Car lamp display-oriented backlight plate splicing and productivity balance scheduling method and system

Abstract

The invention relates to the technical field of data processing, in particular to a backlight plate splicing and productivity balance scheduling method and system for vehicle lamp display, and the method comprises the steps: constructing a splicing plate coupling graph, and extracting a backlight plate combination meeting the splicing plate constraint as a candidate splicing plate set; performing interval rough evaluation on the candidate jointed board set by combining upper and lower bounds of geometric density and coupling cost, and screening out a common candidate jointed board set and a guaranteed bottom recall jointed board set; and constructing a productivity balance model, evaluating the utilization rate and the process rhythm of the candidate jointed board combination on processing equipment such as a laser scribing machine, a film sticking machine and a cutting machine, realizing fine evaluation screening and scheduling simulation, and finally outputting the jointed board combination meeting the requirements of the delivery rate and the resource utilization efficiency. According to the method, the intelligent level of the jointed board scheduling can be effectively improved, the error rejection rate is reduced, and the overall productivity utilization rate and delivery stability of a production line are improved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and in particular to a method and system for scheduling panelization and production capacity balancing of automotive headlight display backlight panels. Background Technology

[0002] Existing technologies for panelizing and balancing production capacity in automotive headlight display backlight panels primarily follow two technical paths. The first employs a serial processing mode: First, static panelizing schemes are generated based on geometric matching and maximizing material utilization. Then, these fixed schemes are used as scheduling inputs, employing heuristic rules or metaheuristic algorithms to allocate equipment tasks and schedule their timing, aiming to meet delivery deadlines and improve equipment utilization. The second approach attempts integrated optimization: By constructing a multi-objective mathematical model, panelizing variables and scheduling variables are considered simultaneously, and optimization algorithms are used to search within a unified solution space, aiming to synergistically optimize indicators such as material costs and production efficiency.

[0003] For example, Chinese invention patent application CN119990698A discloses a capacity-balancing intelligent factory scheduling system based on MES, including: a preprocessing module that intelligently selects production strategies based on the current production mode, improving the flexibility and accuracy of production decisions; and a data execution module that ensures the stable execution of the production plan while promptly detecting and responding to changes in the production plan, avoiding chaos and delays in the production process. Furthermore, the introduction of a change scheduling module and a plan change module makes the production plan change process more scientific and efficient.

[0004] For example, Chinese invention patent CN114548573B discloses a method, terminal, and storage medium for intelligent splicing of sheet materials, belonging to the field of intelligent workshop manufacturing. The method includes acquiring data information of the sheet materials to be spliced ​​and the splicing target range. The data information of the sheet materials to be spliced ​​includes: the length of the sheet materials and the number of sheet materials corresponding to the length of the sheet materials. The splicing target range is the target length range that the spliced ​​sheet materials should be within. The method involves intelligent splicing processing of the data information of the sheet materials to be spliced, resulting in several splicing combinations of sheet materials. The length of each splicing combination of sheet materials is within the target length range.

[0005] Because existing technologies for automotive headlight display backlight panel panelizing generally rely on static geometric dimension matching rules, they ignore the deep coupling relationships between different backlight panels in terms of processing paths, process flows, and equipment occupancy. This leads to seemingly reasonable panelizing combinations frequently experiencing line changeover conflicts, process misalignments, and uneven equipment loads in actual production, severely restricting scheduling efficiency and delivery achievement rates. Furthermore, traditional solutions mostly use fixed scoring models and static thresholds for combination screening, failing to identify combinations with high scoring uncertainty but significant potential value. This results in many panelizing schemes with capacity optimization potential being prematurely eliminated in the initial screening stage, affecting the flexibility and resilience of the overall scheduling strategy. Summary of the Invention

[0006] To address the technical problem of insufficient scheduling flexibility caused by the accidental rejection of high-potential panelization schemes in existing technologies, this invention provides a method and system for panelization and capacity balancing scheduling of automotive lighting display backlight panels. The technical solution is as follows: On the one hand, a method for scheduling panelization and production capacity balancing for automotive lighting display backlight panels is provided, which includes: Step 1: Construct a panel coupling graph based on order demand data. Using each backlight panel as a node in the panel coupling graph, establish edges with mutual interference weights and constraint potential weights between any nodes that satisfy the panel constraints. Perform structural analysis on the panel coupling graph to identify the set of nodes that meet the panel combination requirements. Mark the set of nodes that meet the panel combination requirements as candidate panel combinations to obtain a candidate panel set that reflects panel compatibility and scheduling feasibility.

[0007] Step 2: Combine the upper and lower bounds of geometric density and coupling cost to perform a coarse evaluation of the candidate panel set in intervals. Based on the coarse evaluation results, obtain the ordinary candidate panel set that can directly enter the regular selection channel and the minimum recall panel set that needs to be reviewed. Construct a capacity balance model and evaluate the ordinary candidate panel set and the minimum recall panel set through the capacity balance model to determine the final refined evaluation candidate panel set.

[0008] Step 3: Perform detailed evaluation on each panel combination in the candidate panel set, and perform simulated scheduling based on the production line model and real-time capacity data to generate a panel capacity balance scheduling result that meets the target delivery rate and equipment utilization rate.

[0009] On the other hand, a scheduling system for panelization and production capacity balancing of automotive lighting display backlight panels is provided, including: The panel coupling modeling module is used to construct a panel coupling graph based on order demand data. Each backlight panel is used as a node in the panel coupling graph. Edges with mutual interference weights and constraint potential weights are established between any nodes that meet the panel constraints. The panel coupling graph is subjected to structural analysis processing to identify the set of nodes that meet the panel combination requirements. The set of nodes that meet the panel combination requirements is marked as a candidate panel combination, thereby obtaining a candidate panel set that reflects panel compatibility and scheduling feasibility.

[0010] The candidate combination screening module is used to perform interval-based coarse evaluation of the candidate panel set by combining the upper and lower bounds of geometric density and coupling cost. Based on the coarse evaluation results, ordinary candidate panel sets that can directly enter the regular selection channel and minimum recall panel sets that need to be reviewed are obtained. A capacity balance model is constructed, and the ordinary candidate panel set and minimum recall panel set are evaluated and processed by the capacity balance model to determine the final refined evaluation candidate panel set that enters the fine evaluation process.

[0011] The scheduling evaluation generation module is used to perform evaluation processing on each panel combination in the evaluation candidate panel set, perform simulated scheduling based on the production line model and real-time capacity data, and generate a panel capacity balance scheduling result that meets the target delivery rate and equipment utilization rate.

[0012] The beneficial effects of the technical solutions provided in the embodiments of the present invention include at least the following: (1) By establishing a panel coupling graph with the backlight panel as the node and the panel relationship as the edge, and simultaneously applying hard constraints based on geometric boundaries and device compatibility for millisecond-level filtering, feasible panel solutions can be quickly and accurately identified from a massive number of combinations. This dual-drive approach of graph structure analysis and real-time constraint verification, compared with existing methods that rely on sequential judgment or single-index screening, not only improves the screening speed from hours to milliseconds, but also avoids potential process conflicts in subsequent scheduling in advance, ensuring the feasibility of the panel solution from the source, and achieving a 1+1>2 effect in efficiency and quality.

[0013] (2) An interval-based scoring method based on the upper and lower bounds of geometric density and coupling cost is adopted, combined with a backup recall mechanism that actively identifies high-potential or high-uncertainty combinations, to construct a two-layer evaluation system of main channel screening + edge optimal solution fallback. Compared with the problem of potential optimal solutions being mistakenly rejected due to the use of a single deterministic value in the prior art, this invention can not only perform fast channel processing for clearly high-quality combinations, but also retain review opportunities for special combinations at the decision boundary, which significantly improves the decision flexibility and robustness of the system, and better solves the contradiction between not missing optimal solutions and controlling computational cost in complex scheduling scenarios.

[0014] (3) Based on the candidate set output by the front-end coarse evaluation, a fine evaluation and selection is carried out through the capacity balance model. At the same time, the uncertainty threshold of the coarse evaluation stage is adjusted according to the feedback of the fine evaluation results, forming a closed-loop learning capability of front-end evaluation-back-end verification-parameter tuning. Compared with the static system in the prior art with fixed parameter settings and lack of self-optimization capability, the present invention enables the panel scheduling system to continuously optimize its screening strategy according to actual production feedback. This not only ensures the accuracy of a single decision, but also has the long-term evolution capability to adapt to changes in the production environment, and better solves the problem of performance degradation caused by long-term operation of the system.

[0015] (4) In the refinement stage, based on the production line model and real-time capacity data, a multi-dimensional scheduling simulation is performed on the panel assembly, including equipment utilization rate, process balance rate, and delivery achievement rate. This integrated processing of panel assembly optimization and capacity embedding verification, compared with the existing technology that treats material optimization and capacity scheduling as two independent sequential processes, can directly produce a scheduling scheme that is deeply consistent with the real-time status of the production line and takes into account both material utilization and equipment load. It fundamentally solves the industry pain point of panel scheme optimization on paper and difficulty in implementation in actual production, and maximizes the overall production efficiency. Attached Figure Description

[0016] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0017] Figure 1 This is a flowchart of the method for balancing and scheduling production capacity of backlight panel for vehicle headlight display provided in an embodiment of the present invention; Figure 2 This is a schematic diagram of the structure of the automotive headlight display backlight panel panel and production capacity balancing scheduling system provided in an embodiment of the present invention; Figure 3 This is a flowchart of panel coupling analysis and candidate set generation provided in an embodiment of the present invention; Figure 4 This is a flowchart of the fine evaluation scheduling and dynamic self-optimization provided in the embodiments of the present invention. Detailed Implementation

[0018] The technical solution of the present invention will now be described with reference to the accompanying drawings.

[0019] In embodiments of the present invention, words such as "exemplarily," "for example," etc., are used to indicate that something is an example, illustration, or description. Any embodiment or design described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or designs. Specifically, the use of the word "exemplary" is intended to present the concept in a concrete manner. Furthermore, in embodiments of the present invention, the meaning expressed by "and / or" can be both, or either one.

[0020] In this embodiment of the invention, sometimes a subscript such as W1 may be written in a non-subscript form such as W1. When the difference is not emphasized, the meaning they express is the same.

[0021] To make the technical problems, technical solutions and advantages of the present invention clearer, a detailed description will be given below in conjunction with the accompanying drawings and specific embodiments.

[0022] This invention provides a method for scheduling and balancing production capacity for automotive lighting display backlight panel panelization, such as... Figure 1 The flowchart shown is for the panelization and capacity balancing scheduling method of automotive headlight display backlight panels. The processing flow of this method may include the following steps: Step 1: Construct a panel coupling graph based on order demand data. Using each backlight panel as a node in the panel coupling graph, establish edges with mutual interference weights and constraint potential weights between any nodes that satisfy the panel constraints. Perform structural analysis on the panel coupling graph to identify the set of nodes that meet the panel combination requirements. Mark the set of nodes that meet the panel combination requirements as candidate panel combinations to obtain a candidate panel set that reflects panel compatibility and scheduling feasibility.

[0023] Step 2: Combine the upper and lower bounds of geometric density and coupling cost to perform a coarse evaluation of the candidate panel set in intervals. Based on the coarse evaluation results, obtain the ordinary candidate panel set that can directly enter the regular selection channel and the backup recall panel set that needs to be reviewed in order to prevent false rejection. Construct a capacity balance model and evaluate the ordinary candidate panel set and the backup recall panel set through the capacity balance model to determine the final refined evaluation candidate panel set.

[0024] Step 3: Perform detailed evaluation on each panel combination in the candidate panel set, and perform simulated scheduling based on the production line model and real-time capacity data to generate a panel capacity balance scheduling result that meets the target delivery rate and equipment utilization rate.

[0025] Traditional solutions mostly employ fixed scoring models and static thresholds for combined screening, failing to identify combinations with high scoring uncertainty but significant potential value. In a specific example, during a scheduling process, there exists a panel combination consisting of two different backlight panel models, A and B. This combination has only average geometric fit, resulting in a low initial score in the traditional scoring model. Furthermore, its slightly complex processing path leads to significant fluctuations in the coarse scoring model's assessment of its processing efficiency—high score uncertainty—and thus it is marked as a low-priority combination for elimination. However, further analysis reveals that backlight panels A and B share the switching window between the laser scribing machine and the laminating machine. Moreover, under the current equipment usage conditions on the production line, this combination can avoid peak periods in the cutting section, significantly shortening the overall production cycle time. In other words, while this combination's structural potential was not accurately identified in the initial scoring, it demonstrates significantly better delivery performance and equipment load balancing effects than conventional combinations in the capacity simulation model, possessing extremely high capacity optimization value.

[0026] In this embodiment of the invention, the panel coupling diagram is a graph structure model used to depict the combination relationship between different backlight panel models and their impact on the capacity balance of the scheduling system.

[0027] Specifically, the process of establishing edges with mutual interference weights and constraint potential weights between any nodes that satisfy the panel constraints, using each backlight panel as a node in the panel coupling graph, is as follows: The node has backlight panel attributes, including the backlight panel model code (used to identify the function type or form factor version), geometric dimensions (including length, width and board boundary), quantity requirement (representing the total order demand for this model of backlight panel), and delivery cycle (including the earliest production start time and the latest delivery time).

[0028] The panel constraints are used to evaluate whether any nodes have the physical and technological feasibility to be combined and processed in the same panel process, specifically including geometric compatibility, fixture adaptability, and process consistency.

[0029] The aforementioned geometric compatibility refers to the ability of any nodes to be arranged on the same panel in a two-dimensional panel layout, without overlapping, occlusion, or violation of the minimum safety margin, and to meet the panel layout rules. The compatibility can be determined by calculating the geometric contours, aspect ratios, and overlap of the bounding boxes of the two backlight panels after rotation.

[0030] The aforementioned fixture adaptability refers to whether any node can be fixed by the same tooling fixture, and whether the machining process will result in insufficient clamping force or fixture deformation due to dimensional differences; the judgment criteria may include parameters such as the effective clamping range of the fixture and the clamping tolerance.

[0031] The above process consistency refers to whether the key parameters (such as scribing depth, number of film layers, cutting method, etc.) between any nodes in the subsequent processing flow are consistent or can be processed on the same line, so as to avoid processing failure, frequent line change or path readjustment due to inconsistent parameters.

[0032] Establish edges between any nodes that satisfy the puzzle constraint conditions, with mutual interference weights to characterize the degree of puzzle interference between nodes and constraint potential weights to characterize the ability of puzzle combinations to adjust under the resource and time dimensions.

[0033] The specific calculation process for the above mutual interference weights is as follows: ;ρ i，j D represents the mutual interference weight between backlight panels i and j. geom (i, j) is the geometric dimension difference index between backlight panels i and j, reflecting the edge interference risk caused by the dimension difference between the two backlight panels in the panel. It is obtained by extracting the aspect ratio and safety distance from CAD data and establishing a normalized difference function.

[0034] D path (i, j) represents the process path conflict degree between backlight panels i and j, which is used to measure the degree of interference caused by the spatial overlap and temporal overlap of the paths during the panelization process. By inputting the process paths of backlight panels i and j in the panelization process into the process simulation platform or CAM software, the processing path trajectories and time axes of the two backlight panels under key processes are analyzed, the length of their spatial overlap segment and the duration of their temporal overlap are statistically analyzed, and the normalized overlap index is used to quantify the process path conflict degree caused by spatial and temporal conflicts in the panelization process.

[0035] D fixture (i, j) represents the clamping tension inconsistency between backlight panel i and j. The clamping force feedback data of each clamping point is collected in real time using a simulation platform. The maximum and minimum values ​​of the clamping force ratio or standard deviation and other statistical quantities are calculated and normalized as the quantitative result of the clamping tension inconsistency.

[0036] α1 is the weighting coefficient for the geometric dimension difference index, α2 is the weighting coefficient for the process path conflict degree, and α3 is the weighting coefficient for the fixture tension inconsistency degree. The values ​​of α1, α2, and α3 can be obtained by setting them through expert experience or by fitting and training with historical data. i and j represent the numbers of any two different backlight panels in the set of panels to be assembled. The set of backlight panels is denoted as: B={b1, b2, ..., bn}, bi represents the i-th backlight panel in the set, bj represents the j-th backlight panel in the set, and i≠j.

[0037] The specific calculation process for the aforementioned constraint potential weights is as follows: ;Φ i，j R represents the constraint potential weights of backlight panels i and j.time (i, j) represents the overlap rate of the time windows of backlight panels i and j, indicating whether the two backlight panels can be produced together within the delivery tolerance. By extracting the order delivery range of backlight panels i and j (i.e., their earliest production time and latest delivery time), the intersection length of their time intervals is calculated, and the intersection length is divided by the smaller or larger value of the lengths of their time intervals to obtain the normalized overlap ratio, which is used as the time window overlap rate.

[0038] R capacity (i, j) represents the compatibility of available production capacity for backlight panel i and j (e.g., whether they occupy the same equipment or have parallel workstations). The process equipment and workstation allocation information of the two backlight panels are obtained through the MES system. Combined with real-time production capacity data, the degree of overlap of the equipment or parallel workstations that can be shared in the same period is calculated, and the compatibility of available production capacity is normalized.

[0039] R priority (i, j) represents the customer priority matching score between backlight boards i and j (whether they are at the same priority level, or whether mixed-use is allowed). The customer priority field (e.g., high, medium, low) of the orders corresponding to backlight boards i and j is extracted from the order system. This is then combined with the panelization strategy rule base to determine whether they belong to the same priority level, or whether the current process strategy allows mixed-use of orders with different priorities. If the priorities are completely identical, the matching score is recorded as 1 (complete matching); if mixed-use is allowed but there are priority differences, the corresponding matching score is assigned according to the strategy settings; if the strategy prohibits mixed-use, the matching score is 0.

[0040] β1 is the weighting coefficient for time window overlap rate, β2 is the weighting coefficient for equipment available capacity suitability, and β3 is the weighting coefficient for customer priority suitability. β1, β2, and β3 can be determined by fitting historical scheduling success rates or by setting rules.

[0041] In a specific example embodiment, for backlight panel A (delivery time 3 days, size 120×80mm, high priority) and backlight panel B (delivery time 5 days, size 125×85mm, medium priority), the mutual interference weight is calculated to be 0.62 and the constraint potential weight is 0.78 by the above formula. The two meet the threshold conditions (mutual interference weight threshold is 0.7 and constraint potential weight threshold is 0.75), so the panel combination edge can be established.

[0042] Specifically, the splicing diagram is subjected to structural analysis, and the specific processing procedure is as follows: Based on the relationship between nodes and edges in the panel coupling graph, substructure extraction is performed on the panel coupling graph to identify a connected subgraph composed of several nodes with connected edges. Each node in the connected subgraph represents a backlight panel with paneling feasibility, and any two nodes are connected to each other through edges representing paneling compatibility. For each set of nodes in the connected subgraph, it is determined whether the set of nodes as a whole meets the paneling combination requirements.

[0043] If the set of nodes in a certain connected subgraph satisfies the requirements for panel combination, then the backlight panel corresponding to that set of nodes is marked as a candidate panel combination, and all candidate panel combinations are organized into a candidate panel set; if the set of nodes in a certain connected subgraph does not satisfy the requirements for panel combination, then the backlight panel corresponding to that set of nodes is marked as an invalid panel combination and is removed.

[0044] By using the above-mentioned method for extracting subgraphs and determining the effectiveness of combinations based on the structural relationship of the panel coupling graph, it is possible to quickly identify backlight panel sets that have the potential for combination processing in terms of structural compatibility and process adaptability under the large-scale order panel requirements. This effectively eliminates invalid combinations with serious physical conflicts or tight resource scheduling, thereby improving the quality of panel solutions.

[0045] Compared to traditional pairwise judgment or rule hard coding methods, this method combines graph structure topology and weight quantification features, which not only improves the efficiency and accuracy of panel combination screening, but also provides a structured candidate space for subsequent interval scoring, resource assessment and scheduling simulation. This enables the synergistic optimization of panel feasibility and capacity controllability, and enhances the execution stability of the overall production plan and the utilization rate of production line resources.

[0046] Furthermore, the specific judgment process for identifying the set of nodes that meet the requirements for panel assembly is as follows: The mutual interference weight and constraint potential weight of any side in each connected subgraph are compared with the mutual interference weight threshold and constraint potential weight threshold, respectively. The mutual interference weight threshold refers to the upper limit of the mutual interference weight extracted from the management database within a specified range, and the constraint potential weight threshold refers to the lower limit of the constraint potential weight extracted from the management database within a specified range.

[0047] The management database is used to centrally store various constraint parameters and evaluation thresholds required in the panelization analysis and capacity balancing scheduling process. These thresholds include, but are not limited to, control parameters across multiple dimensions such as mutual interference weight thresholds, constraint potential weight thresholds, uncertainty adjustment thresholds, and equipment utilization target values. The settings of these thresholds are not fixed and manually configured, but rather dynamically generated based on multi-source information such as historical production data, equipment operation logs, order scheduling records, and actual delivery status, through statistical analysis, rule mining, and model learning.

[0048] If the mutual interference weight of any side in a connected subgraph is greater than or equal to the mutual interference weight threshold, then the node set in the connected subgraph as a whole does not meet the requirements for panel combination. If the constraint potential weight of any side in a connected subgraph is less than the constraint potential weight threshold, then the node set in the connected subgraph as a whole does not meet the requirements for panel combination. If the mutual interference weight of any side in a connected subgraph is less than the mutual interference weight threshold, and the constraint potential weight is greater than or equal to the constraint potential weight threshold, then the node set in the connected subgraph as a whole meets the requirements for panel combination.

[0049] By adopting the dual threshold joint judgment mechanism of mutual interference weight and constraint potential weight, it is possible to effectively screen out panel combinations that may cause scheduling failure under the circumstances of geometric space conflict, fixture process incompatibility or resource scheduling tension, thereby improving the physical feasibility and plan deliverability of candidate panel solutions.

[0050] Specifically, the candidate tile set is coarsely evaluated by combining the upper and lower bounds of geometric density and coupling cost. The specific scoring process is as follows: The upper bound of geometric density, reflecting the balance of image utilization in each candidate panel combination, is obtained by using the ratio of the area of ​​the largest single backlight panel in each candidate panel combination to the total area of ​​the combination; specifically: ;D max S is the upper bound of the geometric density, ranging from (0, 1]. max S represents the area of ​​the largest backlight panel in this candidate panel combination. sum The system automatically analyzes the geometric dimensions of each backlight panel in the assembly to calculate the sum of the areas of all backlight panels in the panel assembly, and extracts S through the area calculation module. max S sum .

[0051] By determining the theoretically most compact layout area achievable through the minimum safety margin requirement in each candidate panel combination, a lower bound for the geometric density reflecting the maximum possible area utilization efficiency of each candidate panel combination is obtained; specifically: ;D min S is the lower bound of geometric density. theoretical_min To consider the most compact total area that can be arranged under the minimum safety margin, the system calculates it by combining the shape of each backlight panel, margin requirements, and arrangement optimization algorithms (such as minimum boundary rectangle and approximation of optimal packing).

[0052] Geometric density upper bound D max As a representative of area distribution uniformity, and the lower bound of geometric density D minThe multidimensional coarse screening and boundary assessment of candidate panel combinations jointly reflect the maximum equilibrium potential and the ultimate utilization efficiency, respectively. These two have different physical meanings and do not have a fixed numerical relationship. In common scenarios, it is normal for the lower bound value to be greater than the upper bound value, demonstrating the innovation and engineering applicability of this multidimensional coarse evaluation scheme.

[0053] By weighting the process switching complexity of each backlight panel in each candidate panel combination with the processing equipment occupancy time, an upper bound for the coupling cost reflecting the process switching cost of each candidate panel combination is obtained; specifically: C max As an upper bound for coupling cost, C (i) switch The normalized score for the process switching complexity of the i-th backlight panel is obtained from the corresponding model switching sequence mapping table in the equipment database. This table, indexed by backlight panel model pairs, records the standardized data of adjustment steps and complexity scores required during process switching, providing a basis for automated scheduling and process compatibility assessment; T (i) machine Let α be the normalized processing time of the i-th backlight panel on the specified equipment, and β be empirical weighting coefficients (e.g., α=0.6, β=0.4), obtained through historical production line optimization.

[0054] By calculating the historical panel collinearity rate, path overlap rate, and process compatibility index of the backlight panels in each candidate panel combination, a lower bound for the coupling cost reflecting the minimum interference level of each candidate panel combination is obtained; specifically: C min This is a lower bound for coupling cost; the smaller the value, the more compatible the coupling.

[0055] R colinear To determine the historical collinearity rate of backlight panels in the assembly, the process path trajectory data of each backlight panel in the historical panel record database is extracted. The similarity of all backlight panel process paths is calculated by pairwise pairwise using trajectory similarity algorithms such as dynamic time warping. Finally, the average of the similarity scores is taken as the collinearity rate index reflecting the consistency of the assembly process paths.

[0056] R overlap The processing path overlap rate represents the proportion of paths overlapping in the same process segment. By extracting the processing path trajectory and process segment time axis of each backlight panel from the historical panel record library, the overlapping area of ​​the spatial trajectory and time interval of all backlight panel pairs in the same process segment is determined. The overlap ratio of trajectory points or duration is statistically calculated and the average value is taken as a standardized indicator to measure the degree of overlap of the processing paths of panel assembly.

[0057] I compatAs a process compatibility index, the core process parameters of each backlight panel in the key process section are extracted from the historical panel record library. The consistency of all parameter pairs is calculated by using similarity measurement methods such as parameter normalization and Euclidean distance, and the average value is taken as the overall compatibility score to measure the consistency and compatibility of the backlight panels in the panel assembly in terms of process parameters.

[0058] γ, δ, and η are empirical parameter weights that can be optimized through actual production line backtracking; historical panel collinearity rate, path overlap rate, and process compatibility index can be calculated by extracting logs such as processing paths and temperature trajectories from the historical panel record library, and then using similarity measurement and compatibility model.

[0059] The coarse upper bound values ​​for each candidate panel combination are obtained based on the upper bound of geometric density and the lower bound of coupling cost; specifically: S upper To coarsely evaluate the upper bound, a coarse lower bound for each candidate panel combination is obtained based on the lower bound of geometric density and the upper bound of coupling cost; specifically: S lower For the rough evaluation of the lower bound, w1 and w2 are the weights of geometric utilization and process cost, satisfying w1+w2=1. For example, w1=0.5, which can be adjusted according to the bottleneck focus of the production line.

[0060] Through the aforementioned interval-based coarse scoring method, this invention achieves a dual-dimensional fusion evaluation mechanism of geometric and process characteristics in the process of panelization and capacity balancing scheduling for automotive headlight display backlight panels. This method not only accurately reflects the balance between panel utilization efficiency and process switching costs in the early screening stage, but also effectively characterizes the uncertainty range of the combination through upper and lower bound interval expressions, thus avoiding the problems of traditional single-value scoring models being susceptible to abnormal data and having large screening errors. With this interval-based coarse scoring system, the system can significantly improve the stability and accuracy of panel selection while maintaining computational efficiency, enabling subsequent capacity balancing models to perform optimization calculations based on a more reliable input set. Ultimately, this achieves a coordinated balance between panel compatibility, production cycle time, and equipment load, improving the resource utilization and delivery achievement rate of the entire production line.

[0061] Specifically, the analysis process for obtaining the ordinary candidate panel set that can directly enter the regular selection channel and the backup recall panel set that requires in-depth review based on the preliminary evaluation results is as follows: All candidate puzzle pieces are sorted in descending order according to the upper bound of the coarse evaluation of each candidate puzzle piece combination. The candidate puzzle pieces whose ranking of the upper bound of the coarse evaluation is within the preset target number range are selected as the ordinary candidate puzzle piece set. For candidate puzzle pieces whose ranking is outside the preset target number range, if the upper bound of the coarse evaluation of a certain candidate puzzle piece combination is higher than the lowest value of the lower bound of the coarse evaluation in the ordinary candidate puzzle piece set, then the candidate puzzle piece combination is divided into the minimum recall puzzle piece set. The preset target number is automatically determined by the system based on computing resources, scheduling window and equipment load.

[0062] The coarse evaluation uncertainty is obtained by using the upper and lower bounds of the coarse evaluation of candidate puzzle combinations. The coarse evaluation uncertainty refers to the result of subtracting the lower bound from the upper bound of the coarse evaluation of candidate puzzle combinations. For candidate puzzle combinations whose ranking is outside the preset target number range, if the coarse evaluation uncertainty of a candidate puzzle combination is higher than the coarse evaluation uncertainty threshold, the candidate puzzle combination is classified into the minimum recall puzzle set. The coarse evaluation uncertainty threshold is dynamically generated by the empirical statistical results stored in the management database and is used to constrain the recall scope in the coarse evaluation stage.

[0063] By employing the aforementioned hierarchical screening and recall strategy based on interval coarse scoring results, this invention achieves dual optimization of the accuracy and robustness of the candidate panel set. This method not only ensures that high-scoring combinations are prioritized for subsequent evaluation stages but also prevents combinations with potentially optimal performance but high scoring uncertainty from being prematurely eliminated through a recall mechanism. This significantly improves the overall recall rate and decision security of the system in complex multi-model panel scenarios. Furthermore, by dynamically adjusting the uncertainty threshold, the system can adaptively control the size of the backup recall pool based on real-time production line feedback. This ensures the retention of optimal solutions while avoiding the waste of redundant computing resources, making the panel selection and capacity balancing scheduling process more efficient, controllable, and stable.

[0064] like Figure 3The flowchart for panel coupling analysis and candidate set generation provided in this embodiment of the invention clearly illustrates the complete process from panel coupling analysis to candidate set generation. A panel coupling graph is established with each backlight panel as a node. Then, preliminary screening is performed by determining whether the connected subgraphs meet the panel combination requirements. This determination leads the process in two directions: if the determination is yes, the panel combination is recorded as a candidate panel combination; if the determination is no, it is marked as an invalid panel combination and deleted. Panel combinations selected through this screening collectively form a candidate panel set, completing the preliminary screening. The upper and lower bounds of geometric density and coupling cost are calculated to obtain the upper and lower interfaces of the program. This step establishes a quantitative evaluation interval for each panel combination. Based on this, the system sorts the candidate panel combinations in descending order according to the coarse evaluation upper bound value, selects candidate panel combinations within a preset target number range, and divides them into a normal candidate panel set. This sorting and selection mechanism ensures that high-quality combinations are given priority. For candidate panel combinations ranked outside the preset target number range, the process sets a double safety net mechanism. The algorithm determines whether the upper bound of the coarse evaluation of candidate puzzle combinations is higher than the lowest value of the lower bound of the coarse evaluation in the ordinary candidate puzzle set. If the condition is met, the puzzle is included in the minimum recall set; otherwise, it is deleted. Simultaneously, it checks whether the uncertainty of the coarse evaluation of candidate puzzle combinations is higher than the coarse evaluation uncertainty threshold. If this condition is met, the puzzle is also included in the minimum recall set; otherwise, it is deleted. This design effectively prevents the omission of potentially high-quality solutions.

[0065] Furthermore, the evaluation and processing of the ordinary candidate panel set and the guaranteed recall panel set using the capacity balancing model is as follows: Using the ordinary candidate panel set and the guaranteed recall panel set as input, the system comprehensively evaluates each panel combination through a capacity balance model, outputs a refined candidate panel set, and performs refined evaluation on each panel combination in the refined candidate panel set.

[0066] The set of ordinary candidate puzzle pieces selected through the coarse evaluation stage and the set of guaranteed recall puzzle pieces are used as the model input, denoted as set P. norm With P recall The two are combined into a complete set of candidate pieces for panel evaluation, P. eval =P norm ∪P recall The capacity balance model aims to balance each panel combination p i ∈P eval The quantitative scoring primarily considers the following dimensions: Delivery target fulfillment rate (G) di This indicates whether the panel combination can meet the order delivery requirements. The calculation method can include the ratio of order demand covered by a single panel to the corresponding delivery date; batch process synchronization index G. si This assesses whether there are synchronization conflicts in the manufacturing processes of the different backlight panel models included in the panel assembly; resource conflict penalty item G. ciPenalty weighting is set for situations such as processing cycle conflicts and cross-process resource competition.

[0067] The following is an example of a model scoring function: ; Among them, τ1 is the weighting coefficient of the delivery target satisfaction rate indicator, τ2 is the weighting coefficient of the batch process synchronization indicator, and τ3 is the weighting coefficient of the resource conflict penalty item. τ1, τ2 and τ3 can be called from the management database according to the business strategy.

[0068] The model applies to each puzzle combination p i ∈P eval Calculate its overall score S i The system sorts the puzzle pieces according to their scores from highest to lowest. Based on a preset number or score threshold, the system selects a set of puzzle pieces P as the candidate set for refined evaluation. refined This is used for the next stage of detailed evaluation and simulated scheduling.

[0069] Simultaneously, dynamic self-optimization processing is performed on the preliminary evaluation stage based on the refined evaluation candidate panel set.

[0070] This step introduces a scoring mechanism based on multi-objective capacity balancing to achieve comprehensive performance analysis of candidate panelization combinations, significantly improving the controllability and engineering adaptability of the screening stage. Compared to traditional panelization screening methods based on a single scoring dimension or heuristic rules, this invention explicitly models key indicators such as delivery time fulfillment, equipment load balancing, and process coordination, and implements punitive control over resource conflicts. This ensures that panelization combinations entering the fine-tuning stage not only have structural rationality but also operability for production scheduling. Furthermore, through the parameterized design of a unified scoring function, the system possesses good scalability and dynamic adjustment capabilities, adapting to the panelization strategy optimization needs under different production cycles or business scenarios, enhancing the intelligence and robustness of the entire panelization scheduling system.

[0071] Specifically, the dynamic self-optimization process for the coarse scoring stage based on the refined evaluation candidate panel set is as follows: Based on the number of puzzle combinations originating from the minimum recall puzzle set in the refined candidate puzzle set, the minimum acceptance rate is obtained by dividing the number of minimum puzzle combinations included in the refined candidate puzzle set by the total number of puzzle combinations included in the minimum recall puzzle set. After obtaining minimum acceptance rate data for multiple consecutive target periods (e.g., 5 or 10 consecutive statistical periods), linear regression analysis is performed on this time series data. The slope of the resulting regression line is the minimum acceptance slope. This slope quantitatively reflects the changing trend of the effectiveness of the minimum recall strategy within consecutive periods.

[0072] If the guaranteed adoption slope is not lower than the guaranteed adoption slope threshold, the uncertainty threshold in the coarse scoring stage is reduced to expand the retention range of potential usable combinations; if the guaranteed adoption slope is lower than the guaranteed adoption slope threshold, the uncertainty threshold is increased to reduce the resource consumption of the guaranteed recall panel set; the guaranteed adoption slope threshold is used to control whether to trigger the dynamic boundary point of threshold adjustment, and its stable range is generally determined by the historical fluctuation range, such as a floating window of ±5%.

[0073] Specifically: U= U is the adjusted uncertainty threshold, UT is the uncertainty threshold for the current period, r is the minimum adoption slope, R is the minimum adoption slope threshold, and μ1 and μ2 represent the adjustment step size parameters in different intervals, which can be calibrated through simulation or experience. The dynamic self-optimization of the uncertainty threshold can be achieved through this formula.

[0074] This invention constructs a dynamic adaptive mechanism for the coarse scoring stage based on refined evaluation feedback, achieving closed-loop feedback adjustment between the scoring strategy and actual adoption results. This improves the system's flexibility in identifying candidate panel combinations and enhances overall resource allocation efficiency. Compared to traditional static scoring strategies, which may lead to the mistaken rejection of high-potential panel combinations or redundant calculations due to lenient scoring, this invention can adjust the size and retention range of the safety net recall pool in real time based on historical adoption trends, effectively reducing global performance fluctuations caused by accumulated scoring errors. Furthermore, this adjustment mechanism features a time sliding window characteristic, adapting to real-world scenarios such as batch fluctuations and order structure changes, enhancing the robustness, adaptability, and engineering practical value of the panel combination screening and scheduling modeling system.

[0075] like Figure 4 As shown in the flowchart of the refined evaluation scheduling and dynamic self-optimization provided in this embodiment of the invention, the process begins with a comprehensive evaluation of the capacity balance model. The evaluation results output a set of candidate panels for refined evaluation, which then undergoes multi-dimensional scheduling simulation analysis. Finally, the final panel capacity balance scheduling result is generated through scheduling simulation. Simultaneously, the system executes a feedback optimization mechanism in parallel: calculating the minimum acceptance rate and minimum acceptance slope in the set of candidate panels for refined evaluation, and dynamically adjusting based on the judgment result that the minimum acceptance slope is not lower than the minimum acceptance slope threshold. If the judgment is yes, the uncertainty threshold is reduced, and the retention range is increased; if the judgment is no, the uncertainty threshold is increased, and resource consumption is reduced. This optimization result is used as a parameter update and fed back to the coarse evaluation stage, forming a continuous improvement closed loop.

[0076] Furthermore, the specific process of performing refined evaluation on each puzzle combination in the refined evaluation candidate puzzle set is as follows: Based on the production line model and real-time capacity data of the automotive lighting display backlight panel production line, a multi-dimensional scheduling simulation analysis is performed on each candidate panel combination in the refined evaluation candidate panel set.

[0077] The production line model includes, but is not limited to, equipment capacity parameters, process cycle time parameters, resource constraint parameters, and workstation switching time parameters, which are used to characterize the dynamic capacity characteristics of the production line. Equipment capacity parameters are used to describe the types of processes supported by different equipment, maximum processing capacity, and current available load. Process cycle time parameters are used to describe the cycle time characteristics such as the standard time required for each processing process, switching interval, and synchronization window. Resource constraint parameters are used to define the feasibility range of scheduling resources such as raw materials, tooling fixtures, and operators. Workstation switching time parameters are used to characterize the switching time consumption and conversion mode of equipment or workstations during the assembly and switching process.

[0078] Based on real-time capacity data, the available time window and load status of each device are updated, and the scheduling feasibility verification and timing allocation are performed for each candidate panel combination in the refined evaluation candidate panel set.

[0079] The scheduling feasibility verification refers to the matching verification of all processing steps included in the panel assembly with the real-time available time window and dynamic load status of each piece of equipment in the production line model. The specific verification content includes: process dependency verification, equipment capacity and conflict verification, and time window feasibility verification.

[0080] The aforementioned time-series allocation refers to, under the premise of verifying the feasibility of the aforementioned scheduling, clearly assigning specific execution equipment to each process in the panel assembly, calculating optimized start and end times, thereby forming a conflict-free and executable detailed production sequence plan. The specific allocation process includes: equipment assignment, time window calculation, and buffer time insertion.

[0081] Based on the simulation results of the production line model, the equipment utilization rate, process balance rate, and delivery achievement rate of each panel assembly within the processing cycle are calculated. A comprehensive evaluation function, weighted by multiple indicators, is constructed based on these factors to uniformly measure the overall performance of each panel assembly in terms of equipment resource utilization, process coordination, and delivery target achievement. This comprehensive evaluation function can be constructed using linear weighting, piecewise scoring, fuzzy functions, or learning optimization strategies, with the goal of finding the optimal solution under multi-objective constraints. For example, a comprehensive evaluation function can be constructed using a linear weighted summation method. F is the comprehensive evaluation function, SU is the equipment utilization rate, calculated as the ratio of the sum of the actual processing time of all occupied equipment to the total available time of the equipment during the simulation cycle, ξ1 is the weighting coefficient of equipment utilization, SB is the process balance rate, calculated as the reciprocal of the standard deviation of the cycle time of each process on the critical process path of the panel assembly, ξ2 is the weighting coefficient of process balance rate, SD is the delivery achievement rate, which is a binary or piecewise function. If the simulation completion time of the panel assembly is earlier than the latest delivery time of the order, it is recorded as 1 (or full score); if there is a delay, points are deducted according to the degree of delay according to preset rules, ξ3 is the weighting coefficient of delivery achievement rate. ξ1, ξ2 and ξ3 are a set of configurable strategic parameters stored in the system management database, which can be dynamically called and adjusted according to different business objectives (such as pursuing maximum capacity or ensuring on-time delivery).

[0082] Based on the scoring results of the comprehensive evaluation function, the panel combinations in the candidate panel set are screened. Under the premise of meeting the constraints of target delivery rate and equipment utilization rate, several panel combinations with the best or second best scores are extracted to form the final output panel capacity balance scheduling result, which is issued to the production line scheduling system as an actual production instruction.

[0083] In one specific embodiment, to address issues such as uneven process flow, low equipment resource utilization, and difficulty in guaranteeing delivery time during the production scheduling of automotive headlight display backlight panels, this system uses a refined candidate panel set as input and combines it with the real-time status of the production line to conduct capacity balance optimization simulation. This production line mainly includes three core processes: the laser engraving process is handled by three laser processing machines, namely high-speed laser engraving machine No. 1, high-speed laser engraving machine No. 2, and high-speed laser engraving machine No. 3; the film application process is completed by automatic film application machines No. 1 and No. 2; and the cutting process is performed by three high-speed cutting machines, namely intelligent precision cutting machine No. 1, intelligent precision cutting machine No. 2, and intelligent precision cutting machine No. 3.

[0084] First, based on the constructed production line model, the cycle time parameters, maximum working time, current idle window, and resource occupancy status of each piece of equipment are initialized. For example, in the current shift, high-speed laser engraving machine No. 2 has a 2.8-hour idle window, with an engraving cycle time of 11.9 seconds per panel; automatic film applicator No. 1 has 2.5 hours of available time, with a film applicator time of 7.5 seconds per unit; and intelligent precision cutting machine No. 3, due to its lower load, has a 4.0-hour idle period, with a cutting cycle time of 9.8 seconds. The system inputs the geometric structure information of each panel combination (such as the total length of the laser trajectory, the film applicator area, and the cutting path length) into the scheduling simulation module, dynamically calculates the processing time of each combination on different machines based on the above parameters, and selects the optimal path according to the equipment status.

[0085] A production line model is a structured data computation model built on the concept of digital twins. It provides a high-fidelity virtual environment for scheduling simulation by digitally abstracting the physical production line of automotive lighting display backlight panels in multiple layers and dimensions. The core components of the production line model include: an equipment resource library, a process knowledge base, a real-time status monitoring layer, a capacity calculation engine, and a set of constraint rules. The equipment resource library fully defines the static capability boundaries of all processing units (such as laser engraving machines, laminating machines, and cutting machines), including inherent parameters such as equipment codes, supported processes, maximum processing area, standard operating cycle time, and rated working time. The process knowledge base systematically stores manufacturing specifications, clearly defining standardized process routes for various backlight panels, logical dependencies between processes, and matching rules and compatibility constraints between processes and equipment, fixtures, and materials. The real-time status monitoring layer... The status monitoring layer dynamically reflects the operational pulse of the production system, continuously collecting and updating information such as the current idle time window of each device (e.g., the availability of the laser engraving machine within the next 2 hours), the queue of tasks to be executed, and the real-time occupancy status of fixtures and materials; the capacity calculation engine performs time window calculation, cycle time balance analysis, and load balancing optimization based on logical rules and real-time status, and inserts dynamic buffer time into the scheduling plan to improve its robustness; the constraint rule set clarifies the hard constraints that the scheduling must comply with (such as equipment capacity limitations and resource uniqueness) and the soft constraints used for optimization (such as delivery deadline preferences and equipment utilization targets).

[0086] The model is based on a core computational foundation comprised of logical structure data and real-time status data. During scheduling simulation, the system inputs the geometric information of the panel assembly and the process requirements into the model. By matching the possibilities provided by the logical structure data with the executability provided by the real-time status data, it dynamically calculates processing time, selects the optimal equipment path, and verifies scheduling feasibility and allocates conflict-free timing sequences in virtual space. This allows for accurate prediction and optimization of the production plan's execution effect before it is put into actual production.

[0087] Taking a panel assembly A as an example, the system calculates that its laser engraving requires approximately 0.9 hours on high-speed laser engraving machine number two, film application needs to be completed on automatic film application machine number one, taking 0.4 hours, and the cutting task is assigned to intelligent precision cutting machine number three, taking approximately 0.6 hours. The system further checks whether there are scheduling conflicts between these three machines within the time window of the current shift and verifies the feasibility of the production scheduling results for different assemblies. During the simulation, the system dynamically evaluates the average equipment utilization rate (e.g., 88.2%), process load balance rate (e.g., 91.5%), and the probability of completion within the preset delivery deadline for assembly A during the processing cycle, i.e., the delivery achievement rate (e.g., 96.7%). Based on the above indicators, a comprehensive evaluation function is obtained, with equipment utilization rate, process load balance rate, and delivery achievement rate as the main variables, and weights set to 0.4, 0.3, and 0.3 respectively. Finally, the score for panel assembly A is calculated to be 91.07 points.

[0088] The system sorts all candidate panel combinations according to their scores, selects the highest-scoring combinations, and forms the final panel capacity balancing scheduling result set. During this process, the system rationally allocates equipment load to avoid overloading or idling of critical equipment; simultaneously, it achieves load balancing between processes to reduce the possibility of any particular process becoming a capacity bottleneck; and it combines delivery capability assessment to ensure that task combinations are practically executable. This embodiment shows that the integrated simulation and scoring mechanism not only improves the overall resource utilization of equipment but also achieves the optimization goals of controllable delivery time and balanced processes, providing a feasible, stable, and effective capacity balancing solution for multi-batch intelligent scheduling of automotive lighting display backlight panels.

[0089] Another embodiment of the present invention provides a panelization and capacity balancing scheduling system for automotive lighting display backlight panels. The system includes: a panelization coupling modeling module, a candidate combination screening module, a scheduling refinement generation module, and a management database.

[0090] The panel coupling modeling module and the candidate combination screening module are connected, and the candidate combination screening module and the scheduling evaluation generation module are connected. The panel coupling modeling module, the candidate combination screening module, and the scheduling evaluation generation module are all connected to the management database. The management database is used to store various parameters involved in the panel and capacity balancing scheduling system for automotive lighting display backlight panels.

[0091] The panel coupling modeling module is used to construct a panel coupling graph based on order demand data. Each backlight panel is used as a node in the panel coupling graph. Edges with mutual interference weights and constraint potential weights are established between any nodes that meet the panel constraints. The panel coupling graph is subjected to structural analysis processing to identify the set of nodes that meet the panel combination requirements. The set of nodes that meet the panel combination requirements is marked as a candidate panel combination, thereby obtaining a candidate panel set that reflects panel compatibility and scheduling feasibility.

[0092] The candidate combination screening module is used to perform interval-based coarse evaluation of the candidate panel set by combining the upper and lower bounds of geometric density and coupling cost. Based on the coarse evaluation results, a normal candidate panel set that can directly enter the regular selection channel and a backup recall panel set that is set up to prevent false rejection and requires key review are obtained. A capacity balance model is constructed, and the normal candidate panel set and the backup recall panel set are evaluated and processed by the capacity balance model to determine the final refined evaluation candidate panel set that enters the fine evaluation process.

[0093] The scheduling evaluation generation module is used to perform evaluation processing on each panel combination in the evaluation candidate panel set, perform simulated scheduling based on the production line model and real-time capacity data, and generate a panel capacity balance scheduling result that meets the target delivery rate and equipment utilization rate.

[0094] It should be understood that the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, or B alone. A and B can be singular or plural. Additionally, the character " / " in this article generally indicates an "or" relationship between the preceding and following related objects, but it can also represent an "and / or" relationship. Please refer to the context for a more accurate understanding.

[0095] In this invention, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of a single item or a plurality of items. For example, at least one of a, b, or c can represent: a, b, c, ab, ac, bc, or abc, where a, b, and c can be a single item or multiple items.

[0096] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. A method for scheduling panel layout and production capacity balancing for automotive headlight display backlight panels, characterized in that, The method includes: Step 1: Construct a panel coupling graph based on order demand data. Using each backlight panel as a node in the panel coupling graph, establish edges with mutual interference weights and constraint potential weights between any nodes that satisfy the panel constraints. Perform structural analysis on the panel coupling graph to identify the set of nodes that meet the panel combination requirements. Mark the set of nodes that meet the panel combination requirements as candidate panel combinations to obtain a candidate panel set that reflects panel compatibility and scheduling feasibility. Step 2: Combine the upper and lower bounds of geometric density and coupling cost to perform interval-based coarse evaluation on the candidate panel set. Based on the coarse evaluation results, obtain the ordinary candidate panel set that can directly enter the regular selection channel and the minimum recall panel set that needs to be reviewed. Construct a capacity balance model and evaluate the ordinary candidate panel set and the minimum recall panel set through the capacity balance model to determine the final refined evaluation candidate panel set. Step 3: Perform detailed evaluation on each panel combination in the candidate panel set, and perform simulated scheduling based on the production line model and real-time capacity data to generate a panel capacity balance scheduling result that meets the target delivery rate and equipment utilization rate.

2. The method for panelization and production capacity balancing scheduling of automotive headlight display backlight panels according to claim 1, characterized in that, The process of establishing edges with mutual interference weights and constraint potential weights between any nodes that satisfy the panel constraints, using each backlight panel as a node in the panel coupling graph, is as follows: The node has backlight panel attributes, including the backlight panel's model code, geometric dimensions, quantity requirements, and delivery cycle. The panel constraints are used to evaluate whether there is physical and technological feasibility for combining and processing any nodes in the same panel process, specifically including geometric compatibility, fixture adaptability, and process consistency. Establish edges between any nodes that satisfy the puzzle constraint conditions, with mutual interference weights to characterize the degree of puzzle interference between nodes and constraint potential weights to characterize the ability of puzzle combinations to adjust under the resource and time dimensions.

3. The method for panelization and production capacity balancing scheduling of automotive headlight display backlight panels according to claim 1, characterized in that, The structural analysis and processing of the panel coupling diagram is performed as follows: Based on the relationship between nodes and edges in the panel coupling graph, substructure extraction is performed on the panel coupling graph to obtain a connected subgraph composed of nodes with connected edge relationships. The nodes of the connected subgraph represent backlight panels that are feasible for panel assembly. For each set of nodes in the connected subgraph, determine whether the set of nodes as a whole satisfies the requirements for panel combination; If the set of nodes in a certain connected subgraph satisfies the requirements for panel combination, then the backlight panel corresponding to the set of nodes is marked as a candidate panel combination, and all candidate panel combinations are organized into a candidate panel set. If the set of nodes in a certain connected subgraph does not meet the requirements for panel combination, the backlight panel corresponding to that set of nodes is marked as an invalid panel combination and is removed.

4. The method for panelization and production capacity balancing scheduling of automotive headlight display backlight panels according to claim 3, characterized in that, The process of identifying the set of nodes that meet the requirements for panel assembly is as follows: The mutual disturbance weight and constraint potential weight of each edge in the connected subgraph are compared with the mutual disturbance weight threshold and constraint potential weight threshold, respectively. The mutual interference weight threshold refers to the upper limit of the mutual interference weight within a specified range, and the constraint potential weight threshold refers to the lower limit of the constraint potential weight within a specified range. If the mutual interference weight on any side of a connected subgraph is greater than or equal to the mutual interference weight threshold, then the set of nodes in the connected subgraph as a whole does not meet the requirements for panel combination. If the constraint potential weight on any side of a connected subgraph is less than the constraint potential weight threshold, then the set of nodes in the connected subgraph as a whole does not meet the requirements for panel combination. If the mutual interference weight of any side in a connected subgraph is less than the mutual interference weight threshold, and the constraint potential weight is greater than or equal to the constraint potential weight threshold, then the set of nodes in the connected subgraph as a whole satisfies the panel combination requirement.

5. The method for panelization and production capacity balancing scheduling of automotive headlight display backlight panels according to claim 1, characterized in that, The candidate panel set is coarsely evaluated by combining the upper and lower bounds of geometric density and coupling cost. The specific scoring process is as follows: The upper bound of geometric density, which reflects the balance of the area utilization of each candidate panel combination, is obtained by using the ratio of the area of ​​the largest single backlight panel in each candidate panel combination to the total area of ​​the combination. By using the theoretically most compact layout area achievable by the minimum safety margin requirement in each candidate panel combination, we obtain the lower bound of geometric density to reflect the maximum possible area utilization efficiency of each candidate panel combination. By weighting the process switching complexity of each backlight panel in each candidate panel combination and the processing equipment occupation time, an upper bound of coupling cost is obtained to reflect the process switching cost of each candidate panel combination. By calculating the historical panel collinearity, path overlap rate and process compatibility index of the backlight panels in each candidate panel combination, a lower bound of coupling cost is obtained to reflect the minimum interference level of each candidate panel combination. The coarse upper bound value of each candidate panel combination is obtained based on the upper bound of geometric density and the lower bound of coupling cost. The coarse lower bound values ​​for each candidate panel combination are obtained based on the lower bound of geometric density and the upper bound of coupling cost.

6. The method for panelization and production capacity balancing scheduling of automotive headlight display backlight panels according to claim 1, characterized in that, The analysis process for obtaining the ordinary candidate panel set that can directly enter the regular selection channel and the backup recall panel set that requires in-depth review based on the preliminary evaluation results is as follows: All candidate puzzle pieces are sorted in descending order according to the upper bound of the coarse evaluation of each candidate puzzle piece combination. The candidate puzzle pieces whose ranking of the upper bound of the coarse evaluation is within the preset target number range are selected as the ordinary candidate puzzle piece set. For candidate puzzle combinations whose ranking is outside the preset target number range, if the upper limit of the coarse evaluation of a candidate puzzle combination is higher than the lowest value of the lower limit of the coarse evaluation in the ordinary candidate puzzle set, then the candidate puzzle combination is classified into the guaranteed recall puzzle set. The coarse evaluation uncertainty is obtained by using the upper bound and lower bound of the coarse evaluation of candidate puzzle combinations. For candidate puzzle pieces whose ranking is outside the preset target number range, if the coarse evaluation uncertainty of a candidate puzzle piece is higher than the coarse evaluation uncertainty threshold, then the candidate puzzle piece is classified into the minimum recall puzzle piece set.

7. The method for panelization and production capacity balancing scheduling of automotive headlight display backlight panels according to claim 1, characterized in that, The evaluation and processing of the ordinary candidate panel set and the guaranteed recall panel set using the capacity balance model is as follows: Using the ordinary candidate panel set and the guaranteed recall panel set as input, the capacity balance model is used to comprehensively evaluate each panel combination, output the refined candidate panel set, and then refine each panel combination in the refined candidate panel set. Simultaneously, dynamic self-optimization processing is performed on the preliminary evaluation stage based on the refined evaluation candidate panel set.

8. The method for panelization and production capacity balancing scheduling of automotive headlight display backlight panels according to claim 7, characterized in that, The dynamic self-optimization process based on the refined evaluation candidate panel set for the coarse evaluation stage is as follows: Based on the number of puzzle combinations in the guaranteed recall set of the refined candidate puzzle set, the guaranteed adoption rate is obtained; Perform linear regression on the guaranteed adoption rate for consecutive target periods and obtain the guaranteed adoption slope; If the guaranteed adoption slope is not lower than the guaranteed adoption slope threshold, then the uncertainty threshold in the coarse scoring stage is reduced to expand the range of potentially usable combinations to be retained. If the guaranteed adoption slope is lower than the guaranteed adoption slope threshold, the uncertainty threshold is increased to reduce the resource consumption of the guaranteed recall panel set.

9. The method for panelization and production capacity balancing scheduling of automotive headlight display backlight panels according to claim 1, characterized in that, The specific process of performing refined evaluation on each puzzle combination in the refined evaluation candidate puzzle set is as follows: Based on the production line model and real-time capacity data of the automotive lighting display backlight panel production line, a multi-dimensional scheduling simulation analysis is performed on each candidate panel combination in the refined evaluation candidate panel set. The production line model includes equipment capacity parameters, process cycle time parameters, resource constraint parameters, and workstation switching time parameters, which are used to characterize the dynamic capacity characteristics of the production line. Based on real-time capacity data, update the available time window and load status of each device, and perform scheduling feasibility verification and time sequence allocation for each candidate panel combination in the refined evaluation candidate panel set. Based on the simulation results of the production line model, the equipment utilization rate, process balance rate and delivery achievement rate of each panel combination within the processing cycle are calculated, and a comprehensive evaluation function is constructed accordingly. Based on the scoring results of the comprehensive evaluation function, the optimal or second-best panel combinations under the constraints of target delivery rate and equipment utilization rate are selected to form the final panel capacity balance scheduling result.

10. A scheduling system for panelizing and balancing production capacity of automotive lighting display backlight panels, employing the scheduling method for panelizing and balancing production capacity of automotive lighting display backlight panels as described in any one of claims 1-9, characterized in that, The system includes: The panel coupling modeling module is used to construct a panel coupling graph based on order demand data. Each backlight panel is used as a node in the panel coupling graph. Edges with mutual interference weights and constraint potential weights are established between any nodes that meet the panel constraints. The panel coupling graph is subjected to structural analysis processing to identify the set of nodes that meet the panel combination requirements. The set of nodes that meet the panel combination requirements is marked as a candidate panel combination, thereby obtaining a candidate panel set that reflects panel compatibility and scheduling feasibility. The candidate combination screening module is used to perform interval-based coarse evaluation of the candidate panel set by combining the upper and lower bounds of geometric density and coupling cost. Based on the coarse evaluation results, ordinary candidate panel sets that can directly enter the regular selection channel and minimum recall panel sets that need to be reviewed are obtained. A capacity balance model is constructed, and the ordinary candidate panel set and minimum recall panel set are evaluated and processed by the capacity balance model to determine the final refined evaluation candidate panel set that enters the refined evaluation process. The scheduling evaluation generation module is used to perform evaluation processing on each panel combination in the evaluation candidate panel set, perform simulated scheduling based on the production line model and real-time capacity data, and generate a panel capacity balance scheduling result that meets the target delivery rate and equipment utilization rate.

Images