpopulation, denoted as the operating parameters of the exploratory subpopulation; determining the pheromone importance factor and the composite heuristic information importance factor of the mining subpopulation, denoted as the operating parameters of the mining subpopulation; wherein the pheromone importance factor of the mining subpopulation is greater than the pheromone importance factor of the exploratory subpopulation, and the composite heuristic information importance factor of the exploratory subpopulation is greater than the composite heuristic information importance factor of the mining subpopulation.

[0026] It should be noted that the core of mining subpopulations is to deeply explore the currently discovered high-quality solution regions. Therefore, the importance factor of pheromone mining subpopulations is crucial.α m Higher, for example α m =2.0; Importance factor of composite heuristic information from mining subpopulations β m Lower, for example β m =1.0. Ants in the mining subpopulation, when choosing paths, place greater emphasis on the historically accumulated pheromone intensity. They tend to repeatedly follow paths proven to lead to optimal solutions, optimizing these solutions on a small scale. In contrast, the exploring subpopulation is more inclined to seek entirely new solutions. Therefore, the pheromone importance factor of the exploring subpopulation... α e Lower, for example α e =1.0; Importance factor of composite heuristic information in exploring subpopulations β m Higher, for example β e =2.0. Based on this, ants exploring subpopulations rely less on historical pheromones when making decisions, and instead focus more on the immediate benefits provided by heuristic information, such as prioritizing processes that can be started immediately. This increases the chances of the ant colony algorithm discovering new and better scheduling schemes, ensuring the globality of the search.

[0027] It should be noted that in ant colony optimization, pheromones represent the historical merits of a path, while heuristics represent the current attractiveness of a path. Pheromone importance factors and composite heuristic importance factors are used to adjust the weights of pheromones and heuristics in the decision-making process, thereby balancing the exploratory (finding new paths) and exploitative (optimizing known paths) aspects of the ant colony optimization algorithm.

[0028] Step S2: Construct scheduling paths in parallel for ants exploring subpopulations and ants mining subpopulations, and determine the probability of an ant choosing a processing step based on pheromones and compound heuristic information.

[0029] Among them, the composite heuristic information and the candidate process j The bottleneck equipment's expected waiting time, remaining total processing time, and process changeover cost are all related; the ant population in the exploration sub-species selects the processing process to be completed. j The probability is also related to a perturbation factor, which is inversely proportional to the structural overlap of the current global optimal path. The ants in the mining subpopulation make decisions based on probability.

[0030] Each ant in one of the two subpopulations, the exploratory subpopulation and the mining subpopulation, starts from the initial process and chooses the next processing step according to the roulette wheel method; this step is denoted as the process to be processed. jThe product of pheromone values ​​and composite heuristic information is used to select the processing step. j The probability is directly proportional to the probability.

[0031] It should be noted that the value of pheromones is recorded as the concentration of pheromones.

[0032] In one embodiment, calculating composite heuristic information includes: selecting candidate processes. j The expected available time of the required equipment, the total time of all subsequent processes of the workpiece to which the process belongs, and the equipment adjustment time caused by changing the workpiece or process are weighted and summed, and the reciprocal is taken. For ants exploring the subpopulation, when calculating the selection probability, an additional perturbation factor is multiplied. The perturbation factor is determined by calculating the overlap between the candidate path segment and the current optimal solution path. The higher the overlap, the smaller the factor value, which reduces the probability of selecting a known good path segment and prompts the ants to try new paths.

[0033] Specifically, the composite heuristic information and the candidate process j The expected waiting time for bottleneck equipment, the remaining total processing time, and the process changeover cost are all related. In one embodiment, calculating composite heuristic information includes: targeting the current process... i Transfer to any candidate process j Obtain the execution procedure j Required equipment M j ; Calculation process j Expected device wait time EWT ij The waiting time value is the device M j The time to complete all assigned tasks and the current process i The difference in completion times, where if the waiting time value is negative, the waiting time is determined to be 0; calculate the difference including candidate processes. j Total remaining processing time for all unscheduled processes, including RPT j ; Determine from the current process i Switch to candidate process j Process switching costs ST j ; Calculate the waiting time for the device EWT ij The remaining total processing time RPT j and the process switching cost ST j The reciprocal of the weighted sum yields the composite heuristic information: ,in, or ij The composite heuristic information, β 1.β 2 and β 3 is a preset weighting coefficient, and β 1+ β 2+ β 3 = 1.

[0034] Assuming the current process i If the process is completed at time 50, the ant needs to move on to the next process. j and process k Select Next. Pending Processes j Equipment required M A ,equipment M A If the equipment is not idle until the 60th time, then the expected waiting time is 10 time intervals. And the process... k Required equipment M B If the device is idle by the 45th time, then... M B The waiting time is 0.

[0035] Furthermore, assuming the inclusion of candidate processes... j The remaining total processing time for all unscheduled processes, including those involving processes, is 200. k The remaining total processing time is 220. Furthermore, from the current process... i Switch to process j The switching cost ST is 2, and the cost of switching to process k is 8. If the weighting coefficients... β 1, β 2, β 3. Satisfy: β 1 = 0.5 β 2 = 0.3, β 3 = 0.2, select the candidate process. j The evaluation value was 65.4; Selected process k The evaluation score is 67.6. Pending process. j The heuristic information value is the candidate process. j The reciprocal of the evaluation value, therefore the candidate process j Ants are more likely to choose alternative processes because they receive greater heuristic information. j .

[0036] In one embodiment, the switching cost is time, which is the time required to switch from one process to another.

[0037] In an optional embodiment, the ants in the exploratory subpopulation select the processing steps. j The probabilities include: for ants in the exploration subpopulation k In selecting candidate processesj Previously, the calculation of the constructed partial path was performed. P k Compared with the current globally optimal path L gb The number of identical process pairs is denoted as the structural overlap. SO ( k ); Calculate the ant's perturbation factor based on the structural overlap. D ( k The calculation formula is as follows: ; Calculate ants k by process i Select process j probability The calculation formula is as follows: .

[0038] in, l For all optional next steps, α , β and c To preset weights, or ij From the process i Transfer to process j Heuristic information values, t ij ( t ) is in the t In the next iteration, the path segment ( i , j The value of pheromones accumulated on the surface, or il From the process i Transfer to process l Heuristic information values, t il ( t ) is in the t In the next iteration, the path segment ( i , l The value of pheromones accumulated on the surface.

[0039] Assuming the currently known globally optimal path L gb The sequence of steps is A to B and then to C. Ant 1 in the exploratory subpopulation is constructing a path, and the path ant 1 has already traversed is also from A to B, which is similar to the globally optimal path. L gb There is a common process pair between them, therefore the structural overlap corresponding to ant 1 is... SO The value is 1, which is the perturbation factor corresponding to ant 1. D The value is 0.5.

[0040] When this ant 1 is from the process B When selecting the next process, that is, when selecting a candidate process, regardless of whether the process recommended by the optimal path is chosen... C Or should we choose other processes? D The calculated transition probability will be multiplied by a perturbation factor less than 1, such as 0.5. c The power of this. In contrast, if the constructed path of another ant 2 exploring the subgroup is similar to the globally optimal path... L gb Completely different, the path structure overlap is 0, while the perturbation factor is 1. In this case, ant 1's decision-making is inhibited, its choice behavior becomes more random, and it has a greater chance of deviating from the known optimal path; while ant 2 is unaffected. Therefore, the more similar the exploring ants are to the optimal solution, the more likely they are to exhibit mutation behavior, preventing the exploring subpopulation from prematurely imitating the optimal solution and ensuring the breadth of the search.

[0041] Step S3: After all ants in the exploring subpopulation and all ants in the mining subpopulation have completed path construction, update the globally optimal path and update the pheromone.

[0042] The process of updating the global optimal path includes: attempting to replace the bottleneck segment in the current global optimal path with the corresponding process segment in the optimal path of the exploration subpopulation, so as to update the global optimal path; updating pheromones includes enhancing the pheromones of the path segments on the updated global optimal path, rewarding the exploration path with the greatest difference from the global optimal path, and imposing a penalty on the worst path in the current iteration; and updating the pheromone lower limit for the next iteration based on the coefficient of variation and kurtosis of the pheromone values ​​of each path segment in the global optimal path.

[0043] It should be noted that after all ants have generated complete scheduling paths, the path leading to the current global optimum is found. L gb The longest critical process segment in total time is identified as the bottleneck segment; simultaneously, the optimal path discovered by the exploratory subpopulation in this iteration is found; the process segment processing the same workpiece in the optimal path of the exploratory subpopulation is used to replace the globally optimal path. L gb The bottleneck segment is identified, forming a new candidate path. If the total time of the new path is shorter, the path is updated to the new globally optimal path. L gb To update pheromones, first, volatilize pheromones at a uniform rate across all paths, then only on the new globally optimal path. L gb Add pheromones to the path segments traversed; find the globally optimal path. L gbThe exploration path with the greatest difference in path structure is given a small pheromone reward; the worst path generated by all ants in this round is identified, and an additional, larger proportion of pheromone is released onto the path segment traversed by the ant; in addition, the globally optimal path is calculated. L gb The dispersion and peak sharpness of pheromone values ​​in each path segment are considered. If the values ​​indicate that pheromone is overly concentrated, the lower limit of pheromone for the next iteration is increased; otherwise, the lower limit is decreased.

[0044] In an optional embodiment, the attempt to replace the bottleneck segment in the current globally optimal path with the corresponding process segment in the optimal path of the exploratory subpopulation includes: obtaining the bottleneck device with the longest total load time in the current globally optimal path. M b Extract the bottleneck device from the current globally optimal path. M b The sequence of processes is denoted as S gb Extract the bottleneck device allocated to the optimal path in the current exploration subpopulation. M b The sequence of processes is denoted as S es Construct a candidate scheduling path, and identify the bottleneck devices within the candidate scheduling path. M b The process sequence above adopts S es The process sequence on other equipment remains consistent with the current global optimal path. Based on process dependencies and equipment occupancy constraints, the start and finish times of all affected processes in the candidate scheduling path are recalculated to obtain a complete and feasible scheduling scheme. If the total completion time of the candidate scheduling path is less than the total completion time of the current global optimal path, the global optimal path is updated to the candidate scheduling path.

[0045] For example, after a certain iteration, the globally optimal path is obtained. L gb The total completion time is 300 minutes. The bottleneck equipment is... M 3. The path with the highest total workload is the globally optimal path. L gb Bottleneck equipment allocated in the middle M 3. Process sequence S gb These are process 1, process 5, and process 8.

[0046] At the same time, the optimal path was found by exploring subpopulations. L es Although the total completion time may be 310 minutes, which is inferior to the globally optimal path. Lgb However, in the best path L es In the scheduling scheme, the equipment M 3. Process sequence S es It refers to process 5, process 1, and process 8, that is... S es ={Process 5, Process 1, Process 8}. A replacement operation will be performed, generating a candidate path. L cand In the candidate paths, the device M 1 and equipment M The process arrangement on non-bottleneck equipment maintains the globally optimal path. L gb The original state, only the bottleneck equipment M The execution sequence of the processes on step 3 has been changed from process 1, process 5, process 8 to process 5, process 1, process 8. This is because the equipment... M 3. Re-verify and calculate the start and end times of all affected processes due to changes in the process sequence to ensure the feasibility of the new solution. If, after calculation, the candidate path... L cand The total completion time has changed to 295 minutes, which is less than the previous 300 minutes, so the candidate path will be used. L cand Replace the current globally optimal path L gb Become the new globally optimal path L gb .

[0047] In an optional embodiment, the pheromone update includes pheromone enhancement only on path segments of the updated globally optimal path, rewarding the exploration path with the greatest difference from the globally optimal path, and imposing a penalty evaporation on the worst path in the current iteration, including steps S301 to S303: Step S301: For all path segments ( i , j Perform global pheromone evaporation, where in the (t+1)th iteration, the updated pheromone is... The formula for calculating the updated pheromone is: , r The global volatility coefficient is set to a preset value. The pheromones before the update.

[0048] Here, it is assumed that the initial pheromone value of path segment A to B is 10, and the global volatility coefficient is... r The value is 0.1. After step S301 is executed, the pheromones first uniformly decay to 9, which is the natural forgetting of pheromones.

[0049] Step S302: Enhance the globally optimal path and reward the explored path. The update formula is as follows: Among them, the increment In the path segment ( i , j Located on the globally optimal path L gb When the value is positive and it is close to the globally optimal path L gb The total completion time is inversely proportional to the increment. In the path segment ( i , j Located on the globally optimal path L gb The exploration path with the greatest difference L ed The value is positive and the increment is... Exploration Path L ed The total completion time is inversely proportional to the total completion time.

[0050] Among them, if the worst path in this iteration L w It also includes path segment A to B, and the penalty volatility coefficient. r p If the pheromone level is 0.2, then the pheromone level of the path segment will be penalized to 7.84. Conversely, if the worst path does not pass through this path segment, the pheromone level will remain at 9.8. This sequential update method ensures that the attractiveness of high-quality paths is enhanced, low-quality paths are suppressed, and novel potential paths are also given attention.

[0051] Step S303: For the first t Worst path in the next iteration L w Path segment on ( i , j A penalty evaporation is applied to obtain the pheromone in the current iteration; where the pheromone obtained in the (t+1)th iteration is obtained. The formula is: ,in r p This is the penalty evaporation coefficient; for the (t+1)th iteration, for paths not belonging to the worst path in the current iteration... L w The path segment in which the pheromone value remains at 0. .

[0052] Among them, if the worst path in this iteration L w It also includes path segment A to B, and the penalty volatility coefficient. rp If the pheromone level is 0.2, then the pheromone level of the path segment will be penalized to 7.84. Conversely, if the worst path does not pass through this path segment, the pheromone level will remain at 9.8. This sequential update method ensures that the attractiveness of high-quality paths is enhanced, low-quality paths are suppressed, and novel potential paths are also given attention.

[0053] To avoid premature convergence, in an optional embodiment, updating the pheromone lower bound for the next iteration based on the coefficient of variation and kurtosis of the pheromone values ​​distributed across each path segment of the global optimal path includes: extracting the pheromone values ​​of all path segments on the current global optimal path to form a set; and calculating the coefficient of variation of the set. CV with kurtosis Kurt According to the preset target coefficient of variation CV 0 and target kurtosis Kurt 0 Update pheromone lower limit, including the update of the first t The lower bound of pheromones in the next iteration is obtained to obtain the first... t The formula for the lower limit of pheromones after +1 iterations is: ,in l 1 and l 2 is for adjusting the step size coefficient. t min ( t ) is the first t The lower bound of pheromones in the next iteration. t min ( t +1) is the first t +1 iterations of pheromone lower bound t It is a positive integer.

[0054] It should be noted that when a better solution cannot be found for a long time, it usually indicates a globally optimal path. L gb The pheromone distribution on the path becomes extreme. For example, the globally optimal path... L gb The pheromone values ​​of the included path segments may become 8.0, 8.1, 7.9, and 8.05. These values ​​are very close, leading to a high coefficient of variation. CV Extremely low, far less than the target value CV 0 indicates that the ant population is highly concentrated.

[0055] Another scenario is that the pheromone value changes to 15.0, 0.2, 0.3, and 14.8, resulting in kurtosis. Kurt The value is abnormally high, far exceeding the target kurtosis. Kurt 0 (i.e., the desired kurtosis) KurtThe target value indicates that pheromones are excessively concentrated on a very few path segments, which also limits search diversity. Assuming the current pheromone lower limit... t min The kurtosis is 0.1. Kurt The calculated value is 8, while the target kurtosis is... Kurt 0 is 3, adjustment coefficient l 2 is 0.05. Calculate the new pheromone lower limit. t min ( t +1) will be approximately 0.125. By raising the minimum threshold for pheromones from 0.1 to 0.125, higher base pheromones can be forcibly assigned to path segments that are almost abandoned, preventing them from being completely ignored due to low pheromones and increasing the likelihood of escaping local optima.

[0056] Step S4: Iterate until the preset termination condition is met, and then determine the global optimal path obtained in the last iteration as the production control scheme.

[0057] The entire process of path construction and pheromone updating described above is treated as an iterative loop and repeated until a preset termination condition is met. For example, the total number of iterations reaches a preset upper limit of 500, or the globally optimal path is found. L gb The evaluation metric showed no improvement after 100 consecutive iterations; iteration stopped when any termination condition was met, at which point the globally optimal path was determined. L gb The processing sequence and equipment allocation scheme represented are the output production control scheme.

[0058] In Embodiment 2, the present invention provides a control system for cardboard box production, comprising the following modules: a parameter setting module, used to acquire a cardboard box production task containing N processes, initialize an ant colony into an exploration sub-swarm and a mining sub-swarm, and set operating parameters for the exploration sub-swarm and the mining sub-swarm respectively; a path construction module, used for the two sub-swarms to construct scheduling paths in parallel, and determine the probability of the ant selecting the next process j based on pheromones and composite heuristic information; the composite heuristic information is determined based on the expected waiting time of the bottleneck equipment of the candidate process j, the remaining total processing time, and the process switching cost; wherein the exploration sub-swarm introduces a disturbance factor inversely proportional to the structural overlap of the current globally optimal path when making a selection, and the mining sub-swarm makes a decision based on probability; a path and pheromone updating module, used for... After all ants have completed path construction, the globally optimal path and pheromone are updated: The process attempts to replace the bottleneck segment in the current globally optimal path with the corresponding process segment from the optimal path of the exploratory subpopulation, generating a new globally optimal path with better evaluation metrics; pheromone updates are performed, including pheromone enhancement only on path segments of the updated globally optimal path, rewarding the exploratory path with the greatest difference from the globally optimal path, and imposing a penalty on the worst path in the current iteration; the pheromone lower limit for the next iteration is updated based on the coefficient of variation and kurtosis of the pheromone values ​​distributed on each path segment of the globally optimal path; a scheme output module is used to repeatedly execute the path construction and update steps until a preset termination condition is met, outputting the final globally optimal path as a production control scheme.

[0059] The composite heuristic information is determined based on the expected waiting time of the bottleneck equipment of the candidate process j, the remaining total processing time, and the process switching cost. The exploratory subpopulation introduces a perturbation factor inversely proportional to the structural overlap of the current globally optimal path during selection.

[0060] Those skilled in the art will conceive of many modifications, alterations, and alternatives without departing from the spirit and essence of this invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A control method for paper box production, characterized in that, include: Get included N The paper box production task for each process involves obtaining an ant colony and initializing it into an exploration sub-species and a mining sub-species, then determining the operating parameters for these sub-species. N It is a positive integer; Ants exploring subpopulations and ants mining subpopulations construct scheduling paths in parallel, and determine the processing steps to be performed by ants based on pheromones and compound heuristic information. j The probability of composite heuristic information and candidate processes; where composite heuristic information and candidate processes are related. j The bottleneck equipment's expected waiting time, remaining total processing time, and process changeover costs are related; the selection of processing processes by ant subpopulations is explored. j The probability is also related to the perturbation factor, which is inversely proportional to the structural overlap of the current global optimal path. Ants mining subpopulations make decisions based on probability. After all ants exploring the subpopulation and all ants mining the subpopulation have completed path construction, update the globally optimal path and update the pheromone. The process involves replacing the bottleneck segment in the current global optimal path with the corresponding process segment in the optimal path of the exploration subpopulation to update the global optimal path; pheromone enhancement is applied to the path segments on the updated global optimal path, the exploration path with the greatest difference from the global optimal path is rewarded, and the worst path in the current iteration is penalized by evaporation; the pheromone lower limit for the next iteration is updated based on the coefficient of variation and kurtosis of the pheromone values ​​on each path segment in the global optimal path. The process is iterated until a preset termination condition is met, and then the globally optimal path obtained in the last iteration is determined as the production control scheme.

2. The control method for paper box production according to claim 1, characterized in that, Computational heuristic information includes: For the current process i Transfer to any candidate process j Obtain the execution procedure j Required equipment M j ; Calculation process j Expected device wait time EWT ij The waiting time value is the device M j The time to complete all assigned tasks and the current process i The difference between the completion times, where if the value of the waiting time is negative, then the waiting time is determined to be 0; The calculation includes candidate processes. j Total remaining processing time for all unscheduled processes, including RPT j ; Determine from the current process i Switch to candidate process j Process switching costs ST j ; Calculate the waiting time for the device. EWT ij The remaining total processing time RPT j and the process switching cost ST j The reciprocal of the weighted sum is used to obtain the composite heuristic information, wherein the composite heuristic information is calculated. η ij The formula is: ,in, β 1. β 2 and β 3 is a preset weighting coefficient, and β 1+ β 2+ β 3 = 1.

3. The control method for paper box production according to claim 1, characterized in that, The method by which the exploratory subpopulation introduces a perturbation factor inversely proportional to the structural overlap of the current globally optimal path during selection is as follows: for ants in the exploratory subpopulation... k In selecting candidate processes j Previously, the calculation of the constructed partial path was performed. P k Compared with the current globally optimal path L gb The number of identical process pairs is denoted as the structural overlap. SO ( k ); Calculate the ant's perturbation factor based on the structural overlap. D ( k The calculation formula is as follows: ; Calculate ants k by process i Select process j probability The calculation formula is as follows: ； l For all optional next steps, α , β and γ To preset weights, η ij From the process i Transfer to process j Heuristic information values, τ ij ( t ) is in the t In the next iteration, the path segment ( i , j The value of pheromones accumulated on the surface, η il From the process i Transfer to process l Heuristic information values, τ il ( t ) is in the t In the next iteration, the path segment ( i , l The value of pheromones accumulated on the surface.

4. The control method for paper box production according to claim 1, characterized in that, The attempt to replace the bottleneck segment in the current globally optimal path with the corresponding process segment in the optimal path of the exploratory subpopulation includes: Find the bottleneck device with the longest total load time in the current globally optimal path. M b ; Extract the bottleneck device from the current globally optimal path. M b The sequence of processes is denoted as S gb Extract the bottleneck device allocated to the optimal path in the current exploration subpopulation. M b The sequence of processes is denoted as S es ; Construct a candidate scheduling path, with bottleneck devices in the candidate scheduling path. M b The process sequence above adopts S es The process sequence on other equipment remains consistent with the current globally optimal path; Based on process dependencies and equipment occupancy constraints, the start and finish times of all affected processes in the candidate scheduling path are recalculated to obtain a complete and feasible scheduling scheme. If the total completion time of the candidate scheduling path is less than the total completion time of the current global optimal path, the global optimal path will be updated to a candidate scheduling path.

5. A control method for paper box production according to claim 1, characterized in that... Pheromone updates include: For all path segments ( i , j Perform global pheromone evaporation, where in the (t+1)th iteration, the updated pheromone is... The formula for calculating the updated pheromone is: , ρ The global volatility coefficient is set to a preset value. The pheromones before the update; The globally optimal path is enhanced, and the explored path is rewarded. The update formula is as follows: ； Among the increments In the path segment ( i , j Located on the globally optimal path L gb When the value is positive and it is close to the globally optimal path L gb The total completion time is inversely proportional to the increment. In the path segment ( i , j Located on the globally optimal path L gb The exploration path with the greatest difference L ed When the value is positive, the increment is... Exploration Path L ed The total completion time is inversely proportional to the total completion time; For the worst path in the current iteration L w Path segment on ( i , j A penalty evaporation is applied to obtain the pheromone in the current iteration; where the pheromone obtained in the (t+1)th iteration is obtained. The formula is: ,in ρ p This is the penalty evaporation coefficient; for the (t+1)th iteration, for paths not belonging to the worst path in the current iteration... L w The path segment in which the pheromone value remains at 0. .

6. The control method for paper box production according to claim 1, characterized in that, Update the pheromone lower bound for the next iteration, including: Extract the pheromone values ​​of all path segments on the current globally optimal path to form a set; Calculate the coefficient of variation of the set. CV with kurtosis Kurt ; Based on the preset target coefficient of variation CV 0 and target kurtosis Kurt 0 Update pheromone lower limit, including the update of the first t The lower bound of pheromones in the next iteration is obtained to obtain the first... t The formula for the lower limit of pheromones after +1 iterations is: ,in λ 1 and λ 2 is for adjusting the step size coefficient. τ min ( t ) is the first t The lower bound of pheromones in the next iteration. τ min ( t +1) is the first t +1 iterations of pheromone lower bound t It is a positive integer.

7. The control method for paper box production according to claim 1, characterized in that, The operational parameters for determining the exploration and mining subpopulations include: The pheromone importance factor and the composite heuristic information importance factor of the exploratory subpopulation are determined and denoted as the operating parameters of the exploratory subpopulation; The pheromone importance factor and the composite heuristic information importance factor of the mining subpopulation are determined and denoted as the operating parameters of the mining subpopulation. Among them, the pheromone importance factor of the mining subpopulation is greater than that of the exploration subpopulation, and the composite heuristic information importance factor of the exploration subpopulation is greater than that of the mining subpopulation.

8. A control system for paper box production, characterized in that, include: The parameter setting module is used to obtain parameters including... N In the paper box production task of each process, the ant colony is initialized into an exploration sub-species and a mining sub-species, and the operating parameters are set for the exploration sub-species and the mining sub-species respectively. The path construction module is used to construct scheduling paths in parallel by two sub-populations. It determines the probability of an ant selecting the next process j based on pheromones and composite heuristic information. The composite heuristic information is determined based on the expected waiting time of the bottleneck equipment of the candidate process j, the remaining total processing time, and the process switching cost. The exploration sub-population introduces a perturbation factor that is inversely proportional to the structural overlap of the current global optimal path when making a selection, while the mining sub-population makes a decision based on probability. The path and pheromone update module is used to update the globally optimal path and update the pheromone after all ants have completed path construction: it attempts to replace the bottleneck segment in the current globally optimal path with the corresponding process segment in the optimal path of the exploration subpopulation to generate a new globally optimal path with better evaluation indicators; it performs pheromone updates, including enhancing pheromone only on path segments of the updated globally optimal path, rewarding the exploration path with the greatest difference from the globally optimal path, and imposing a penalty on the worst path in the current iteration; and it updates the pheromone lower limit for the next iteration based on the coefficient of variation and kurtosis of the pheromone values ​​on each path segment of the globally optimal path. The solution output module is used to repeatedly execute the path construction and update steps until the preset termination conditions are met, and output the final global optimal path as the production control solution.

9. A control system for paper box production according to claim 8, characterized in that, The composite heuristic information is determined based on the expected waiting time of the bottleneck equipment for the candidate process j, the remaining total processing time, and the process switching cost.

10. A control system for paper box production according to claim 8, characterized in that, The exploration subpopulation is selected by introducing a perturbation factor that is inversely proportional to the structural overlap of the current global optimal path.

Images