A method for intelligent scheduling and path optimization of steel coils in a cold rolling mill

By generating candidate coil sequences and mapping them to a spatiotemporal path network, identifying bottleneck constraints and feeding them back to the candidate coil sequences for coordinated correction, the problem of disconnect between production process cycle time and warehouse path occupancy in cold rolling mill coil scheduling is solved, and the continuity and consistency of coil scheduling instructions are achieved.

CN123117645APending Publication Date: 2026-11-17YANGZHOU HENGRUN OCEAN HEAVY IND CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611307415.2
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-08-27
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

In the scheduling of steel coils in cold rolling mills, there is a lack of two-way closed-loop constraints between the production process cycle time and the storage area path occupancy. This results in the production scheduling results and path execution results not being able to correct each other within the same control cycle. This leads to tasks that can theoretically enter the line but cannot be reached on time by the actual path. Furthermore, when the path encounters a bottleneck section, it can only wait locally or temporarily reassign the task, and cannot be fed back to the production sequence for correction.

Method used

By acquiring the process status, storage location status, downstream unit process control cycle time, and storage area path occupancy status of steel coils in the cold rolling mill, candidate steel coil sequences are generated and mapped to a spatiotemporal path network. Bottleneck constraints are identified and fed back to the candidate steel coil sequences for linkage correction, generating scheduling control instructions that include the steel coil execution order, path node occupancy time period, and unit inlet buffer order.

Benefits of technology

It achieves the same scheduling process for production cycle time and warehouse path occupancy status, can identify bottleneck sections and conflict time slots and make corrections before execution, reduce the disconnect between production queues and warehouse execution routes, maintain the same control constraints between unit entrance and actual reachable path in warehouse, and avoid repeated changes to locked steel coils and steel coils that have entered the unit entrance.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123117645A_ABST
    Figure CN123117645A_ABST
Patent Text Reader

Abstract

This invention relates to the field of process control technology, specifically to an intelligent scheduling and path optimization method for steel coils in a cold rolling mill. The method acquires the process status, storage location status, downstream unit process control cycle time, and storage area path occupancy status of the steel coils. Based on the process status and unit cycle time, it generates candidate steel coil sequences and allowable arrival windows. The candidate sequences are mapped to a spatiotemporal path network composed of storage location nodes, path segment nodes, and unit entrance nodes. The method deduces the node occupancy periods for each steel coil at different outbound times, identifies bottleneck constraints formed by channel overlap, entrance queuing, and stacking issues, and feeds these bottleneck constraints back to the candidate steel coil sequences. The method then performs linked corrections on the steel coil release order, outbound time, arrival window, and unit entrance buffer order, generating scheduling control commands. This invention enables process cycle time constraints and path occupancy constraints to mutually correct each other within the same control cycle, reducing execution deviations caused by feasible scheduling but unreachable paths.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of process control technology, specifically to a method for intelligent scheduling and path optimization of steel coils in a cold rolling mill. Background Technology

[0002] In cold-rolled mills, steel coils continuously flow between processes such as pickling, rolling, annealing, leveling, rewinding, and packaging. The scheduling objects typically possess attributes such as steel grade, specifications, surface finish, upstream completion time, downstream unit entry time, and warehouse location. Existing process control methods generally rely on production planning. The planning system generates steel coil processing queues based on the production cycle time of downstream units, specification transition sequence, and delivery sequence, and then distributes these queues to the warehouse execution side. The warehouse execution side queries the current warehouse location based on the steel coil number, and, combined with warehouse access routes, hoisting equipment status, and unit entry buffer status, determines the steel coil outbound sequence and transport route. The basic characteristic of this approach is that the production sequence is determined first, followed by warehouse path matching. Path calculation is usually based on the shortest distance, the fewest transfers, or the current availability of access routes. Since production queues and path execution are two relatively independent control links, the production side focuses more on process continuity, while the transport side focuses more on whether a single steel coil can reach the target unit. There is a lack of joint constraint expression between the two based on the same time coordinate.

[0003] In conventional technical solutions, steel coil path optimization often employs warehouse node diagrams or channel network diagrams. The system abstracts warehouse locations, channel intersections, cross-zone connection points, and unit entrances as path nodes, and traversable relationships as path segments. When a steel coil needs to be retrieved, the system searches for traversable paths based on the coil's current location and the target unit entrance location, selecting the path with the shortest distance, least waiting time, or fewest transfers, considering current task occupancy. For obstructed steel coils, the system typically arranges for repacking, moving the obstructed coil to a temporary warehouse location before retrieving the target coil. For entrance queuing or channel conflicts, the system usually handles them by delaying retrieval, temporary waiting, re-finding paths, or manually adjusting the task order. This type of solution can handle the handling of a single steel coil from its warehouse location to the unit entrance, but path calculations are often performed after the production queue has been determined. The allowable arrival window formed by the process cycle time is not fully converted into time occupancy constraints for path nodes, and channel congestion, repacking chains, and entrance queuing generated on the path side are difficult to promptly constrain the production queue.

[0004] Another common approach is to introduce a rolling scheduling mechanism into the scheduling system. The system reads the unit status, warehouse status, and incomplete task status at fixed intervals, recalculating the release time and handling sequence for coils that have not yet been executed. This method typically locks tasks that have already started executing and reorders those that haven't, mitigating the impact of on-site deviations on subsequent tasks. Some systems also set priorities based on the risk of material shortages at the unit inlet, coil waiting time, or warehouse congestion, adjusting the coil release order through priority ranking. While this approach can alleviate local anomalies, its reordering is still primarily based on the production plan or current execution status, lacking a unified analysis of the coupling relationship between path segments, warehouse obstructions, inlet buffers, and process windows. When channel conflicts arise from multiple coils sharing the same path segment in adjacent time periods, simply increasing the priority of one coil may shift the conflict to subsequent coils; when the inlet buffer capacity is limited, simply releasing coils earlier may cause channel waiting or warehouse congestion.

[0005] The main technical problem with existing technologies is the lack of a two-way closed-loop constraint between production cycle time and warehouse path occupancy in cold rolling mill coil scheduling. This results in the inability of scheduling results and path execution results to be mutually corrected within the same control cycle. Although the coil sequence generated on the production side meets the continuous processing requirements of downstream units, it fails to simultaneously consider the location of the coil in the warehouse, obstruction relationships, channel occupancy, and entrance buffer positions. This easily leads to tasks that are theoretically possible to enter the line but cannot arrive on time on the actual path. Although the path side can find passable routes for specified coils, when encountering bottleneck sections, stacking chains, or entrance queues, it can usually only wait locally or temporarily reassign them, and cannot feed back bottleneck sections, conflict time slots, and affected coil sets to the production sequence for local correction. As a result, scheduling control commands repeatedly deviate from the actual execution status, making it difficult to maintain consistency between the unit entrance buffer sequence, coil exit time, and path node occupancy periods. Summary of the Invention

[0006] The purpose of this invention is to provide an intelligent scheduling and path optimization method for steel coils in cold rolling mills, which can solve the problems mentioned in the background art.

[0007] To achieve the above objectives, the technical solution adopted by the present invention is as follows: A method for intelligent scheduling and path optimization of steel coils in a cold rolling mill includes: acquiring the process status, storage location status, downstream unit process control cycle time, and storage area path occupancy status of steel coils in the cold rolling mill; generating candidate steel coil sequences and corresponding allowed arrival windows based on the process status and the downstream unit process control cycle time; mapping the candidate steel coil sequences to a spatiotemporal path network composed of storage location nodes, path segment nodes, and unit entrance nodes, and deducing the node occupancy periods of each steel coil at different outbound times; identifying bottleneck constraints formed by channel overlap, entrance queuing, and stacking association based on the node occupancy periods; feeding back the bottleneck constraints to the candidate steel coil sequences, and performing linked corrections on the steel coil release order, outbound time, and arrival window, generating scheduling control instructions that include the steel coil execution order, path node occupancy periods, and unit entrance buffer order.

[0008] Preferably, generating a candidate coil sequence and corresponding allowed arrival windows based on the process status and the downstream unit process control cycle time includes: establishing a status record for each coil, the status record including at least the current storage location, target process, upstream completion time, downstream unit cycle time, specification connection attribute, inventory obstruction level, and remaining waiting tolerance; forming a candidate coil set according to constraints such as the same target process, continuous specification connection, and upstream completion time meeting the entry conditions; determining the allowed arrival window of the coils in the candidate coil set according to the downstream unit cycle time and inlet buffer capacity, and retaining multiple replaceable candidate sorting positions within the allowed arrival window.

[0009] Preferably, mapping the candidate steel coil sequence to a spatiotemporal path network includes: defining storage locations, storage area passage intersections, cross-area connection points, and unit entrances as path nodes with time occupancy attributes; defining the passability relationship between adjacent path nodes as path segments; forming continuous occupancy segments for each candidate steel coil according to the outbound time, handling duration, stacking occupancy time, and entrance waiting time; and writing the continuous occupancy segments into the corresponding path nodes and path segments, so that the process arrival window of the same steel coil is associated with the path occupancy period in the same time coordinate.

[0010] Preferably, identifying bottleneck constraints based on the node occupancy time period includes: comparing the occupancy segments of different steel coils on the same path segment, the same storage channel, and the same unit entrance buffer position in the same time coordinate; when the occupancy segments overlap, and the steel coils corresponding to the overlap cannot fall into their respective allowed arrival windows by adjusting the outbound time, the overlapping position is recorded as a bottleneck segment; steel coils that have a path succession relationship, stacking obstruction relationship, or entrance queuing relationship with the bottleneck segment are classified into the affected steel coil set, and the bottleneck segment, conflict time segment, and affected steel coil set are combined into the bottleneck constraint.

[0011] Preferably, determining the allowable arrival window for steel coils within the candidate steel coil set includes: using the downstream unit process control cycle time as a benchmark, calculating the earliest and latest arrival times for each candidate sorting position based on the unit inlet buffer capacity; converting the upstream completion time, current warehouse release time, and inventory obstruction level of the steel coil into an outbound time boundary; overlapping the outbound time boundary with the earliest and latest arrival times to form an effective arrival window for the steel coil to participate in sorting; when the same sorting position corresponds to multiple effective arrival windows, retaining steel coils with consistent process connection attributes and comparable path occupancy differences as candidate replacement objects.

[0012] Preferably, forming the continuous occupancy segment includes: generating a candidate path chain according to the passable path from the current storage location of the steel coil to the entrance of the target unit; assigning start and end times for occupancy to storage location release, passage entry, intersection point passage, cross-zone connection, and entrance buffer in the candidate path chain; establishing a successor association between the temporary occupancy segment of the steel coil to be moved out and the outbound occupancy segment of the target steel coil for the purpose of repacking steel coils; setting mutually exclusive occupancy markers for path segments that cannot be passed in parallel within the same candidate path chain, and writing the mutually exclusive occupancy markers into the spatiotemporal path network.

[0013] Preferably, feeding back the bottleneck constraint to the candidate coil sequence includes: extracting coils that are still not released and whose process connection attributes meet the requirements of adjacent sorting positions, based on the sorting position of the affected coil set in the candidate coil sequence; performing a partial reordering on the extracted coils, with the sorting of locked coils and coils that have entered the unit inlet as the boundary; recalculating the effective arrival window and path occupancy segment for the reordered coils; when there is still overlap in the occupancy of the same bottleneck segment after the partial reordering, including the preceding path segment connected to the bottleneck segment in the feedback range and redefining the coil release time.

[0014] Preferably, the redefined release time of the steel coil includes: based on the conflict time slice of the bottleneck section, tracing upstream along the path to the first storage location node or channel waiting node that can wait without obstructing the release of other candidate steel coils; setting the traced storage location node or channel waiting node as the release correction node; calculating the release delay or advance based on the path occupancy segment of the steel coil from the release correction node to the target unit entrance; and simultaneously writing the release delay or advance into the steel coil's release time, effective arrival window, and unit entrance buffer sequence, so that the reorder adjustment and path occupancy adjustment are completed within the same control cycle.

[0015] Preferably, mutual exclusion occupancy markers are set for path segments that cannot be passed in parallel within the same candidate path chain, including: establishing a mutual exclusion set according to the one-way passage relationship, intersection point sharing relationship, and entrance buffer continuation relationship of the path segments; when the continuous occupied segments of the target steel coil have time overlap with the registered occupied segments in the mutual exclusion set, the overlapping segments are split into waitable segments and non-waitable segments; the original candidate path chain is retained for the waitable segments and the start and end times of the corresponding node occupancy are adjusted; a detour candidate path chain is generated for the non-waitable segments, and the occupied segments of the detour candidate path chain are synchronously included in the subsequent bottleneck constraint identification.

[0016] Preferably, after simultaneously writing the release delay or advance amount into the steel coil's outbound time, effective arrival window, and unit inlet buffer sequence, the method further includes: performing a consistency check on the corrected steel coil sequence before issuing the scheduling control command; the consistency check includes comparing whether the outbound time of the same steel coil falls within the available outbound time boundary, whether the path node occupancy period avoids the mutually exclusive occupancy mark, and whether the unit inlet buffer sequence is within the effective arrival window; when any check item is not satisfied, the corresponding steel coil and its adjacent sorting positions are re-added to the affected steel coil set, and the scheduling control command is regenerated according to the local reordering method defined in claim 7.

[0017] Compared with the prior art, the beneficial effects of the present invention are as follows: 1. This invention incorporates the coil process status, warehouse location status, downstream unit process control cycle time, and warehouse path occupancy status into a single scheduling process. It then performs linked corrections based on candidate coil sequences, allowable arrival windows, spatiotemporal path networks, and bottleneck constraints. This transforms the time requirements for production process cycle time into path node occupancy periods, and converts channel overlap, entrance queuing, and stacking correlations derived from path deduction into correction criteria for candidate coil sequences. The coil release sequence, outbound time, path node occupancy periods, and unit entrance buffer sequence are generated by the same closed-loop process, reducing the disconnect between the production-side queue and the warehouse execution-side route. For coils where scheduling is feasible but the path is blocked, the scheduling process can identify bottleneck sections and conflicting time slices before execution is issued, and correct the sorting and arrival windows within the unreleased coil range, ensuring that the coils required for unit entrance and the actual reachable path in the warehouse maintain the same control constraints.

[0018] 2. This invention, by establishing status records for candidate coil sets, creating time occupancy attributes for storage nodes and path segments, setting mutually exclusive occupancy markers for path segments that cannot be passed in parallel, and performing consistency checks on outbound time, effective arrival window, and inlet buffer order after local reordering, can narrow the scope of readjustment tasks and avoid repeated modifications to locked coils and coils already in the unit inlet. For coils involving stacking, establishing a successor-successor association between temporary occupancy segments and target coil outbound occupancy segments allows stacking, channel waiting, and inlet buffering to be identified and corrected in the same time coordinate. Generating detour candidate path chains for non-waitable conflict segments and incorporating them into subsequent bottleneck constraint identification can reduce new conflicts caused by local detours, ensuring the continuity of coil scheduling instructions during rolling updates. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the overall closed-loop control process of the intelligent scheduling and path optimization method for cold rolling mill coils according to the present invention. Figure 2 This is a flowchart illustrating the generation process of the candidate steel coil sequence and the allowed arrival window in this invention. Figure 3 This is a flowchart of the spatiotemporal path network construction and bottleneck constraint identification process of the present invention; Figure 4 This is a flowchart of the bottleneck constraint feedback, local reordering, release correction and consistency verification of the present invention. Detailed Implementation

[0020] refer to Figure 1 In one embodiment, the intelligent scheduling and path optimization method for cold-rolled steel coils is applied to the process control scenario of steel coils flowing from the storage area to the downstream unit inlet. The controlled objects are the steel coils to be shipped out, the storage location, the path segment, and the buffer position at the unit inlet. The method acquires the process status, storage location status, downstream unit process control cycle time, and storage area path occupancy status of the steel coils within the same control cycle. The process status includes the current process of the steel coil, the target process, the upstream completion time, whether the conditions for entering the target process are met, and the specification connection relationship with adjacent steel coils. The storage location status includes the storage location of the steel coil, the storage location, and the storage location status. Whether the coil is occupied, whether the coil is blocked by other coils, and the accessibility relationship from the storage location to the adjacent channel, the downstream unit process control cycle includes the continuous time position of the target unit that can receive the coil within the current control cycle, the occupancy order of the inlet buffer position, and the allowable arrival interval between adjacent coils. The storage area path occupancy status includes the occupancy status of storage location nodes, channel intersection nodes, cross-area connection nodes, path segments, and unit inlet nodes in each time slice. As a result, the coil sequence on the production side and the path execution on the storage area side are no longer processed separately, but are formed into a set of rectifiable scheduling objects in the same data view.

[0021] In this embodiment, the generation of candidate coil sequences is based on process status and downstream unit process control cycle time. The method first screens coils whose target process is consistent with the downstream unit and whose upstream completion time meets the entry conditions. Then, it generates candidate coil sequences based on specification connection relationship, unit inlet buffer capacity and remaining waiting tolerance. Each coil in the candidate coil sequence is assigned an allowed arrival window. The allowed arrival window is not simply determined by the production plan time, but is jointly limited by the unit receiving time position, inlet buffer idle time and the time when the coil meets the conditions for leaving the warehouse. When the candidate coil sequence is generated, several replaceable coils under the same process position are retained, so that subsequent path deduction can locally correct the coil release order without disrupting the continuous process control of the unit.

[0022] In this embodiment, the storage area is represented as a spatiotemporal path network. Storage location nodes, storage area channel intersections, cross-area connection points, and unit entrance nodes all have time occupancy attributes. The passability relationship between adjacent nodes is represented as path segments. When the method maps candidate steel coil sequences to the spatiotemporal path network, it takes the current storage location of the steel coil as the starting point and the downstream unit entrance as the ending point to deduce the path node occupancy time period and path segment occupancy time period for the same steel coil at multiple feasible outbound times. The occupancy time period includes the time when the steel coil is released from the storage location, the time when it enters the channel, the time when it passes through the intersection node, the time when it crosses the cross-area connection, the time when it enters the entrance buffer position, and the temporary occupancy time caused by stacking. This allows the allowable arrival window of the steel coil and the actual reachability state of the storage area path to be compared in the same time coordinate.

[0023] In this embodiment, the bottleneck constraint is obtained by comparing the time periods occupied by nodes. The method checks whether there is overlap in the occupancy of different steel coils on the same path segment, the same storage channel, and the same unit entrance buffer position in the same time coordinate. If the overlap can be eliminated by adjusting the outbound time within the allowed arrival window, the corresponding adjustment is written into the candidate steel coil sequence. If the overlap cannot be eliminated by adjusting the outbound time, the overlapping position is recorded as the bottleneck segment, the time of the overlap is recorded as the conflict time slice, and the steel coils that have a path succession relationship, stacking block relationship, or entrance queuing relationship with the bottleneck segment are included in the affected steel coil set. The bottleneck segment, conflict time slice, and affected steel coil set together constitute the bottleneck constraint fed back to the candidate steel coil sequence.

[0024] In this embodiment, after the bottleneck constraint is fed back to the candidate coil sequence, the method does not reorder all coils. Instead, it performs a linkage correction within the affected coil set and the coils that are still in the unreleased state. The linkage correction includes partially reordering the coil release order, advancing or delaying the coil outbound time, shrinking or shifting the allowed arrival window, rearranging the unit inlet buffer order, and rewriting the path node occupancy time. The coils that have entered the execution stage and the coils that have entered the unit inlet maintain their original order. The regenerated scheduling control instruction includes the coil execution order, the path node occupancy time, and the unit inlet buffer order, so that the process cycle time, outbound time, and path occupancy together form a closed-loop control result.

[0025] In this embodiment, the following comprehensive occupancy conflict value can be used to quantitatively deduce the occupancy status of steel coils and paths in the candidate steel coil sequence:

[0026] in, Indicates the first The overall occupancy conflict value of each candidate steel coil. Indicates the first The set of path nodes and path segments traversed by each candidate steel coil. This indicates other steel coil sets that have been registered for occupation or are being simulated simultaneously. Indicates the first One candidate steel coil in the path object At the start of the occupancy, Indicates the first One candidate steel coil in the path object When the occupancy ends, and These represent the start and end times of occupancy for other steel coils on the same path object, respectively. Indicates the first The queue length formed when candidate steel coils enter the unit inlet Indicates the first The number of stacking associations corresponding to each candidate steel coil. Indicates the entry queuing weight. Indicates the weight of the stacking association. Used to retrieve larger values. Used to take the smaller value, if the first One candidate steel coil occupies a certain path segment from minute 40 to minute 46, and another steel coil occupies the same path segment from minute 43 to minute 49. Therefore, their overlapping occupancy in that path segment is 3 minutes, from minute 43 to minute 46. 2 and =1, Take 2, If we choose 3, then the conflict contribution of this segment, along with the entry queuing and stacking association, will participate in the sorting correction.

[0027] In this embodiment, the scheduling control command is not directly fixed after it is generated. Instead, a consistency check is performed before the end of the current control cycle. The check includes whether the outbound time of the same steel coil falls within its available outbound time boundary, whether the path node occupancy period avoids the mutual exclusion occupancy mark, and whether the unit inlet buffer sequence is within the allowed arrival window. If any inconsistency exists, the corresponding steel coil and its adjacent steel coils in the candidate steel coil sequence are re-incorporated into the affected steel coil set, and bottleneck constraint feedback and local correction are continued until the steel coil execution sequence, path node occupancy period, and unit inlet buffer sequence in the scheduling control command are consistent with each other. This embodiment makes the theoretically accessible tasks executable at the path level through bidirectional correction of process cycle time and path time and space occupancy, and makes the path bottleneck obtain a correctable position at the scheduling level.

[0028] refer to Figure 2 Preferably, to provide a feasible data foundation for the candidate coil sequence, a status record is established for each coil. The status record includes at least the current storage location, target process, upstream completion time, downstream unit cycle time, specification connection attribute, inventory obstruction level, and remaining waiting tolerance. The current storage location is used to determine the starting node of the coil in the storage area network. The target process is used to determine the downstream unit or process entry point into which the coil should enter. The upstream completion time is used to determine whether the coil can enter the next process. The downstream unit cycle time is used to determine the time position when the unit continuously receives coils. The specification connection attribute is used to determine whether the coil can be continuous with the coils before and after it in the candidate coil sequence. The inventory obstruction level is used to indicate the obstruction relationship that needs to be handled before the coil leaves the storage from the current location. The remaining waiting tolerance is used to limit the range in which the coil can be delayed or advanced within the current control cycle.

[0029] In this embodiment, the candidate coil set is formed according to the constraints of having the same target process, continuous specification connection, and upstream completion time meeting the entry conditions. Specifically, when multiple coils have the same target process and can all enter the same downstream unit, the method compares whether adjacent coils meet the process connection rules of the unit in terms of thickness, width, steel type, or surface grade, and excludes coils whose upstream completion time is later than the unit's acceptable time and cannot pass through the waiting entry window. The remaining coils enter the candidate coil set. The coils in the candidate coil set are not immediately fixed in a unique sorting position, but correspond to multiple alternative sorting positions within the range allowed by the unit's inlet buffer capacity, providing alternative objects for subsequent path bottleneck feedback.

[0030] Preferably, the allowable arrival window is determined by the downstream unit cycle time and the inlet buffer capacity. The method uses the steel coil currently being processed or about to be processed by the unit as the time reference, and forms several candidate sorting positions according to the time position of the unit continuously receiving steel coils. For each candidate sorting position, the earliest arrival time and the latest arrival time are calculated. The earliest arrival time is limited by the idle time of the inlet buffer, and the latest arrival time is limited by the continuous production continuity time of the unit. If the time boundary for the steel coil to be released from the warehouse formed by the upstream completion time, warehouse release time and obstruction handling time cannot overlap with the earliest arrival time and the latest arrival time of the candidate sorting position, then the steel coil will not participate in the competition for that sorting position, thus avoiding invalid calculations in subsequent path deductions on unreachable times.

[0031] In this embodiment, the overlap between the available outbound time boundary and the candidate ranking position is calculated in the following way:

[0032] in, Indicates the first The steel coil participated in the first The effective arrival window is formed when there are candidate sorting positions. Indicates the first The upstream completion time of a steel coil Indicates the first The current release time of each steel coil from its storage location. Indicates the first The earliest arrival time corresponding to each candidate sorting position. Indicates the first The latest time a steel coil can be allowed to wait. Indicates the first The latest arrival time corresponding to each candidate sorting position, if Then the first The steel coil can participate in the first If the nth candidate sorting position, If the upstream completion time of a steel coil is 18 minutes, the warehouse release time is 22 minutes, the earliest arrival time of the candidate sorting position is 25 minutes, the latest time the steel coil is allowed to wait is 50 minutes, and the latest arrival time of the candidate sorting position is 42 minutes, then the effective arrival window is from 25 minutes to 42 minutes.

[0033] Furthermore, when the same candidate sorting position corresponds to multiple valid arrival windows, the method retains steel coils with consistent process connection attributes and comparable path occupancy differences as candidate replacement objects. The comparable path occupancy differences mean that the corresponding steel coils can all be mapped to the same spatiotemporal path network, and comparable outbound time, path node occupancy segment, and inlet buffer entry time can be obtained. If a steel coil is technologically replaceable but its storage location information is incomplete or its path cannot be mapped, it will not be used as a replacement object for that sorting position. If the process connection attributes of a steel coil are inconsistent with those of the steel coils adjacent to the sorting position, even if its path is shorter, it will not enter the replacement set for that sorting position. This embodiment uses status recording, process screening, and valid arrival window overlap processing to ensure that the candidate steel coil sequence simultaneously meets the requirements of production process control and path deduction input.

[0034] In a preferred embodiment, the steel coil status record can be organized according to the data structure shown in Table 1. The fields in the table are used to support the generation of candidate steel coil sets, calculation of effective arrival windows, path occupancy inference, and bottleneck constraint feedback. Each field remains consistent according to the steel coil number within the same control cycle, and can be refreshed according to upstream completion, warehouse location change, and inlet buffer change in different control cycles.

[0035] Table 1. Data Structure of Steel Coil Scheduling Status Field Name Data Meaning Scheduling and usage location Constraint handling methods Current storage location Starting position of steel coils in the warehouse network Path chain generation Convert to starting path node Target process Downstream processing stages that steel coils need to enter Candidate set filtering Matching the unit inlet Upstream completion time The steel coil is ready to proceed to the next process. Effective arrival window calculation As the lower bound of time Downstream unit cycle time Time relationship of continuous receiving of steel coils by the target unit Candidate sorting position determined Formation unit arrival window Specification connection attributes Compatibility with continuous production of adjacent steel coils Candidate steel coil sequence generation Determine the range of alternatives Inventory occlusion levels Obstruction relationships that need to be addressed before shipment Reverse stacking occupancy fragment generation Forming successor-successor relationships Remaining waiting tolerance The time range during which steel coils can be scheduled earlier or later Local ordering and window correction Limit adjustment boundaries refer to Figure 3 In this embodiment, the spatiotemporal path network is constructed with storage locations, storage area passage intersections, cross-area connection points, and unit entrances as path nodes, and the passable relationship between adjacent path nodes as path segments. Each path node and path segment records a time occupancy table. The time occupancy table is within the current control cycle and records the start and end times of occupancy for locked tasks, currently executing tasks, and candidate simulation tasks. A path node represents the occupancy of a steel coil in a certain spatial location, and a path segment represents the movement occupancy of a steel coil between two adjacent spatial locations. The unit entrance node also includes a buffer position order, so that path occupancy and entrance queuing can be simulated in the same network.

[0036] Preferably, for each candidate steel coil, a continuous occupancy segment is formed according to the time of departure, the duration of handling, the time of stacking, and the waiting time at the entrance. The method generates a candidate path chain based on the passable path between the current storage location and the entrance of the target unit. The candidate path chain is arranged in the order of nodes, and the start and end times of storage location release, passage entry, intersection point passage, cross-zone connection, and entrance buffer are written into the corresponding node or segment. If there are multiple candidate path chains for the same steel coil, a set of continuous occupancy segments is generated for each path chain. Subsequently, it is selected whether to retain the segment based on the occupancy conflict value and the allowed arrival window. The continuous occupancy segments make the departure, movement, waiting, and entry of the same steel coil form a traceable time sequence.

[0037] Furthermore, the steel coils involved in the repacking are linked in a successive manner according to the occlusion relationship between the target steel coil and the steel coil being removed. Specifically, when the target steel coil is located deep in the storage area or inside the passageway and is occupied by other steel coils, the method first writes the temporary movement process of the steel coil to be removed into a temporary occupancy segment, and then writes the outbound occupancy segment of the target steel coil into the same candidate path chain. The end time of the temporary occupancy segment is earlier than the start time of the outbound occupancy segment of the target steel coil. The temporary storage node corresponding to the steel coil being removed is also written into the time occupancy table. If the steel coil being removed is simultaneously in other candidate steel coil paths, its temporary occupancy segment is used as the input for subsequent bottleneck constraint identification to avoid the repacking action being ignored in the path deduction.

[0038] In this embodiment, mutually exclusive occupancy markers are set for path segments that cannot be passed in parallel. The mutually exclusive occupancy markers are derived from one-way passage relationships, intersection point sharing relationships, and entrance buffer continuation relationships. When two steel coils have time overlap in the same narrow section of the channel, the same cross-area connection point, or the same entrance buffer position, the method does not allow them to register for occupancy at the same time, and marks the corresponding cross-time slice as mutually exclusive conflict. If a steel coil can wait at the upstream waiting node and still fall into the valid arrival window after waiting, the original candidate path chain is retained and the start and end times of subsequent occupancy are shifted. If it cannot fall into the valid arrival window after waiting, a detour candidate path chain is generated, and the node occupancy segment of the detour chain is written into the spatiotemporal path network.

[0039] In this embodiment, consecutively occupied segments can be formed using the following recursive method:

[0040] in, Indicates the first The first steel coil enters the candidate path chain. At the moment of a path object, Indicates the first The steel coil left the first At the moment of a path object, Indicates the first The steel coil occupies the first The duration of a path object This indicates the additional waiting time caused by mutual exclusion waiting, stacking pre-positioning, or entry buffering. If a steel coil enters the first path object at the 30th minute and occupies the space for 4 minutes, and the additional waiting time caused by mutual exclusion waiting ahead is 2 minutes, then the time of leaving the path object is the 36th minute. The entry time of subsequent path objects is recursively calculated based on the 36th minute.

[0041] Preferably, when forming candidate path chains, the method does not use a single shortest path as the only result. Instead, it performs a constraint search on the traversability relationship between the current storage location and the target unit entrance, retaining several candidate path chains that can meet the valid arrival window and do not form an unwaitable conflict with the locked task. The candidate path chains differ in the order of path nodes, the number of mutually exclusive segments, the entrance buffer continuation relationship, and the successor relationship before and after stacking. Subsequently, the corresponding path chain is selected or modified based on the bottleneck constraint feedback results. If a candidate path chain is short in path length but forms an unwaitable conflict at the entrance buffer position, it is not used first. If another candidate path chain requires detour but can enter the unit entrance within the allowed arrival window, the path chain can be retained to participate in subsequent consistency verification.

[0042] In this embodiment, bottleneck constraint identification is based on the comparison of occupied segments in the same time coordinate. The method compares the occupied segments of different steel coils in the same path segment, the same storage channel, and the same unit entrance buffer position, and determines whether cross-occupancy can be eliminated by adjusting the outbound time or waiting upstream. If the cross-occupancy occurs before the waiting node and is still within the effective arrival window after waiting, the cross-occupancy is converted into waiting correction. If the cross-occupancy occurs in the unwaitable segment or exceeds the effective arrival window after waiting, a bottleneck segment is formed. The bottleneck segment, along with the conflict time slice and the set of affected steel coils, is fed back to the candidate steel coil sequence.

[0043] Furthermore, the determination of the affected steel coil set is not limited to the two steel coils with overlapping occupancy. The method extends along the path succession relationship, stacking blockage relationship, and entrance queuing relationship to include steel coils connected to the bottleneck section and whose occupancy period may change. For example, when the path section of the target steel coil overlaps with another steel coil, if there is a previous stacking steel coil before the target steel coil, then the previous stacking steel coil is also included in the affected steel coil set. If the entry time of subsequent steel coils in the same unit's entrance buffer sequence depends on the entry time of the target steel coil, then the subsequent steel coils are also included in the affected steel coil set. The affected steel coil set thus covers the local scheduling range that will actually change with the bottleneck.

[0044] In this embodiment, the bottleneck constraint can be written into the affected set of steel coils according to the following determination formula:

[0045] in, Represents a path object The corresponding bottleneck steel coil set, Indicates the first A steel coil in the path object The time period occupied on Indicates the first A steel coil in the path object The time period occupied on This represents the intersection of two time periods. To represent the empty set, Indicates the first The steel coil and the first A steel coil in the path object The duration of the non-eliminable crossover. Represents path objects A set of candidate steel coils that have successor, occlusion, or entry continuity relationships. If two steel coils are on the path object... The occupancy periods are from the 50th to the 58th minute and from the 55th to the 62nd minute, respectively. The intersection is from the 55th to the 58th minute, and this period cannot be eliminated by waiting. Therefore, both steel coils enter... .

[0046] refer to Figure 4 Preferably, when feeding back bottleneck constraints to the candidate coil sequence, the method first locates the sorting position of the affected coil set in the candidate coil sequence, and then extracts the coils that are still in the unreleased state and whose process connection attributes meet the requirements of adjacent sorting positions. Locked coils, coils in execution, and coils that have entered the unit inlet do not participate in local reordering. Local reordering is only carried out within the scope of bottleneck influence. Candidate replacement objects must simultaneously meet the following requirements: consistent target process, continuous specification connection, overlapping effective arrival windows, and path occupancy segments that can be recalculated. After the reordering is completed, the effective arrival windows and path occupancy segments are recalculated to avoid introducing new process discontinuities due to changes in the production sequence.

[0047] In this embodiment, the local reordering is performed using a restricted insertion method. The method temporarily removes the steel coils that have not yet been released from the original candidate steel coil sequence in the affected steel coil set, and searches for an insertion position within the affected sorting interval. Each insertion position needs to check the specification connection attributes of the previous and next steel coils, the unit inlet buffer sequence, and the time boundary for delivery. If the effective arrival window formed after insertion is empty, the insertion position is excluded. If the path occupancy segment still has an irreversible overlap with the same bottleneck segment after insertion, the adjacent insertion position is tried or a candidate replacement object is used until a local sequence that meets the process connection and path occupancy requirements is formed.

[0048] Furthermore, when there is still overlap in the occupancy of the same bottleneck section after local reordering, the method includes the preceding path section connected to the bottleneck section in the feedback scope and redefines the release time of the steel coil. Specifically, the method traces upstream along the candidate path chain of the target steel coil to the first storage location node or channel waiting node that can wait without blocking the release of other candidate steel coils. This node is set as the release correction node. The release delay or advance is calculated based on the path occupancy segment between the release correction node and the target unit entrance. The release time of the steel coil, the effective arrival window, and the unit entrance buffer sequence are updated simultaneously to ensure that path waiting is not isolatedly attached to the middle section of the channel.

[0049] In this embodiment, the calculation of the release correction node and the release correction amount can be performed using the following expression:

[0050] in, Indicates the first Release correction node for each steel coil Indicates the first The set of storage location nodes or channel waiting nodes that can be waited for in a candidate path chain for steel coils. Indicates the first The steel coils arrived at the node. At that moment, Indicates the first The reference time for the window when a steel coil enters the target unit inlet. Indicates the first A steel coil at the node The cost of obstructing the release of other candidate steel coils while waiting. Indicates the occlusion cost weight. This represents the node that minimizes the subsequent expression. If the time deviations of a steel coil at two waitable nodes are 3 minutes and 1 minute, respectively, and the occlusion costs are 4 and 8, respectively. If we set it to 1, the calculated values ​​for the two nodes are 7 and 9 respectively. The node to be released for correction is the one with a time deviation of 3 minutes but a lower cost of occlusion.

[0051] In a preferred embodiment, the bottleneck constraint feedback process can be organized according to the control logic in Table 2. The contents listed in the table are used to explain the data source, feedback range and scheduling correction method corresponding to different bottleneck types. In actual execution, the same steel coil may be affected by multiple bottleneck types at the same time. In this case, the union of the affected steel coil sets is used as the local correction range.

[0052] Table 2. Comparison of Bottleneck Constraint Feedback Control Logic Bottleneck types Data source Feedback Scope Scheduling correction method Channel overlap Path segment occupied fragment intersection The same path segment and its preceding waiting nodes Adjust release timing or generate bypass path chain Queues at the entrance Unit inlet buffer position occupancy order Adjacent coils in the same inlet buffer queue Correct the ingress buffer order and arrival window Stacking association Temporary occupation of the section for shielding steel coils Target steel coil and removed steel coil Establish successor-successor occupancy relationship Warehouse backlog Storage location nodes and channel nodes occupy overlapping areas. Current storage location and adjacent outbound channels Select release correction node Unwaitable Conflict Mutual exclusion mark and occupied segment intersection All related steel coils in a mutually exclusive set Exclude the original path chain or include it in the detour chain In this embodiment, a mutually exclusive set is established for path segments that cannot be passed in parallel within the same candidate path chain. The mutually exclusive set consists of one-way passage relationship, intersection point sharing relationship and entrance buffer continuation relationship. When the continuous occupied segment of the target steel coil has time overlap with the registered occupied segment in the mutually exclusive set, the method splits the overlapping segment into a waitable segment and a non-waitable segment. The upstream position corresponding to the waitable segment has a node that can be stopped and does not block the exit of other candidate steel coils after stopping. The position corresponding to the non-waitable segment does not allow stopping or will cause warehouse back-blockage, intersection point blockage or entrance buffer sequence disruption after stopping.

[0053] Preferably, for waitable segments, the method retains the original candidate path chain and adjusts the start and end times of the corresponding node occupancy. During the adjustment, the occupancy time of all subsequent nodes and segments is shifted starting from the waitable node, and the unit entrance entry time after the shift is rechecked to see if it is still within the effective arrival window. For non-waitable segments, the method does not continue to accumulate waiting time on the original path chain, but generates a detour candidate path chain. The detour candidate path chain must re-register all node occupancy segments and participate in bottleneck constraint identification together with the existing task occupancy and candidate task occupancy. If the entrance arrival time of the detour candidate path chain exceeds the effective arrival window, the detour candidate path chain is excluded.

[0054] In this embodiment, the detour candidate path chain does not simply replace the path name. Instead, it uses the current conflict location as the dividing point to reassemble the upstream path segment, the alternative intermediate path segment, and the connecting path segment before the target unit entrance of the target steel coil. After splicing, it checks whether there is a passable relationship between the path segments, whether the intermediate path segment forms an unwaitable conflict with the locked task, and whether the connecting path segment maintains the entrance buffer order. If any check is not satisfied, the intermediate path segment is reselected or the process is returned to the local reordering process. After the detour candidate path chain is included in the subsequent bottleneck constraint identification, it may continue to affect the set of affected steel coils, so that the detour itself does not form new unrecorded conflicts.

[0055] Furthermore, the selection of the detour candidate path chain can be calculated using the following constraint values:

[0056] in, Indicates the first The steel coil uses the first Path constraint values ​​when there are 1 candidate path chain. This indicates the overall occupancy conflict value corresponding to the candidate path chain. Indicates the number of mutex occupancy markers. This indicates the deviation from the effective reach window boundary when entering the unit inlet. This indicates the number of path objects in the candidate path chain that need to be re-registered. , and These represent the weights of mutual exclusion, window offset, and path re-registration, respectively. If a steel coil has two candidate path chains, the first path chain... 4. For 2, For 1, The value is 5, the second path chain. For 6, 0, 0, It is 7. Take 3, Take 2, If we set it to 1, then the constraint value of the first path chain is 17, and the constraint value of the second path chain is 13. Subsequent deductions will prioritize retaining the second path chain.

[0057] In this embodiment, consistency verification is performed before the scheduling control command is issued. The verification does not change the execution state of the locked task, but performs a logical closure check on the corrected steel coil sequence. The check objects include the steel coil's outbound time, effective arrival window, path node occupancy period, mutual exclusion occupancy mark, and unit entrance buffer order. If the outbound time of a steel coil is earlier than the upstream completion time or the storage space release time, it is determined that it does not meet the outbound time boundary. If the occupancy period of a path node and the mutual exclusion occupancy mark intersect in a non-waiting manner, it is determined that its path occupancy is inconsistent. If the time when a steel coil enters the unit entrance exceeds the effective arrival window or disrupts the entrance buffer order, it is determined that its entrance continuity is inconsistent.

[0058] Preferably, when the consistency check finds that any check item is not satisfied, the method will add the corresponding steel coil and its adjacent sorting positions before and after it back to the set of affected steel coils, and regenerate the scheduling control instructions according to the local reordering method. The reason for including the adjacent sorting positions before and after in the set is that the entry buffer order and specification connection attributes usually form constraints between adjacent steel coils. If only a single steel coil is adjusted, it may cause new process continuity problems or path occupancy problems in adjacent positions. When regenerating the scheduling control instructions, the method still keeps the order of the locked steel coils and the steel coils that have entered the unit entry unchanged, and only recalculates the effective arrival window, path occupancy segment and release correction node within the set of affected steel coils.

[0059] In this embodiment, consistency verification can be represented by closure verification markers. Specifically, an outbound boundary marker, a path mutual exclusion marker, and an entry window marker are established for each steel coil. The outbound boundary marker is satisfied when the outbound time is within the available outbound time boundary. The path mutual exclusion marker is satisfied when all occupied segments avoid the mutual exclusion occupation marker. The entry window marker is satisfied when the entry time is within the valid arrival window and the buffer order does not conflict. Only when all three markers are satisfied can the scheduling instruction corresponding to the steel coil enter the issueable set. If any marker is not satisfied, the scheduling instruction corresponding to the steel coil is returned to the bottleneck constraint feedback process.

[0060] Furthermore, within the same control cycle, the candidate coil sequence may undergo multiple bottleneck constraint feedbacks. The method restricts repeated changes by freezing boundaries. The frozen boundaries include coils that have already started execution, coils that have completed the pre-stacking action, and coils that have entered the unit inlet buffer position. The path node occupancy time of the coils within the frozen boundaries is used as a fixed occupancy for subsequent simulations. The coils outside the frozen boundaries participate in local reordering, release time correction, or detour path chain generation based on the affected coil set. Freezing boundaries ensures that scheduling recalculation will not extend to tasks that have already been irreversibly executed on-site, keeping the control commands consistent with the on-site state.

[0061] In a preferred embodiment, the data content of the scheduling control instruction includes coil identifier, execution order, outbound time, starting warehouse location, target unit entrance, candidate path chain identifier, path node occupancy period, release correction node, entrance buffer location, and abnormal rollback flag. The abnormal rollback flag is used to record the bottleneck type triggered by the coil during consistency verification and the reason for re-adding it to the affected coil set. When the flag is read in subsequent control cycles, it can first check whether the same type of bottleneck still exists. If the bottleneck has been eliminated, the coil is restored to a normal candidate coil. If the bottleneck still exists, it continues to participate in the simulation according to the locally corrected boundary.

[0062] In this embodiment, the scheduling process works by mapping time constraints on the production side to space occupancy on the path side, and then feeding back bottleneck constraints on the path side to candidate sequences on the production side. Specifically, the process status and downstream unit process control cycle time determine the time range within which the steel coil can enter the unit, while the storage location status and storage area path occupancy status determine whether the steel coil can actually reach the target unit entrance within that time range. The spatiotemporal path network unifies the two types of constraints into node occupancy time periods. Bottleneck constraints transform unavoidable path conflicts into sequence correction inputs. Local reordering and release time correction rewrite the correction results into the spatiotemporal path network until the scheduling control command passes the consistency check.

[0063] In this embodiment, if a target unit requires the first steel coil to enter the entrance buffer position between the 20th and 35th minutes and the second steel coil to enter the entrance buffer position between the 36th and 50th minutes during continuous production, and both candidate steel coil A and candidate steel coil B meet the requirements of the target process and specification connection, and the current storage location of steel coil A is close to the unit entrance but forms an unwaitable conflict with the locked task at the intersection point, while the current storage location of steel coil B is farther away but can enter the entrance buffer position at the 42nd minute through another path, then the method will not select steel coil A solely based on distance. Instead, it will map steel coil B to the second candidate sorting position based on the effective arrival window, mutual exclusion occupancy marker, and path constraint value, and retain steel coil A in the subsequent feasible window.

[0064] Preferably, if steel coil A is blocked by other steel coils, and the blocking steel coil itself also belongs to the subsequent candidate steel coil set, the method writes the temporary occupancy segment of the blocking steel coil into the spatiotemporal path network, and determines whether the temporary storage location of the blocking steel coil can be connected with the subsequent outbound path. If it can be connected, the re-stacking movement of the blocking steel coil is not considered as an independent invalid occupancy, but is entered into the path deduction as a pre-occupancy segment for subsequent outbound preparation. If it cannot be connected, the temporary storage location of the blocking steel coil only participates in bottleneck identification as an occupancy constraint. In this embodiment, the re-stacking process is incorporated into the scheduling closed loop through the association between successive and subsequent processes.

[0065] Furthermore, if the same bottleneck segment appears repeatedly in multiple control cycles, the method writes the historical conflict time slice of the bottleneck segment as a known occupancy tendency into the spatiotemporal path network without changing the basic process. This allows subsequent candidate path chains to prioritize avoiding the high-conflict time slice of the segment. The historical conflict time slice is not used as a fixed prohibition condition, but rather participates in the calculation of the comprehensive occupancy conflict value and path constraint value. If the bottleneck segment is actually idle at a certain moment and the candidate steel coil can still meet the effective arrival window, the path deduction can still retain the candidate path chain passing through the segment, thereby avoiding the mechanical solidification of local historical conflicts into permanent constraints.

[0066] In this embodiment, the rolling update under abnormal conditions starts with the status refresh of the same control cycle. If the upstream completion time is delayed, the warehouse status changes, or the unit entrance buffer occupancy is extended, the method recalculates the effective arrival window of the affected steel coils and re-registers the path node occupancy period. If the change does not affect the locked steel coils and the steel coils that have entered the unit entrance, only the affected steel coil set outside the frozen boundary is locally corrected. If the change causes a planned steel coil to be unable to enter the unit entrance within the effective arrival window, the steel coil and its adjacent sorting positions are re-included in the candidate replacement calculation to generate a new steel coil execution order and path occupancy period.

[0067] Preferably, the method retains a corrected trajectory for the coil sequence within the same scheduling control cycle. The corrected trajectory records the changes in the coil from the original candidate sorting position to the corrected sorting position, the advancement or delay of the release time, the selection or exclusion of candidate path chains, and the source of bottleneck constraints. The corrected trajectory does not participate in the execution of control commands, but is used to verify whether the same bottleneck recurs in the next control cycle. If the same bottleneck recurs and the corresponding coil set is the same, the feedback range of the preceding path segment is directly expanded. If the same bottleneck recurs but the corresponding coil set has changed, the bottleneck constraint identification is re-performed to prevent the use of outdated affected ranges.

[0068] In this embodiment, data updates are kept consistent using a transactional approach. The process status, storage location status, unit inlet buffer status, and path occupancy status are all snapshotted after entering the same control cycle. Candidate coil sequence generation, spatiotemporal path network writing, bottleneck constraint feedback, and consistency verification are all performed based on the same snapshot. If the on-site status changes during execution, the change is processed in the next control cycle to avoid data inconsistency caused by the same coil's outbound time being based on the old storage location status and the path occupancy period being based on the new storage location status. At the beginning of the next control cycle, tasks that have been issued and entered the execution phase are written to the frozen boundary, and tasks that have not been issued re-participate in the candidate calculation.

[0069] Furthermore, the correction of the unit inlet buffer sequence follows the dual constraints of effective arrival window and specification connection attribute. If a steel coil is delayed in entering the inlet buffer position due to path bottleneck, the method checks whether its subsequent steel coil can enter the previous buffer position in advance. If the subsequent steel coil can be replaced in terms of process connection attribute and there is an overlap in effective arrival window, then the two are partially swapped. If the subsequent steel coil can arrive in advance but the specification connection with the previous steel coil is not continuous, then the swapping is not performed. Instead, the original steel coil path occupancy is adjusted by releasing the correction node or detouring the path chain, thereby avoiding path-side adjustment from disrupting the production process control constraints.

[0070] In this embodiment, the entry buffer position is treated as a path node, and the entry buffer order is treated as the time occupancy order on that path node. Therefore, entry queuing is not a separate production scheduling problem, but a form of path node occupancy conflict. If two steel coils enter the same entry buffer position at adjacent times, the method checks whether the previous steel coil will occupy the entry time of the subsequent steel coil due to the unit's receiving delay. If occupancy overlap occurs, the entry buffer position is recorded as a bottleneck node, and the steel coils that have a successor relationship with it are included in the set of affected steel coils. Subsequently, the arrival window and release time are adjusted in a coordinated manner.

[0071] Preferably, the processing of storage location obstruction levels adopts a recursive expansion method. If the target steel coil is obstructed by one steel coil, a temporary occupation segment is formed. If the target steel coil is obstructed by multiple steel coils, multiple temporary occupation segments are formed sequentially from the outside to the inside according to the obstruction chain. Each temporary occupation segment includes the steel coil that was removed, the temporary storage node, and the start and end times of occupation. If any steel coil that was removed in the obstruction chain also belongs to the candidate steel coil set, the method retains its substitutability in the subsequent candidate sorting position and uses its temporary storage node as the new current storage location to participate in path deduction, so that the storage location obstruction information can be continuously transmitted between steel coil removal and subsequent scheduling.

[0072] In this embodiment, the method can be implemented by the scheduling calculation program in the existing process control platform of the cold rolling mill. The scheduling calculation program receives production plan, warehouse inventory, unit inlet buffer and path task registration data, generates a status snapshot according to the same control cycle, and then forms candidate steel coil sequence, spatiotemporal path network, bottleneck constraints, local correction results and scheduling control instructions. The scheduling control instructions are output to warehouse handling task management and downstream unit inlet buffer management. The implementation does not rely on new hardware structure, and the required data all come from the existing process control data objects in the steel coil process flow and warehouse path execution.

[0073] In this embodiment, the implementation boundary of the method can be configured according to the organization of the cold rolling mill warehouse area. If there are multiple entrance channels in the warehouse area, the spatiotemporal path network includes multiple starting directions and intersection points. If the same target unit corresponds to multiple entrance buffer positions, each buffer position serves as an independent entrance node and shares the same unit receiving cycle. If the steel coil needs to be transferred across areas, the cross-area connection point serves as a connecting node between path segments and participates in the occupancy registration. The configuration change does not change the basic processing logic of candidate steel coil sequence generation, path occupancy deduction, bottleneck constraint feedback, and consistency verification.

[0074] Specifically, during implementation, steel coils within the current control cycle can be divided into locked steel coils, candidate steel coils, and steel coils not currently involved. Locked steel coils include those currently being handled, those that have completed pre-stacking, or those that have entered the unit entrance. Candidate steel coils include those that match the target process and have a valid arrival window. Steel coils not currently involved include those whose upstream processes are incomplete, whose target processes do not match, or those that cannot form a valid arrival window. The spatiotemporal path network registers fixed occupancy for locked steel coils, registers deduced occupancy for candidate steel coils, and only retains storage space occupancy for steel coils not currently involved, without participating in path chain generation, thereby reducing invalid path deduction.

[0075] Furthermore, if there are a large number of candidate steel coils, the method can first group the candidate steel coils according to the target process and the unit entrance, and each group forms a separate candidate steel coil sequence and effective arrival window. Then, cross-group occupancy comparison is performed on the shared path section. Cross-group occupancy comparison is only performed on the channel intersection point, cross-area connection point and the channel before the shared entrance. If cross-group occupancy forms a bottleneck section, the corresponding steel coils are fed back to their respective candidate steel coil sequences for local correction. After correction, the occupancy time period of the shared path section is written back, so that the local scheduling closed loop can still be maintained when multiple units share the warehouse channel.

[0076] In this embodiment, the data reference relationship between each implementation object remains unique. The steel coil status record is indexed by the steel coil identifier, the path node and path segment are indexed by the warehouse network identifier, the candidate sorting position is indexed by the target unit and time position, the bottleneck constraint is indexed by the path object and conflict time slice, and the scheduling control instruction is indexed by the steel coil identifier and control cycle. Any steel coil can only have one confirmed outbound time and one confirmed inbound buffer position within the same control cycle. Multiple candidate path chains can exist, but only the path chain that passes the consistency check is written into the distributable set.

[0077] In this embodiment, after implementation, a data closed loop that can be repeatedly corrected is formed between the candidate steel coil sequence on the production side and the execution of the storage area path. The process status of the steel coil determines the candidate range, the path occupancy deduction determines the bottleneck constraint, the bottleneck constraint determines the local correction range, and the local correction result re-enters the path occupancy deduction. The scheduling control command is issued after consistency verification. The advantage of this embodiment is that it can identify situations where production scheduling is feasible but the path is unreachable before the control command is generated, and maintain the unit entrance buffer sequence consistent with the actual reachability of the storage area through sorting correction, release time correction, and path chain correction within the range of unreleased steel coils.

Claims

1. A method for intelligent scheduling and path optimization of steel coils in a cold rolling mill, characterized in that, include: Obtain the process status, warehouse location status, downstream unit process control cycle time, and warehouse path occupancy status of steel coils from the cold rolling mill; Based on the process status and the downstream unit process control cycle time, a candidate steel coil sequence and a corresponding allowed arrival window are generated; The candidate steel coil sequence is mapped to a spatiotemporal path network consisting of storage location nodes, path segment nodes, and unit entry nodes, and the node occupancy time period of each steel coil at different outbound times is deduced. Based on the node occupancy time period, identify bottleneck constraints formed by channel overlap, entrance queuing, and stacking failure association; The bottleneck constraints are fed back to the candidate steel coil sequence, and the release order, outbound time and arrival window of the steel coils are adjusted in a coordinated manner to generate scheduling control instructions that include the execution order of steel coils, the time period occupied by path nodes and the buffer order of unit inlet.

2. The intelligent scheduling and path optimization method for steel coils in a cold rolling mill according to claim 1, characterized in that, Based on the process status and the downstream unit process control cycle time, a candidate coil sequence and corresponding allowed arrival window are generated, including: Establish a status record for each steel coil, which includes at least the current warehouse location, target process, upstream completion time, downstream unit cycle time, specification connection attribute, inventory obstruction level, and remaining waiting tolerance. A candidate steel coil set is formed based on the constraints that the target process is the same, the specifications can be connected continuously, and the upstream completion time meets the conditions for entering the line. Based on the downstream unit cycle time and inlet buffer capacity, the allowable arrival window for the steel coils in the candidate steel coil set is determined, and multiple alternative candidate sorting positions are reserved within the allowable arrival window.

3. The intelligent scheduling and path optimization method for steel coils in a cold rolling mill according to claim 1, characterized in that, Mapping the candidate steel coil sequence to a spatiotemporal path network includes: The storage location, the intersection of the storage area passage, the cross-area connection point, and the unit entrance are defined as path nodes with time occupancy attributes, and the passability relationship between adjacent path nodes is defined as path segments. For each candidate steel coil, a continuous occupancy segment is formed based on the time of departure from the warehouse, the duration of handling, the time occupied by stacking, and the waiting time at the entrance. The continuously occupied segments are written into the corresponding path nodes and path segments, so that the process arrival window of the same steel coil and the path occupancy period are associated in the same time coordinate.

4. The intelligent scheduling and path optimization method for steel coils in a cold rolling mill according to claim 1, characterized in that, Bottleneck constraints are identified based on the node's occupancy time period, including: Compare the segments occupied by different steel coils on the same path section, the same storage channel, and the same unit inlet buffer position in the same time coordinate; When there is an overlap in occupied segments, and the steel coils corresponding to the overlap cannot fall into their respective allowed arrival windows by adjusting the outbound time, the overlapping position is recorded as a bottleneck segment. Steel coils that have path succession relationships, stacking block relationships, or entrance queuing relationships with the bottleneck section are classified into the affected steel coil set, and the bottleneck section, conflict time slice, and affected steel coil set are combined into the bottleneck constraint.

5. The intelligent scheduling and path optimization method for steel coils in a cold rolling mill according to claim 2, characterized in that, Determining the allowed arrival window for steel coils within the candidate set of steel coils includes: Based on the downstream unit process control cycle time, the earliest and latest arrival times of each candidate sorting position are calculated backward according to the unit inlet buffer capacity. Convert the upstream completion time of steel coils, the current warehouse release time, and the inventory obstruction level into the time boundary for shipment. The overlap between the available outbound time boundary and the earliest and latest arrival times is processed to form an effective arrival window for steel coils to participate in the sorting. When multiple valid arrival windows correspond to the same sorting position, steel coils with consistent process connection attributes and comparable path occupancy differences are retained as candidate replacement objects.

6. The intelligent scheduling and path optimization method for steel coils in a cold rolling mill according to claim 3, characterized in that, Forming the continuous occupied segment includes: Based on the traversable path from the current location of the steel coil to the entrance of the target unit, a candidate path chain is generated in the order of the path nodes. In the candidate path chain, start and end times are allocated for storage space release, channel entry, intersection point passage, cross-regional connection, and entrance buffer, respectively; For steel coils involving stacking, establish a sequential relationship between the temporary occupancy segment of the steel coil to be removed and the outbound occupancy segment of the target steel coil; Set a mutual exclusion occupancy mark for path segments that cannot be passed in parallel within the same candidate path chain, and write the mutual exclusion occupancy mark into the spatiotemporal path network.

7. The intelligent scheduling and path optimization method for steel coils in a cold rolling mill according to claim 4, characterized in that, Feeding the bottleneck constraint back to the candidate steel coil sequence includes: Based on the sorting position of the affected steel coil set in the candidate steel coil sequence, extract the steel coils that are still in an unreleased state and whose process connection attributes meet the requirements of adjacent sorting positions; Using the premise of not changing the order of locked steel coils and steel coils that have entered the unit inlet as the boundary, perform a local reordering on the extracted steel coils; Recalculate the effective arrival window and path segment occupied by the reordered steel coils; When there is still overlap in the occupation of the same bottleneck section during local reordering, the preceding path section connected to the bottleneck section is included in the feedback range, and the release time of the steel coil is redefined.

8. The intelligent scheduling and path optimization method for steel coils in a cold rolling mill according to claim 7, characterized in that, The timing of the steel coil release has been redefined, including: Based on the conflict time slice of the bottleneck section, trace upstream along the path to the first warehouse location node or channel waiting node that can wait without blocking other candidate steel coils from leaving the warehouse; Set the traced storage location node or channel waiting node as the release correction node; Based on the path segment occupied by the steel coil from the release correction node to the inlet of the target unit, calculate the release delay or advance. The release delay or advance amount is simultaneously written into the steel coil's outbound time, effective arrival window, and unit inlet buffer sequence, so that the sequence adjustment and path occupancy adjustment are completed within the same control cycle.

9. The intelligent scheduling and path optimization method for steel coils in a cold rolling mill according to claim 6, characterized in that, Set mutual exclusion markers for path segments that cannot be traversed in parallel within the same candidate path chain, including: Establish mutually exclusive sets based on the one-way traffic relationship of path segments, the sharing relationship of intersection points, and the buffer connection relationship of entrances; When the consecutive occupied segments of the target steel coil overlap with the registered occupied segments in the mutually exclusive set in time, the overlapping segments are split into waitable segments and non-waitable segments; For the waitable segments, the original candidate path chains are retained and the start and end times of the corresponding node occupation are adjusted. For the non-waitable segments, detour candidate path chains are generated, and the occupied segments of the detour candidate path chains are synchronously included in the subsequent bottleneck constraint identification.

10. The intelligent scheduling and path optimization method for steel coils in a cold rolling mill according to claim 8, characterized in that, After simultaneously writing the release delay or advance amount into the steel coil's outbound time, effective arrival window, and unit inlet buffer sequence, it also includes: Before the dispatch control command is issued, a consistency check is performed on the corrected steel coil sequence; The consistency check includes comparing whether the outbound time of the same steel coil falls within the outbound time boundary, whether the path node occupancy period avoids the mutual exclusion occupancy mark, and whether the unit inlet buffer sequence is within the effective arrival window. If any verification item is not satisfied, the corresponding steel coil and its adjacent sorting positions before and after it will be added back to the affected steel coil set, and the scheduling control instruction will be regenerated in accordance with the local reordering method defined in claim 7.