An industrial intelligent control method, device and medium based on cloud-edge-device collaboration

By using a cloud-edge-device collaborative control method, the operating condition phase and safety constraint boundaries are identified, an edge real-time control window is generated, and the strategy trajectory is matched. This solves the access mismatch problem between cloud optimization strategies and edge real-time control, and achieves orderly control takeover and stable execution.

CN123113583APending Publication Date: 2026-11-17BEIJING RES INST OF AUTOMATION FOR MACHINERY IND
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202611460197.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2026-09-20
Publication Date
2026-11-17

AI Technical Summary

Technical Problem

There are issues of access mismatch and takeover uncertainty within the same control cycle between cloud-based optimization strategies and edge real-time control, resulting in unclear takeover scope of control program segments, which affects the stable execution of real-time actions and subsequent calibration.

Method used

By establishing cloud-edge-device collaborative control tasks, binding control program fragments and edge-side hierarchical backoff constraints, identifying operating condition phases, safety constraint boundaries, and control time limit margins, generating an edge real-time control window, and combining it with historical operating condition control records for strategy matching, a cloud-based optimized strategy track and an edge real-time control track are formed, realizing dual-track differential locking and control takeover methods, and generating control execution instructions.

Benefits of technology

It achieves access adaptation between cloud-based optimization strategies and edge real-time control within the same control cycle, ensuring that control execution commands are generated under safety constraints and time limits. It also provides cloud-edge-device collaborative calibration records to ensure the orderly takeover and stable execution of the control system.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN123113583A_ABST
    Figure CN123113583A_ABST
Patent Text Reader

Abstract

This invention discloses an industrial intelligent control method, device, and medium based on cloud-edge-device collaboration, relating to the field of industrial control technology. The method includes: reading the applicable operating conditions and effective boundaries of a strategy based on a cloud-optimized strategy track; determining the strategy access status by combining the current operating condition phase and safety constraint boundaries in the edge real-time control window; forming a strategy access status; locking the edge real-time control track and the cloud-optimized strategy track accessed in the current cycle based on the strategy access status; determining the takeover method based on the difference level and control time limit margin; locking the control program segment corresponding to the current operating condition phase based on the control takeover method; and confirming the segment takeover range by combining the segment start boundary, segment end boundary, and safety constraint boundaries. This invention, by forming a strategy access status, achieves the adaptation and confirmation between the cloud-optimized strategy track and the current operating condition phase and safety constraint boundaries before entering the control cycle.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] This invention relates to the field of industrial control technology, and in particular to an industrial intelligent control method, device and medium based on cloud-edge-device collaboration. Background Technology

[0002] With the development of intelligent manufacturing, industrial internet and edge computing technologies, industrial control systems are gradually evolving from single field control to cloud-edge-device collaborative control. The correlation between controlled objects, control programs, process cycles and production line-level control objectives is constantly improving. The cloud has the ability to analyze historical operating conditions and optimize cross-production line strategies, the edge has the ability to perform real-time field verification and generate control commands, and the device-side actuators have the ability to collect status and provide action feedback. The collaboration of these three has become an important technical direction for improving the intelligence, precision and continuity of industrial production processes.

[0003] In relevant industrial control systems, cloud-based optimization strategies are typically generated based on historical data and production line targets, while edge real-time control relies on the current operating conditions and on-site safety boundaries for execution. The two types of control bases are prone to misalignment in terms of control cycle, applicable boundaries, and execution time limits. If the cloud-based strategy participates in control without judging the current operating condition phase and safety constraint boundaries, it may lead to unclear control program segment takeover scope, insufficient basis for edge backoff triggering, and further affect the stable execution of real-time actions and subsequent calibration. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an industrial intelligent control method based on cloud-edge-device collaboration to solve the problems of access mismatch and takeover uncertainty between cloud optimization strategies and edge real-time control within the same control cycle.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution: In a first aspect, this invention provides an industrial intelligent control method based on cloud-edge-device collaboration, comprising: establishing a cloud-edge-device collaborative control task based on an industrial control object, binding control program segments and edge-side hierarchical backoff constraints, identifying the operating condition phase, safety constraint boundaries, and control time limit margin through edge-side state acquisition and edge verification to obtain an edge real-time control window; extracting edge real-time control basis according to the edge real-time control window, and performing cloud strategy matching between the window state summary and historical operating condition control records and production line-level control targets to generate an edge real-time control track and a cloud-optimized strategy track; reading the applicable operating conditions and policy effectiveness boundaries of the strategy based on the cloud-optimized strategy track, and determining the strategy access status in conjunction with the current operating condition phase and safety constraint boundaries in the edge real-time control window, forming... The system establishes a policy access status. Based on this status, it performs dual-track differential locking on the edge real-time control track and the cloud-optimized policy track for the current cycle access. The takeover method is determined based on the differential level and control time limit margin. The control takeover method is then determined by locking the control program segment corresponding to the current operating phase. The takeover range of the segment is confirmed by combining the segment's start boundary, end boundary, and safety constraint boundary. The correction range of the current control program segment is then locked based on the takeover range. End-side hierarchical backoff constraints are embedded into the segment connection process to obtain control execution instructions. These instructions drive the industrial controlled object to perform real-time actions. End-side execution feedback, policy access status, and control takeover method are correlated and calibrated to form a cloud-edge-end collaborative calibration record.

[0007] As a preferred embodiment of the cloud-edge-device collaborative industrial intelligent control method of the present invention, the steps for obtaining the edge real-time control window are as follows: Taking the control objective of the industrial control object as the starting point of the task, the process cycle is mapped to the control cycle, the allowed action boundary and the expected execution time of the control program segment are bound to the control program segment, and the end-side hierarchical backoff constraint is written into the failure takeover condition of the control program segment to obtain the control task event carrier; The control task event carrier triggers the edge-side state acquisition, and performs edge verification on the sampling timing, execution feedback, available status of control program segments, and communication arrival status to form an edge-side trusted state carrier containing the available status of control program segments, edge-side communication latency, and edge verification time. The current operating condition phase is identified based on the trusted state carrier on the edge side, and the allowed action boundaries and process cycle times in the control task event carrier are read. The current operating condition phase is mapped to the allowed action boundaries to determine the safety constraint boundaries. The control time limit margin is determined based on the edge communication delay, edge verification time, expected execution time of the control program segment, and process cycle times, and an edge real-time control window is generated.

[0008] As a preferred embodiment of the cloud-edge-device collaborative industrial intelligent control method of the present invention, the steps for generating the edge real-time control track and the cloud optimization strategy track are as follows: Based on the edge real-time control window, the current operating condition phase is associated with the safety constraint boundary and control time limit margin, and the available state of the control program segment is incorporated into the association result to obtain the edge real-time control basis. Based on the edge real-time control criteria, edge control instructions are determined, mapped to the instruction target and the execution order of the segments, and the edge control execution boundary is encapsulated according to the safety constraint boundary and the control time limit margin to generate the edge real-time control track. The window status summary is generated using the edge real-time control window. The window status summary is matched with the historical operating condition control records. The matching result is then corrected according to the production line-level control objectives to a cloud-optimized strategy track that is in the same control cycle as the edge real-time control track.

[0009] As a preferred embodiment of the cloud-edge-device collaborative industrial intelligent control method of the present invention, the steps for forming the policy admission state are as follows: Read the applicable operating conditions and effective boundaries of the policy from the cloud-optimized policy track, and read the current operating condition phase and safety constraint boundaries from the edge real-time control window to obtain the policy access evidence body; Based on the verification of the policy access evidence body, the working condition phase consistency and boundary coverage consistency between the cloud-optimized policy track and the edge real-time control window are obtained to obtain the policy boundary hit status. Based on the policy boundary hit status, the cloud-based optimization policy track is assessed for current cycle admission, adjacent cycle delay, and policy freeze status, thus forming the policy admission status.

[0010] As a preferred embodiment of the cloud-edge-device collaborative industrial intelligent control method of the present invention, the steps for obtaining the control takeover mode are as follows: The cloud-optimized policy track for the current cycle is selected based on the policy admission status. When the policy admission status hits either the adjacent cycle's delayed state or the policy freeze state, the edge real-time control track is used as the basis for takeover in the current control cycle. When the cloud-optimized policy track is the current cycle's admission, the cloud-optimized policy track and the edge real-time control track for the current cycle are placed in the same control cycle to lock the consistency of control direction, control amplitude, instruction target, and edge control execution boundary, thus obtaining the dual-track difference locking result. Based on the results of dual-track difference locking, dual-track differences with consistent control direction and control amplitude within the edge control execution boundary are classified into low difference level, dual-track differences with consistent control direction and control amplitude touching the edge control execution boundary are classified into medium difference level, and dual-track differences with inconsistent control direction and inconsistent instruction action objects are classified into high difference level, thus obtaining the difference level. The strategy admission status is used as a prerequisite for takeover, and the takeover hit determination is made in combination with the difference level and control time limit margin. When the strategy admission status hits any of the adjacent cycle delay status and strategy freeze status, and the control time limit margin meets the current cycle execution requirements, the edge real-time control track continues to execute. When the difference level is low and the control time limit margin meets the current cycle execution requirements, the edge real-time control track continues to execute. When the difference level is medium and the control time limit margin meets the segment correction requirements, the cloud optimization strategy track participates in correction. When the difference level is high, the end-side layered backoff constraint intervenes. When the control time limit margin does not meet the segment correction requirements, the end-side layered backoff constraint intervenes, thus obtaining the control takeover mode.

[0011] As a preferred embodiment of the cloud-edge-device collaborative industrial intelligent control method of the present invention, the steps for confirming the segment takeover range are as follows: Based on the control takeover method, the current operating condition phase in the edge real-time control window is read, the control program segment corresponding to the current operating condition phase is retrieved from the control task event carrier, and the control program segment is matched with the segment execution order in the edge real-time control track to obtain the phase segment index. The control program segment start boundary and segment end boundary are read based on the phase segment index, and the segment start boundary and segment end boundary are associated with the safety constraint boundary in the edge real-time control window to obtain the segment boundary set; Based on the fragment boundary set, the takeover range of the control program fragment is locked by boundary overlap, and the fragment range covered by the safety constraint boundary is taken as the allowable takeover range, thus obtaining the fragment takeover range.

[0012] As a preferred embodiment of the cloud-edge-device collaborative industrial intelligent control method of the present invention, the steps for obtaining the control execution command are as follows: Based on the segment takeover range and control takeover method, the correction range of the current control program segment is locked within the segment takeover range to obtain the segment correction range; Based on the segment correction range, the end-side hierarchical backoff constraints in the control task event carrier are written into the segment connection process of the current control program segment, and the edge control instructions, segment execution order, segment correction range and end-side hierarchical backoff constraints in the edge real-time control track are encapsulated to obtain the control execution instructions.

[0013] As a preferred embodiment of the cloud-edge-device collaborative industrial intelligent control method of the present invention, the steps for forming the cloud-edge-device collaborative calibration record are as follows: The system drives industrial control objects to perform real-time actions in the order of segment execution by control execution commands, and collects the execution completion status, post-execution measurement status, actuator feedback status, and safety constraint verification status to form end-side execution feedback. Based on the execution feedback from the end side and the control execution instructions, the execution consistency is verified under the same control cycle. The verified end side execution feedback is then associated with the policy access status and control takeover mode to form an execution association calibration result. Based on the execution correlation calibration results, mark the edge execution deviation, policy access deviation, and takeover method deviation, and encapsulate the control execution command, edge execution feedback, policy access status, control takeover method, and deviation mark to form a cloud-edge-device collaborative calibration record.

[0014] In a second aspect, the present invention provides a computer device, including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the cloud-edge-device collaborative industrial intelligent control method as described in the first aspect of the present invention.

[0015] Thirdly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the cloud-edge-device collaborative industrial intelligent control method as described in the first aspect of the present invention.

[0016] The beneficial effects of this invention are as follows: By forming a strategy admission state, the adaptation and confirmation between the cloud-optimized strategy track and the current operating condition phase and safety constraint boundary are realized before entering the control cycle; by performing dual-track difference locking on the edge real-time control track and the cloud-optimized strategy track of the current cycle admission, and judging the control takeover mode according to the difference level and control time limit margin, the orderly takeover between cloud strategy, edge real-time control and end-side hierarchical backoff in the industrial control system is realized, so that the control execution command can be generated under the condition of meeting the safety constraint boundary and control time limit requirements, and provides a clear basis for cloud-edge-end collaborative calibration records. Attached Figure Description

[0017] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the 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.

[0018] Figure 1 This is a flowchart of an industrial intelligent control method based on cloud-edge-device collaboration.

[0019] Figure 2 A flowchart for the real-time edge control window and dual-track generation.

[0020] Figure 3 This is a flowchart for determining the policy access and takeover hit.

[0021] Figure 4 This is a flowchart illustrating the linkage between the segment takeover range and the calibration execution.

[0022] Figure 5 This is a data chart showing the distribution of control over the different methods of takeover.

[0023] Figure 6 This is a data diagram showing the coverage of fragment takeover range and safety constraint boundaries. Detailed Implementation

[0024] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0025] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0026] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0027] Reference Figures 1-6 This is one embodiment of the present invention, which provides an industrial intelligent control method based on cloud-edge-device collaboration, comprising the following steps: S1. Establish cloud-edge-device collaborative control tasks based on industrial control objects, and bind control program fragments and edge-side hierarchical backoff constraints. Through edge-side status acquisition and edge verification, identify the operating condition phase, safety constraint boundaries and control time limit margin to obtain the edge real-time control window.

[0028] S1.1: Taking the control objective of the industrial control object as the starting point of the task, the process cycle is mapped to the control cycle, the allowed action boundary and the expected execution time of the control program segment are bound to the control program segment, and the end-side hierarchical backoff constraint is written into the failure takeover condition of the control program segment to obtain the control task event carrier; It should be noted that industrial control objects refer to equipment or production line links in the industrial production process that need to be monitored, regulated, and controlled.

[0029] Specifically, taking the control objective of the industrial control object as the starting point of the task, the system reads the operating status requirements and action result requirements of the industrial control object in the current process task, and matches the operating status requirements and action result requirements with the corresponding process cycle. The process cycle defines the control cycle of the control program segment. The system reads the permissible action boundary and the expected execution time of the control program segment, writes the permissible action boundary into the action range field of the control program segment, and writes the expected execution time of the control program segment into the time limit field of the control program segment. The system reads the end-side hierarchical backoff constraint and writes the end-side hierarchical backoff constraint into the failure takeover condition of the control program segment, so that the control program segment can trigger the corresponding end-side hierarchical backoff constraint when it does not meet the permissible action boundary, the expected execution time of the control program segment, and the control objective. The system associates and encapsulates the control objective, control cycle, permissible action boundary, control program segment, expected execution time of the control program segment, and failure takeover condition to obtain the control task event carrier.

[0030] It should be noted that the control objective is the operating state and action result that the industrial control object needs to achieve in the current process task; The process cycle is the time rhythm of an industrial control object during the connection between adjacent processes, and the control cycle is the single control processing time period obtained according to the process cycle. Permitted action boundaries define the range of actions and safety limits that a control program segment is allowed to perform during execution. The control program segment is a program segment used to drive the industrial controlled object to perform control actions under the current operating condition phase; The estimated execution time of a control program segment is the time required for the control program segment to complete the corresponding control action from the start of execution. The end-side hierarchical rollback constraint is a restriction condition for the end-side to take over the control program segment step by step according to the safety level when the control program segment fails. The failure takeover condition is the condition that triggers the end-side hierarchical backoff constraint when the control program segment does not meet the control objective, the allowed action boundary, and the expected execution time of the control program segment.

[0031] S1.2: The control task event carrier triggers the end-side state acquisition, and performs edge verification on the sampling timing, execution feedback, available status of control program segments and communication arrival status to form an end-side trusted state carrier containing the available status of control program segments, end-side communication delay and edge verification time. Specifically, the control cycle, control program segment, permissible action boundaries, and failure takeover conditions are read from the control task event carrier. End-side status acquisition is triggered according to the control cycle, forming the end-side acquisition time, execution feedback, control program segment call status, and communication arrival status. The sampling timing is verified against the end-side acquisition time and edge reception time to ensure it falls within the control cycle. Based on consecutive adjacent execution feedback within the same control cycle, the action state changes corresponding to adjacent execution feedback are verified to ensure they fall within the permissible action boundaries. The order of adjacent execution feedback is also verified to ensure it meets the action connection requirements of the control program segment. When both adjacent execution feedback falls within the permissible action boundaries, and the adjacent execution feedback... When the sequence of actions is consistent with the action connection requirements of the control program segment, the execution feedback is deemed to meet the action connection requirements. The available status of the control program segment is verified according to the call status of the control program segment. The arrival interval from the acquisition time at the end side to the reception time at the edge is recorded according to the communication arrival status, and the end side communication delay is determined. The edge verification time is determined according to the verification interval from the edge reception time to the end of the edge verification. The verification results of the sampling timing, the verification results of the execution feedback, the available status of the control program segment, the communication arrival status, the end side communication delay, and the edge verification time are associated and encapsulated to form an end side trusted status carrier containing the available status of the control program segment, the end side communication delay, and the edge verification time.

[0032] It should be noted that edge status acquisition is the process of acquiring the operating status of the industrial controlled object and the execution status of the control program segment according to the control cycle; Execution feedback refers to the action status returned by the industrial controlled object after executing a control program segment; The control program segment invocation status is the state in which the control program segment is retrieved by the terminal and is ready to be executed. The communication arrival status is the status of the end-side status acquisition results reaching the edge side. The communication arrival status is used to confirm the arrival interval between the end-side acquisition time and the edge reception time, and the arrival interval determines the end-side communication delay. The available state of a control program segment is the state in which the control program segment can be called normally and meets the usage requirements of the current control cycle. End-side communication delay is the time interval between the end-side acquisition time and the edge reception time; Edge verification time is the time interval between the edge reception time and the edge verification end time.

[0033] S1.3: Identify the current operating condition phase based on the trusted state carrier on the edge side, and read the allowed action boundary and process cycle from the control task event carrier. Map the current operating condition phase to the allowed action boundary to determine the safety constraint boundary. Determine the control time limit margin based on the edge communication delay, edge verification time, expected execution time of the control program segment, and process cycle, and generate the edge real-time control window.

[0034] Specifically, the sampling timing, execution feedback, available status of control program segments, end-side communication latency, and edge verification time are read from the edge-side trusted state carrier. The current operating condition phase is identified based on the control cycle position corresponding to the sampling timing and execution feedback. After obtaining the current operating condition phase, the allowed action boundary and process cycle are read from the control task event carrier. The current operating condition phase is mapped to the action range that matches the current operating condition phase within the allowed action boundary, and the matched action range is determined as the safety constraint boundary. The available processing time period of the current control cycle is determined based on the process cycle. The end-side communication latency, edge verification time, and expected execution time of the control program segment are deducted from the time period occupied by the available processing time to obtain the control time limit margin. The current operating condition phase, safety constraint boundary, control time limit margin, and available status of the control program segment are associated to generate an edge real-time control window.

[0035] It should be noted that the current operating condition phase includes the start-up preparation phase, steady-state operation phase, load switching phase, process transition phase, anomaly recovery phase, and shutdown closing phase. The current operating condition phase is determined based on the sampling timing, execution feedback, and the available state of the control program segment. When the control program segment has been retrieved and the execution feedback has not yet formed continuous action feedback, it is identified as the start-up preparation phase. When the execution feedback is continuous within the control cycle and the action state is within the allowable action boundary, it is identified as the steady-state operation phase. When the action result corresponding to the control target changes and the execution feedback shows continuous changes corresponding to the load change, it is identified as the load switching phase. When the process cycle corresponds to the connection of adjacent processes and the execution feedback changes from the previous action state to the next action state, it is identified as the process transition phase. When the execution feedback did not meet the allowable action boundary and recovered to the allowable action boundary after being controlled by the end-side layered back-off constraint, it is identified as the anomaly recovery phase. When the action result corresponding to the control program segment has reached the control target and the subsequent execution feedback enters the end-holding state, it is identified as the shutdown closing phase. The identification of the current operating condition phase is based on the execution feedback within the same control cycle and adjacent control cycles. Continuous action feedback, continuous change, and end-hold all refer to the execution feedback maintaining the same feedback trend, the same switching trend, or the same end-hold state within the preset continuous cycle requirement. The preset continuous cycle requirement is set according to the process cycle time, the expected execution duration of the control program segment, and the end-side status acquisition frequency. When the execution feedback does not exceed the allowable action boundary and corresponds to the current control target, it is considered as valid feedback for current operating condition phase identification. When the same execution feedback simultaneously meets the identification conditions of multiple current operating condition phases, it is prioritized in the following order: abnormal recovery phase, load switching phase, process transition phase, start-up preparation phase, steady-state operation phase, and shutdown termination phase.

[0036] The sampling timing sequence is the chronological order in which the end-side status is acquired within the control cycle; The edge reception time is the time point at which the edge side receives the status acquisition results from the end side. The edge verification end time is the time point at which the edge side completes the verification of sampling timing, execution feedback, available status of control program segments, and communication arrival status. Process cycle time refers to the operating time rhythm of an industrial control object during the connection between adjacent processes, and is used to divide the control cycle and determine the real-time control processing period. Permissible action boundaries refer to the range of motion, speed, position, and safety limits that a control program segment is allowed to achieve when driving an industrial controlled object to perform actions; Control time limit margin refers to the remaining time available for control takeover and segment correction within the current control cycle after deducting end-side communication delay, edge verification time, and the expected execution time of the control program segment. The control time limit margin is determined by deducting end-side communication delay, edge verification time, and the expected execution time of the control program segment from the available processing time corresponding to the current control cycle. When the remaining time after deduction is less than or equal to 0, it is determined that the control time limit margin does not meet the takeover and segment correction requirements of the current control cycle, and the end-side hierarchical backoff constraint is directly entered. When any one of the end-side communication delay, edge verification time, and the expected execution time of the control program segment is missing, a time limit margin pending verification status is generated, and the cloud optimization strategy track is prohibited from participating in the segment correction of the current control cycle.

[0037] S2. Extract the edge real-time control basis from the edge real-time control window, and match the window status summary with the historical operating condition control records and production line-level control targets using cloud strategies to generate edge real-time control tracks and cloud optimization strategy tracks.

[0038] S2.1: Based on the edge real-time control window, associate the current operating condition phase with the safety constraint boundary and control time limit margin, and incorporate the available state of the control program segment into the association result to obtain the edge real-time control basis; Specifically, using the edge real-time control window as the basis for organization, the current operating phase, safety constraint boundaries, control time limit margin, and available status of control program segments corresponding to the edge real-time control window are determined. The current operating phase is mapped to the safety constraint boundaries to determine the usable safety constraint boundaries under the current operating phase, and the control time limit margin is mapped to the current operating phase to determine the remaining time conditions available for control processing under the current operating phase. The available status of control program segments is incorporated into the association result formed by the current operating phase, safety constraint boundaries, and control time limit margin, so that the association result can represent the safety constraint boundaries, control time limit margin, and available status of control program segments under the current operating phase. The current operating phase, safety constraint boundaries, control time limit margin, and available status of control program segments are associated and encapsulated to obtain the edge real-time control basis.

[0039] S2.2: Based on the edge real-time control criteria, determine the edge control instructions, map the edge control instructions to the instruction target and the execution order of the segments, and encapsulate the edge control execution boundary according to the safety constraint boundary and the control time limit margin to generate the edge real-time control track; Specifically, using real-time edge control as the processing basis, the system determines the safety constraint boundary, control time margin, and available status of control program segments under the current operating phase. When the available status of the control program segments meets the requirements of the current control cycle, edge control commands are determined based on the control objectives and safety constraint boundaries corresponding to the current operating phase. According to the control program segments corresponding to the edge control commands, the edge control commands are associated with the industrial control objects that need to perform actions, forming command application objects. The execution order of the segments is determined according to the action connection relationship of the control program segments within the control cycle. The safety constraint boundary is used as the action range limit for the edge control commands, and the control time margin is used as the execution time limit for the edge control commands, encapsulating them to form the edge control execution boundary. The edge control commands, command application objects, segment execution order, and edge control execution boundaries are associated to generate the edge real-time control track.

[0040] It should be noted that the current control cycle usage requirement means that the control program segment can be called normally within the current control cycle, and that the expected execution time, safety constraint boundaries, and control time limit margin of the control program segment all meet the real-time control processing requirements.

[0041] S2.3: Generate a window status summary using the edge real-time control window, match the window status summary with historical operating condition control records, and correct the matching result to a cloud-optimized strategy track that is in the same control cycle as the edge real-time control track according to the production line-level control objectives.

[0042] Specifically, using the edge real-time control window as the basis for organization, the current operating condition phase, safety constraint boundary, control time limit margin, and available status of control program segments in the edge real-time control window are summarized to form a window state summary. The current operating condition phase in the window state summary is matched for consistency with the operating condition phase in historical operating condition control records, retaining historical operating condition control records with consistent operating condition phases. In the historical operating condition control records with consistent operating condition phases, the safety constraint boundary in the window state summary is matched for coverage relationships with the safety constraint boundary in the historical operating condition control records, retaining historical operating condition control records where the safety constraint boundary can cover the current safety constraint boundary. In the historical operating condition control records after safety constraint boundary matching, the control time limit margin in the window state summary is matched for time limit adaptation with the corresponding control execution results of the historical operating condition control records, retaining historical operating condition control records where the control execution results can form effective execution feedback within the control time limit margin, and using the retained historical operating condition control records as the matching results. If no historical operating condition simultaneously meets the requirements of the current operating condition phase, safety constraint boundary, and control time limit margin, the process is considered complete. During control recording, records of missing cloud-optimized strategy tracks and strategy matching failures are generated. The current cycle's edge real-time control track is used as the takeover basis, and the reason for matching failure is written into the cloud-edge-device collaborative calibration record. The matching results are correlated with the production line-level control objectives. When the production line-level control objective is to maintain the cycle time, the control directions, amplitudes, and objects that do not increase the control time margin are prioritized to retain the control directions, amplitudes, and objects that do not increase the control time margin are prioritized to retain the control directions, amplitudes, and objects that fall within the energy consumption constraint range corresponding to the production line-level control objective and can achieve the control objective are prioritized to retain the control directions, amplitudes, and objects that fall within the allowable quality range corresponding to the production line-level control objective and whose measurement status after execution falls within the allowable quality range corresponding to the production line-level control objective are prioritized to retain the control directions, amplitudes, and objects that fall within the allowable quality range corresponding to the production line-level control objective after the measurement status after execution is prioritized to retain the control directions, amplitudes, and objects that fall within the allowable quality range corresponding to the production line-level control objective after the matching results of the production line-level control objective are corrected. The matching results of the production line-level control objective are associated and encapsulated with the applicable operating conditions and effective boundaries of the strategy to generate a cloud-optimized strategy track in the same control cycle as the edge real-time control track.

[0043] It should be noted that the applicable operating conditions of the strategy refer to the range of operating conditions that the cloud-optimized strategy track can participate in controlling, which is used to indicate that the cloud-optimized strategy track is suitable for the corresponding current operating condition phase. The policy effective boundary refers to the control range and security restrictions that the cloud-optimized policy track is allowed to operate when it participates in control. It is used to indicate whether the cloud-optimized policy track can cover the security constraint boundaries in the edge real-time control window.

[0044] S3. Based on the cloud-optimized strategy track, read the applicable working conditions and effective boundaries of the strategy, and combine the current working condition phase and safety constraint boundaries in the edge real-time control window to determine the strategy access status, form the strategy access status, and perform dual-track difference locking on the edge real-time control track and the cloud-optimized strategy track for current period access according to the strategy access status. Then, determine the takeover method based on the difference level and control time limit margin to obtain the control takeover method.

[0045] S3.1: Read the applicable operating conditions and effective boundaries of the policy from the cloud-based optimization policy track, and read the current operating condition phase and safety constraint boundaries from the edge real-time control window to obtain the policy access evidence body; Specifically, using the cloud-based optimized policy track and the edge real-time control window as the basis for organization, the applicable operating conditions and policy effectiveness boundaries of the policy are determined from the cloud-based optimized policy track, and the current operating condition phase and safety constraint boundaries are determined from the edge real-time control window. The applicable operating conditions of the policy are mapped to the current operating condition phase, so that the applicable scope corresponding to the cloud-based optimized policy track can be associated with the actual operating condition phase within the current control cycle. The policy effectiveness boundaries are mapped to the safety constraint boundaries, so that the effective scope corresponding to the cloud-based optimized policy track can be associated with the safety constraint boundaries within the current control cycle. The applicable operating conditions of the policy, the policy effectiveness boundaries, the current operating condition phase, and the safety constraint boundaries are associated and encapsulated to obtain the policy access evidence body.

[0046] S3.2: Based on the policy access evidence body verification, verify the working condition phase consistency and boundary coverage consistency between the cloud-optimized policy track and the edge real-time control window to obtain the policy boundary hit status; Specifically, using the policy access evidence body as the verification basis, the consistency between the policy application condition and the current condition phase is verified to determine the condition phase consistency status; when the condition phase consistency status is consistent, the coverage consistency between the policy effective boundary and the safety constraint boundary is verified to determine the number of boundary items covered by the policy effective boundary and the number of boundary items contained in the safety constraint boundary; based on the condition phase consistency status, the number of boundary items covered by the policy effective boundary and the number of boundary items contained in the safety constraint boundary, the policy boundary hit value corresponding to the policy access evidence body is calculated; when the policy... The policy boundary hit value indicates that the applicable operating condition of the policy is in phase with the current operating condition and the policy effective boundary completely covers the safety constraint boundary. In this case, the cloud-optimized policy track is marked as a fully hit state. When the policy boundary hit value indicates that the applicable operating condition of the policy is in phase with the current operating condition and the policy effective boundary does not completely cover the safety constraint boundary, the cloud-optimized policy track is marked as a partially hit state. When the policy boundary hit value indicates that the applicable operating condition of the policy is not in phase with the current operating condition, the cloud-optimized policy track is marked as a miss state. Finally, the fully hit state, partially hit state, and miss state are taken as the policy boundary hit state.

[0047] The policy boundary hit value corresponding to the policy admission evidence body is calculated using the following expression: ; in, Indicates the first The policy boundary hit value corresponding to each policy admission evidence body; Indicates the sequence number of the policy access evidence body; Indicates the first The phase consistency state of the operating condition corresponding to each strategy admission evidence body is taken when the phase of the applicable operating condition of the strategy is consistent with the current operating condition. When the applicable operating condition of the strategy is out of phase with the current operating condition, take... ; Indicates the first The number of boundary items in a policy access evidence body that cover the security constraint boundary with the policy effective boundary; Indicates the first The number of boundary terms contained in the security constraint boundary of each policy access evidence body.

[0048] It should be noted that before calculating the policy boundary hit value, duplicate boundary items in the policy effective boundary and security constraint boundary are deduplicated. When the same boundary item has conflicting constraint content in the policy effective boundary and security constraint boundary, the corresponding policy admission evidence body is marked as a boundary pending verification state, and the cloud-optimized policy track is suspended from participating in the admission determination of the current control cycle. When the number of boundary items contained in the security constraint boundary is 0, the cloud-optimized policy track is determined to be policy frozen. When the number of boundary items covered by the policy effective boundary is greater than the number of boundary items contained in the security constraint boundary, the number of boundary items covered by the policy effective boundary is taken as the number of boundary items contained in the security constraint boundary, and the corresponding situation is recorded in the boundary configuration verification record.

[0049] Boundary items are constraint items formed by breaking down the policy effective boundary and safety constraint boundary according to the range of action, speed range, position range, and safety limit; the number of boundary items is the number of constraint items; the number of boundary items that the policy effective boundary covers the safety constraint boundary is the number of constraint items in the policy effective boundary that can satisfy the corresponding constraint items of the safety constraint boundary; the number of boundary items contained in the safety constraint boundary is the total number of constraint items in the safety constraint boundary.

[0050] ; It should be noted that the policy boundary hit value ranges from 0 to 1. A policy boundary hit value of 1 indicates that the operating phase between the cloud-optimized policy track and the edge real-time control window is consistent, and the policy effective boundary completely covers the safety constraint boundary. A policy boundary hit value greater than 0 and less than 1 indicates that the operating phase between the cloud-optimized policy track and the edge real-time control window is consistent, but the policy effective boundary does not completely cover the safety constraint boundary. A policy boundary hit value of 0 corresponds to two situations: inconsistent operating phase and consistent operating phase but the policy effective boundary does not cover the boundary items in the safety constraint boundary.

[0051] S3.3: Based on the policy boundary hit status, determine the current period admission, adjacent period delay, and policy freeze status of the cloud optimization policy track to form the policy admission status; Specifically, the status of the cloud-optimized policy track entering the current control cycle is determined based on the policy boundary hit status. When the policy boundary hit status is a full hit, it indicates that the operating phase between the cloud-optimized policy track and the edge real-time control window is consistent, and the policy effective boundary completely covers the safety constraint boundary, thus the cloud-optimized policy track is determined to be admitted in the current cycle. When the policy boundary hit status is a partial hit, it indicates that the operating phase between the cloud-optimized policy track and the edge real-time control window is consistent, but the policy effective boundary does not completely cover the safety constraint boundary, thus the cloud-optimized policy track is determined to be delayed in the adjacent cycle. When the policy boundary hit status is a miss, it indicates that the operating phase between the cloud-optimized policy track and the edge real-time control window is inconsistent, or the operating phase is consistent but the policy effective boundary does not cover the boundary items in the safety constraint boundary, thus the cloud-optimized policy track is determined to be policy frozen. The current cycle admission, adjacent cycle delay, and policy freeze are taken as the policy admission status corresponding to the cloud-optimized policy track.

[0052] S3.4: Based on the policy admission status, filter the cloud-optimized policy track for admission in the current cycle. When the policy admission status hits any of the adjacent cycle's delayed state or the policy freeze state, use the edge real-time control track as the basis for takeover in the current control cycle. When the cloud-optimized policy track is the admission for the current cycle, place the cloud-optimized policy track and the edge real-time control track under the same control cycle to lock the consistency of control direction, control amplitude, instruction target, and edge control execution boundary, and obtain the dual-track difference locking result. Specifically, based on the policy access status, the cloud-optimized policy track is checked for status. When the policy access status matches either the adjacent cycle delay status or the policy freeze status, it is confirmed that the cloud-optimized policy track does not participate in the dual-track difference locking of the current control cycle, and the edge real-time control track is used as the basis for takeover in the current control cycle. When the cloud-optimized policy track is the current cycle's access, the cloud-optimized policy track and the edge real-time control track of the current cycle are mapped to the same control cycle. The control direction, control amplitude, and command target corresponding to the cloud-optimized policy track and the edge real-time control track of the current cycle are determined respectively, and combined with the edge control in the edge real-time control track... Consistent locking is performed on the control execution boundary; when the control direction points to the same action adjustment direction, a consistent locking result for the control direction is formed; when the control amplitude is within the edge control execution boundary, a consistent locking result for the control amplitude is formed; when the command object is the same industrial control object, a consistent locking result for the command object is formed; when the control direction, control amplitude, and command object all satisfy the edge control execution boundary, a consistent locking result for the edge control execution boundary is formed; the consistent locking results for the control direction, control amplitude, command object, and edge control execution boundary are correlated to obtain the dual-track differential locking result.

[0053] S3.5: Based on the results of dual-track difference locking, dual-track differences with consistent control direction and control amplitude within the edge control execution boundary are classified into low difference level, dual-track differences with consistent control direction and control amplitude touching the edge control execution boundary are classified into medium difference level, and dual-track differences with inconsistent control direction and inconsistent instruction action objects are classified into high difference level respectively, thus obtaining the difference level. Specifically, based on the dual-track difference locking results, the control direction consistency locking results and the command target consistency locking results are prioritized for verification. When the control directions are inconsistent, it indicates that the cloud-based optimization strategy track and the edge real-time control track have different adjustment directions for the same control cycle, and the corresponding dual-track difference is classified as a high difference level. When the command targets are inconsistent, it indicates that the cloud-based optimization strategy track and the edge real-time control track act on different industrial control objects, and the corresponding dual-track difference is classified as a high difference level. When the control directions are consistent and the command targets are consistent, the classification is based on the relationship between the control amplitude and the edge control execution boundary. If the control amplitude is within the edge control execution boundary... If the control amplitude is within the control range and does not reach the edge control execution boundary, the corresponding dual-track difference is classified into the low difference level. When the control direction is consistent and the target of the instruction is consistent, the difference is classified according to the relationship between the control amplitude and the upper and lower limits of the edge control execution boundary. When the control amplitude is strictly between the upper and lower limits of the edge control execution boundary, the corresponding dual-track difference is classified into the low difference level. When the control amplitude is equal to the upper limit and equal to the lower limit, the corresponding dual-track difference is considered to have reached the edge control execution boundary and is classified into the medium difference level. When the control amplitude exceeds the upper limit but is lower than the lower limit, the corresponding dual-track difference is classified into the high difference level, thus obtaining the difference level.

[0054] S3.6: The strategy admission state is used as a prerequisite for takeover, and the takeover hit determination is made in combination with the difference level and control time limit margin. When the strategy admission state hits any of the adjacent cycle delay state and strategy freeze state and the control time limit margin meets the current cycle execution requirements, the edge real-time control track continues to execute. When the difference level is low and the control time limit margin meets the current cycle execution requirements, the edge real-time control track continues to execute. When the difference level is medium and the control time limit margin meets the segment correction requirements, the cloud optimization strategy track participates in the correction. When the difference level is high, the end-side layered backoff constraint intervenes. When the control time limit margin does not meet the segment correction requirements, the end-side layered backoff constraint intervenes, thus obtaining the control takeover mode.

[0055] Specifically, after the policy access status and difference level are determined, the takeover conditions for the cloud-optimized policy track to participate in the current control cycle are judged. When the policy access status hits at least one of the adjacent cycle delay status and policy freeze status, it is confirmed that the cloud-optimized policy track does not participate in the current control cycle. The execution requirements of the current cycle are verified in conjunction with the control time limit margin. If the control time limit margin meets the execution requirements of the current cycle, the edge real-time control track continues to execute. If the control time limit margin does not meet the execution requirements of the current cycle, the edge-side layered rollback constraint is intervened. When the policy access status is current cycle access, the takeover hit determination is made based on the difference level and control time limit margin. If the difference level is... If the difference level is low and the control time limit margin meets the current cycle execution requirements, the edge real-time control rail is hit and execution continues, and the control direction, control amplitude, and instruction target corresponding to the cloud-optimized strategy rail are written into the candidate record of the adjacent cycle; if the difference level is medium and the control time limit margin meets the segment correction requirements, the cloud-optimized strategy rail is hit and participates in the correction; if the difference level is high, the end-side layered backoff constraint is hit; if the control time limit margin does not meet the segment correction requirements, the end-side layered backoff constraint is hit; the hit results of edge real-time control rail continuing execution, cloud-optimized strategy rail participating in correction, and end-side layered backoff constraint intervention are determined as the control takeover mode.

[0056] It should be noted that the adjacent cycle candidate records are used to save the content of cloud optimization strategy tracks that did not participate in the execution of the current control cycle under the low difference level, and serve as the candidate basis when generating cloud optimization strategy tracks in the next control cycle. The current cycle execution requirement is that the control time limit margin is not less than the sum of the edge command issuance time, the end-side action initiation time, and the minimum safety verification time. The edge command issuance time is determined based on the communication history of the edge real-time control track sending edge control commands to the end side. The end-side action initiation time is determined based on the execution log of the corresponding actuator of the industrial control object. The minimum safety verification time is determined based on the shortest stable time required for safety constraint boundary verification in the control cycle calibration record. When the policy admission state hits at least one of the adjacent cycle delayed state and policy frozen state, and the control time limit margin does not meet the current cycle execution requirements, the end-side hierarchical backoff constraint is directly triggered. The fragment correction requirement stipulates that the control time limit margin must not be less than the sum of the cloud policy writing time, fragment correction generation time, edge confirmation time, and security constraint verification time. The cloud policy writing time is determined based on the communication history of the cloud-optimized policy track being written to the edge side, the fragment correction generation time is determined based on historical fragment correction records, the edge confirmation time is determined based on the edge confirmation execution log, and the security constraint verification time is determined based on the control cycle calibration record. When the control time limit margin is equal to the sum of the corresponding fragment correction requirements, the cloud-optimized policy track is allowed to participate in the correction. When the control time limit margin is less than the sum of the corresponding fragment correction requirements, the cloud-optimized policy track is prohibited from participating in the correction, and the edge-side hierarchical rollback constraint is triggered.

[0057] It should be noted that the edge real-time control rail continues to execute edge real-time control to maintain the current control cycle, the cloud optimization strategy rail participates in the correction to correct the current control program segment when the control time limit margin meets the segment correction requirements, and the end-side hierarchical backoff constraint intervenes to take over the current control cycle when the difference level is high or the control time limit margin does not meet the segment correction requirements.

[0058] Figure 5 The chart shows the percentage distribution of three control takeover methods: "continuation of edge real-time control track," "participation of cloud-optimized strategy track in correction," and "intervention of edge-side hierarchical rollback constraints." As can be seen from the chart, the percentage of edge real-time control track continuing execution is 89.5%, indicating that in most control cycles, the cloud-optimized strategy track does not directly replace the edge real-time control track. Instead, it participates in correction only after the policy access status, difference level, and control time limit margin meet the judgment conditions. The percentage of cloud-optimized strategy track participating in correction is 1.8%, indicating that the cloud-optimized strategy track only enters segment correction when the current cycle's access, difference level, and control time limit margin requirements are met. The percentage of edge-side hierarchical rollback constraints intervening is 8.7%, indicating that edge-side hierarchical rollback constraints can enter the takeover chain when the difference level does not meet the takeover requirements or the control time limit margin is insufficient.

[0059] S4. Based on the control takeover method, lock the control program segment corresponding to the current operating condition phase, and combine the segment start boundary, segment end boundary and safety constraint boundary to confirm the segment takeover range. Based on the segment takeover range, lock the correction range of the current control program segment, and embed the end-side layered backoff constraint into the segment connection process to obtain the control execution command.

[0060] S4.1: Based on the control takeover method, read the current operating condition phase in the edge real-time control window, retrieve the control program segment corresponding to the current operating condition phase from the control task event carrier, match the control program segment with the segment execution order in the edge real-time control track, and obtain the phase segment index. Specifically, after the control takeover method is determined, the hit results of the edge real-time control rail continuing to execute, the cloud optimization strategy rail participating in correction, and the end-side hierarchical backoff constraint intervention corresponding to the control takeover method are confirmed, and the current operating condition phase is determined from the edge real-time control window; using the current operating condition phase as the segment matching condition, the control program segment corresponding to the current operating condition phase is retrieved from the control task event carrier; the control program segment is matched with the segment execution order in the edge real-time control rail to determine the execution order of the control program segment within the current control cycle; the current operating condition phase, control program segment, segment execution order, and control takeover method are associated to obtain the phase segment index.

[0061] S4.2: Based on the phase segment index, read the segment start boundary and segment end boundary of the control program segment, and associate the segment start boundary and segment end boundary with the safety constraint boundary in the edge real-time control window to obtain the segment boundary set; Specifically, based on the phase segment index, the control program segment corresponding to the current operating condition phase, the segment execution order, and the control takeover method are determined, and the segment start boundary and segment end boundary are determined within the control program segment. The segment start boundary is then matched with the safety constraint boundary in the edge real-time control window to confirm whether the start takeover position of the control program segment falls within the coverage of the safety constraint boundary. Similarly, the segment end boundary is matched with the safety constraint boundary in the edge real-time control window to confirm whether the end takeover position of the control program segment falls within the coverage of the safety constraint boundary. When both the segment start boundary and the segment end boundary fall within the coverage of the safety constraint boundary, the entire control program segment is considered as the segment interval covered by the safety constraint boundary. When the segment start boundary falls within the coverage of the safety constraint boundary but the segment end boundary does not, the segment interval between the segment start boundary and the end position of the safety constraint boundary is considered as the segment interval covered by the safety constraint boundary. When the boundary does not fall within the coverage area of ​​the safety constraint boundary but the segment termination boundary does, the segment interval between the start position of the safety constraint boundary and the segment termination boundary is taken as the segment interval covered by the safety constraint boundary. When neither the segment start boundary nor the segment termination boundary falls within the coverage area of ​​the safety constraint boundary, and the coverage area of ​​the safety constraint boundary is located between the segment start boundary and the segment termination boundary, the control program segment interval corresponding to the coverage area of ​​the safety constraint boundary is taken as the segment interval covered by the safety constraint boundary. When neither the segment start boundary nor the segment termination boundary falls within the coverage area of ​​the safety constraint boundary, and the coverage area of ​​the safety constraint boundary is not located between the segment start boundary and the segment termination boundary, no segment interval covered by the safety constraint boundary is formed. The current operating condition phase, control program segment, segment execution order, control takeover method, segment start boundary, segment termination boundary, safety constraint boundary, and segment interval covered by the safety constraint boundary are associated and encapsulated to obtain a segment boundary set.

[0062] S4.3: Based on the fragment boundary set, the takeover range of the control program fragment is locked by boundary overlap, and the fragment range covered by the safety constraint boundary is taken as the allowable takeover range, thus obtaining the fragment takeover range; Specifically, the manageable interval of a control program segment is determined based on the segment start and end boundaries in the segment boundary set. The manageable interval is then overlapped with the segment interval covered by the safety constraint boundaries in the segment boundary set for boundary locking. When the segment interval covered by the safety constraint boundary completely falls within the manageable interval, the segment interval covered by the safety constraint boundary is determined as the allowed takeover interval. When the segment interval covered by the safety constraint boundary partially overlaps with the manageable interval, the overlapping portion is determined as the allowed takeover interval. When the segment interval covered by the safety constraint boundary does not overlap with the manageable interval, the overlapping portion is determined as the allowed takeover interval. When takeover intervals overlap, no allowable takeover interval is formed; when no allowable takeover interval is formed, the generation of fragment correction range stops, and fragment correction range locking is no longer based on allowable takeover intervals; the control takeover mode is mapped to end-side hierarchical backoff constraint intervention, and end-side backoff instructions are directly generated based on end-side hierarchical backoff constraints. End-side backoff instructions are used to make the current control program segment enter the corresponding state among safe shutdown state, maintain the current position state, withdraw the safe position state, and wait for manual confirmation state; the control program segment, segment execution order, control takeover mode, and allowable takeover interval are associated and encapsulated to obtain the fragment takeover range.

[0063] S4.4: Based on the segment takeover range and control takeover method, lock the correction range of the current control program segment within the segment takeover range to obtain the segment correction range; Specifically, the control program segment, allowed takeover interval, and control takeover method within the segment takeover range are determined. When the control takeover method is for the edge real-time control rail to continue execution, the control program segment interval corresponding to the edge real-time control rail within the allowed takeover interval is locked as the correction range of the current control program segment to maintain the continued execution of the edge real-time control rail within the current control cycle. When the control takeover method involves the cloud-optimized strategy rail participating in the correction, the control program segment interval associated with the control direction, control amplitude, and instruction target corresponding to the cloud-optimized strategy rail within the allowed takeover interval is locked as the correction range of the current control program segment to limit the scope of the cloud-optimized strategy rail's participation in the correction. When the control takeover method involves the intervention of end-side hierarchical backoff constraints, the control program segment interval corresponding to the end-side hierarchical backoff constraints within the allowed takeover interval is locked as the correction range of the current control program segment to limit the segment range of the end-side hierarchical backoff constraints. The control program segment, allowed takeover interval, control takeover method, and correction range of the current control program segment are associated and encapsulated to obtain the segment correction range.

[0064] Figure 6The diagram illustrates the positional relationship between the control program segment, the coverage area of ​​the safety constraint boundary, and the segment correction range. Solid lines represent control program segments, dashed lines represent the coverage area of ​​the safety constraint boundary, and dotted lines represent the segment correction range of this invention. As can be seen from the diagram, the segment correction range does not cover the entire control program segment but is limited to the coverage area of ​​the safety constraint boundary and its manageable interval. This indicates that after determining the control takeover method, this invention further combines the segment start boundary, segment end boundary, and safety constraint boundary to perform boundary overlap locking on the manageable interval, thereby limiting the actual correction range.

[0065] S4.5: Based on the segment correction range, write the end-side hierarchical backoff constraint in the control task event carrier into the segment connection process of the current control program segment, and encapsulate the edge control instructions, segment execution order, segment correction range and end-side hierarchical backoff constraint in the edge real-time control track to obtain the control execution instructions.

[0066] Specifically, the current control program segment and allowed takeover interval corresponding to the segment correction range are determined, and the edge-side hierarchical rollback constraints are retrieved from the control task event carrier; the edge-side hierarchical rollback constraints are written into the segment connection process of the current control program segment within the allowed takeover interval, so that the segment connection process switches to the edge-side hierarchical rollback constraints when the current control program segment does not meet the control objective, allowed action boundary, and expected execution duration of the control program segment; the edge control instructions and segment execution order in the edge real-time control rail are determined, and the edge control instructions and segment execution order are mapped to the segment correction range; when the control takeover method is cloud-optimized strategy rail participating in the correction, the data is read... The control direction, control amplitude, and instruction target are retrieved from the cloud-based optimization strategy track. These parameters are then written into the segment correction content of the control execution instruction, and the corresponding edge control instruction content is replaced within the segment correction range. When the control takeover method is to continue execution of the edge real-time control track, the edge control instruction content is retained. When the control takeover method is to intervene with edge-side hierarchical backoff constraints, the edge-side hierarchical backoff constraints are written into the control execution instruction. The edge control instruction, segment execution order, segment correction range, segment correction content, and edge-side hierarchical backoff constraints are associated and encapsulated to obtain the control execution instruction.

[0067] It should be noted that the segment start boundary is the position where the control program segment begins execution or takes over within the current control cycle; The segment termination boundary is the position where the control program segment ends execution or terminates takeover within the current control cycle; The coverage of safety constraint boundaries refers to the range of motion, speed, position, and corresponding safety limits that the safety constraint boundaries can cover within the control program segment. The starting position of the safety constraint boundary is the position where the coverage of the safety constraint boundary begins to be covered in the control program segment; The termination position of the safety constraint boundary is the position where the coverage of the safety constraint boundary ends within the control program segment. The segment range covered by the safety constraint boundary is the segment range of the control program segment that falls within the coverage area of ​​the safety constraint boundary; The takeoverable section is the section of the control program segment that is defined by the segment start boundary and the segment end boundary and is capable of performing takeover processing. The takeover-permitted interval is the segment interval within the takeover-permitted interval that overlaps with the segment interval covered by the safety constraint boundary and is allowed to be taken over. The scope of segment takeover is the takeover processing range formed by associating the control program segment, the segment execution order, the control takeover method, and the allowed takeover intervals. The fragment correction range is the scope of the current control program fragment correction effect locked according to the control takeover method within the fragment takeover range.

[0068] S5. Drive the industrial control object to perform real-time actions according to the control execution command, and perform correlation calibration of the end-side execution feedback, policy access status and control takeover mode to form a cloud-edge-end collaborative calibration record.

[0069] S5.1: Drive industrial control objects to perform real-time actions according to the segment execution sequence by control execution instructions, and collect the execution completion status, post-execution measurement status, actuator feedback status and safety constraint verification status to form end-side execution feedback; Specifically, the execution order of the current control program segments is determined according to the execution sequence of the control execution instructions, and the edge control instructions are applied to the corresponding industrial control objects, enabling the industrial control objects to perform real-time actions within the segment correction range. During the execution of the current control program segment, it is confirmed whether the real-time action has reached the end position corresponding to the current control program segment based on the segment execution sequence, forming an execution completion state. After the real-time action is executed, the operating status and action results of the industrial control object corresponding to the action range, speed range, position range, and safety limits are collected to form the post-execution measurement state. At the same time, the action response returned by the actuator is collected to form the actuator feedback state. The action results corresponding to the post-execution measurement state are compared with the action range in the safety constraint boundary. The system performs a range check, comparing the speed state corresponding to the post-execution measurement state with the speed range in the safety constraint boundary, the position state corresponding to the post-execution measurement state with the position range in the safety constraint boundary, the safety-related states corresponding to the post-execution measurement state with the safety limits in the safety constraint boundary, and the consistency check between the actuator feedback state and the post-execution measurement state. When all checks meet the safety constraint boundaries, a "check passed" safety constraint check state is formed; when any check fails to meet the safety constraint boundaries, a "check failed" safety constraint check state is formed. The execution completion state, post-execution measurement state, actuator feedback state, and safety constraint check state are associated and encapsulated to form the edge execution feedback.

[0070] S5.2: Based on the execution feedback and control execution instructions, perform execution consistency verification under the same control cycle, and associate the verified execution feedback with the policy admission status and control takeover mode to form execution association calibration results; Specifically, the process involves mapping the edge execution feedback and control execution commands to the same control cycle, verifying the execution completion status against the execution sequence of segments in the control execution commands to confirm whether the industrial controlled object has reached the end position corresponding to the current control program segment according to the segment execution sequence; verifying the post-execution measurement status and actuator feedback status against the edge control commands in the control execution commands to confirm whether the action result of the industrial controlled object and the action response returned by the actuator correspond to the edge control commands; verifying the action result corresponding to the post-execution measurement status against the segment correction range in the control execution commands to confirm whether the real-time action falls within the segment correction range, and verifying whether the real-time action meets the safety constraint boundary in conjunction with the safety constraint review status; after verification, associating the verified edge execution feedback with the policy access status to confirm the access status of the cloud-optimized policy track corresponding to the edge execution feedback, associating the verified edge execution feedback with the control takeover method to confirm the hit results in the edge real-time control track continuing execution, the cloud-optimized policy track participating in correction, and the edge layered backoff constraint intervention corresponding to the edge execution feedback; and associating and encapsulating the verified edge execution feedback, policy access status, and control takeover method to form the execution association calibration result.

[0071] S5.3: Based on the execution correlation calibration results, mark the edge execution deviation, policy access deviation, and takeover method deviation, and encapsulate the control execution command, edge execution feedback, policy access status, control takeover method, and deviation mark to form a cloud-edge-edge collaborative calibration record.

[0072] Specifically, deviations are marked based on the verified edge execution feedback, policy access status, and control takeover method in the execution correlation calibration results. The verified edge execution feedback determines whether the control execution commands are executed within the same control cycle according to the segment execution order, segment correction range, and safety constraint boundaries. An edge execution deviation is marked when any of the following occurs: the execution completion status has not reached the end position corresponding to the current control program segment, the post-execution measurement status has not fallen within the segment correction range, or the safety constraint review status is "review failed." Access deviation is determined based on the correlation between edge execution feedback and policy access status. If the policy access status is "access in the current cycle" but the edge execution feedback does not reflect the action result after the cloud-optimized policy track participates in the correction, or if the policy access... If the state is either the adjacent cycle delay state or the strategy freeze state, but the edge execution feedback shows that the cloud-optimized strategy track participates in the current control cycle, a strategy admission deviation is marked. Takeover deviation is determined based on the correlation between edge execution feedback and control takeover mode. When the control takeover mode hits one of the following results: edge real-time control track continues execution, cloud-optimized strategy track participates in correction, or edge layered backoff constraint intervention, but the edge execution feedback does not show the corresponding action state under the hit result, a takeover mode deviation is marked. Edge execution deviation, strategy admission deviation, and takeover mode deviation are summarized into a deviation marker. The control execution command, edge execution feedback, strategy admission state, control takeover mode, and deviation marker are then associated and encapsulated to form a cloud-edge-device collaborative calibration record.

[0073] It should be noted that the execution completion status is the end confirmation status formed after the industrial control object executes the current control program segment in the segment execution order; The post-execution measurement status is the real-time data collected after the execution of the action, which shows the operating status and action results of the industrial control object. The actuator feedback status is the action response status returned by the actuator during the real-time action execution process; The safety constraint verification status is the status formed after verifying the post-execution measurement status and the actuator feedback status against the safety constraint boundary. The safety constraint review status that has passed the review is the state formed when the action result, speed status, position status, operating status corresponding to the safety limit, and actuator feedback status all meet the safety constraint boundaries. The safety constraint review status that fails the review is the status formed when there are situations in the action result, speed status, position status, operating status corresponding to the safety limit, and actuator feedback status that do not meet the safety constraint boundary. The associated calibration result is the calibration result formed by associating and encapsulating the verified edge execution feedback, policy access status, and control takeover mode; the edge execution deviation is the deviation formed when the control execution instructions are not executed in the same control cycle according to the segment execution order, segment correction range, and safety constraint boundary. Policy admission bias is the deviation that occurs when the correspondence between the edge execution feedback and the policy admission status is inconsistent. The takeover method deviation is the deviation that occurs when the correspondence between the end-side execution feedback and the control takeover method is inconsistent. The deviation marker is a marker formed by summarizing the end-side execution deviation, strategy access deviation, and takeover method deviation.

[0074] This embodiment also provides a computer device applicable to the industrial intelligent control method based on cloud-edge-device collaboration, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the industrial intelligent control method based on cloud-edge-device collaboration proposed in the above embodiment.

[0075] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0076] This embodiment also provides a storage medium storing a computer program. When executed by a processor, the program implements the cloud-edge-device collaborative industrial intelligent control method proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0077] In summary, this invention achieves the adaptation and confirmation between the cloud-optimized strategy track and the current operating condition phase and safety constraint boundaries before entering the control cycle by forming a strategy admission state; by locking the edge real-time control track and the cloud-optimized strategy track for the current cycle admission with dual-track differences, and determining the control takeover mode based on the difference level and control time limit margin, it realizes the orderly takeover between cloud strategy, edge real-time control and end-side layered backoff in the industrial control system, so that control execution commands can be generated under the condition of meeting safety constraint boundaries and control time limit requirements, and provides a clear basis for cloud-edge-end collaborative calibration records.

[0078] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An industrial intelligent control method based on cloud-edge-device collaboration, characterized in that, include: Based on industrial control objects, cloud-edge-device collaborative control tasks are established, and control program fragments and edge-side hierarchical backoff constraints are bound together. By acquiring edge-side status and verifying edge verification, the operating condition phase, safety constraint boundaries and control time limit margin are identified to obtain the edge real-time control window. Extract the edge real-time control basis from the edge real-time control window, and match the window status summary with the historical working condition control records and production line-level control targets using cloud strategies to generate edge real-time control tracks and cloud optimization strategy tracks. Based on the cloud-optimized strategy track, the applicable working conditions and effective boundaries of the strategy are read. Combined with the current working condition phase and safety constraint boundary in the edge real-time control window, the strategy access status is determined and a strategy access status is formed. Based on the strategy access status, the edge real-time control track and the cloud-optimized strategy track for current period access are locked with dual-track difference. Based on the difference level and control time limit margin, the takeover method is determined and the control takeover method is obtained. Based on the control takeover method, the control program segment corresponding to the current operating condition phase is locked. Combined with the segment start boundary, segment end boundary and safety constraint boundary, the segment takeover range is confirmed. Based on the segment takeover range, the correction range of the current control program segment is locked. The end-side layered backoff constraint is embedded into the segment connection process to obtain the control execution command. The system drives industrial controlled objects to perform real-time actions based on control execution commands, and correlates and calibrates edge execution feedback, policy access status, and control takeover methods to form a cloud-edge-device collaborative calibration record.

2. The industrial intelligent control method based on cloud-edge-device collaboration as described in claim 1, characterized in that, The steps to obtain the edge real-time control window are as follows: Taking the control objective of the industrial control object as the starting point of the task, the process cycle is mapped to the control cycle, the allowed action boundary and the expected execution time of the control program segment are bound to the control program segment, and the end-side hierarchical backoff constraint is written into the failure takeover condition of the control program segment to obtain the control task event carrier; The control task event carrier triggers the edge-side state acquisition, and performs edge verification on the sampling timing, execution feedback, available status of control program segments, and communication arrival status to form an edge-side trusted state carrier containing the available status of control program segments, edge-side communication latency, and edge verification time. The current operating condition phase is identified based on the trusted state carrier on the edge side, and the allowed action boundaries and process cycle times in the control task event carrier are read. The current operating condition phase is mapped to the allowed action boundaries to determine the safety constraint boundaries. The control time limit margin is determined based on the edge communication delay, edge verification time, expected execution time of the control program segment, and process cycle times, and an edge real-time control window is generated.

3. The industrial intelligent control method based on cloud-edge-device collaboration as described in claim 2, characterized in that, The steps for generating the edge real-time control track and the cloud optimization strategy track are as follows: Based on the edge real-time control window, the current operating condition phase is associated with the safety constraint boundary and control time limit margin, and the available state of the control program segment is incorporated into the association result to obtain the edge real-time control basis. Based on the edge real-time control criteria, edge control instructions are determined, mapped to the instruction target and the execution order of the segments, and the edge control execution boundary is encapsulated according to the safety constraint boundary and the control time limit margin to generate the edge real-time control track. The window status summary is generated using the edge real-time control window. The window status summary is matched with the historical operating condition control records. The matching result is then corrected according to the production line-level control objectives to a cloud-optimized strategy track that is in the same control cycle as the edge real-time control track.

4. The industrial intelligent control method based on cloud-edge-device collaboration as described in claim 1 or 3, characterized in that, The steps for establishing the policy admission state are as follows: Read the applicable operating conditions and effective boundaries of the policy from the cloud-optimized policy track, and read the current operating condition phase and safety constraint boundaries from the edge real-time control window to obtain the policy access evidence body; Based on the verification of the policy access evidence body, the working condition phase consistency and boundary coverage consistency between the cloud-optimized policy track and the edge real-time control window are obtained to obtain the policy boundary hit status. Based on the policy boundary hit status, the cloud-based optimization policy track is assessed for current cycle admission, adjacent cycle delay, and policy freeze status, thus forming the policy admission status.

5. The industrial intelligent control method based on cloud-edge-device collaboration as described in claim 4, characterized in that, The steps for obtaining control and takeover are as follows: The cloud-optimized policy track for the current cycle is selected based on the policy admission status. When the policy admission status hits either the adjacent cycle's delayed state or the policy freeze state, the edge real-time control track is used as the basis for takeover in the current control cycle. When the cloud-optimized policy track is the current cycle's admission, the cloud-optimized policy track and the edge real-time control track for the current cycle are placed in the same control cycle to lock the consistency of control direction, control amplitude, instruction target, and edge control execution boundary, thus obtaining the dual-track difference locking result. Based on the results of dual-track difference locking, dual-track differences with consistent control direction and control amplitude within the edge control execution boundary are classified into low difference level, dual-track differences with consistent control direction and control amplitude touching the edge control execution boundary are classified into medium difference level, and dual-track differences with inconsistent control direction and inconsistent instruction action objects are classified into high difference level, thus obtaining the difference level. The strategy admission status is used as a prerequisite for takeover, and the takeover hit determination is made in combination with the difference level and control time limit margin. When the strategy admission status hits any of the adjacent cycle delay status and strategy freeze status, and the control time limit margin meets the current cycle execution requirements, the edge real-time control track continues to execute. When the difference level is low and the control time limit margin meets the current cycle execution requirements, the edge real-time control track continues to execute. When the difference level is medium and the control time limit margin meets the segment correction requirements, the cloud optimization strategy track participates in correction. When the difference level is high, the end-side layered backoff constraint intervenes. When the control time limit margin does not meet the segment correction requirements, the end-side layered backoff constraint intervenes, thus obtaining the control takeover mode.

6. The industrial intelligent control method based on cloud-edge-device collaboration as described in claim 5, characterized in that, The steps for confirming the scope of the segment takeover are as follows: Based on the control takeover method, the current operating condition phase in the edge real-time control window is read, the control program segment corresponding to the current operating condition phase is retrieved from the control task event carrier, and the control program segment is matched with the segment execution order in the edge real-time control track to obtain the phase segment index. The control program segment start boundary and segment end boundary are read based on the phase segment index, and the segment start boundary and segment end boundary are associated with the safety constraint boundary in the edge real-time control window to obtain the segment boundary set; Based on the fragment boundary set, the takeover range of the control program fragment is locked by boundary overlap, and the fragment range covered by the safety constraint boundary is taken as the allowable takeover range, thus obtaining the fragment takeover range.

7. The industrial intelligent control method based on cloud-edge-device collaboration as described in claim 6, characterized in that, The steps to obtain the control execution instruction are as follows: Based on the segment takeover range and control takeover method, the correction range of the current control program segment is locked within the segment takeover range to obtain the segment correction range; Based on the segment correction range, the end-side hierarchical backoff constraints in the control task event carrier are written into the segment connection process of the current control program segment, and the edge control instructions, segment execution order, segment correction range and end-side hierarchical backoff constraints in the edge real-time control track are encapsulated to obtain the control execution instructions.

8. The industrial intelligent control method based on cloud-edge-device collaboration as described in claim 1 or 7, characterized in that, The steps for forming the cloud-edge-device collaborative calibration record are as follows: The system drives industrial control objects to perform real-time actions in the order of segment execution by control execution commands, and collects the execution completion status, post-execution measurement status, actuator feedback status, and safety constraint verification status to form end-side execution feedback. Based on the execution feedback from the end side and the control execution instructions, the execution consistency is verified under the same control cycle. The verified end side execution feedback is then associated with the policy access status and control takeover mode to form an execution association calibration result. Based on the execution correlation calibration results, mark the edge execution deviation, policy access deviation, and takeover method deviation, and encapsulate the control execution command, edge execution feedback, policy access status, control takeover method, and deviation mark to form a cloud-edge-device collaborative calibration record.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the industrial intelligent control method based on cloud-edge-device collaboration as described in any one of claims 1 to 8.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the industrial intelligent control method based on cloud-edge-device collaboration as described in any one of claims 1 to 8.