Virtualized orchestration system and method for industrial control

Abstract

This invention relates to the field of virtualization orchestration technology, specifically to a virtualization orchestration system and method for industrial control. The system includes a task extraction module, a node behavior monitoring module, a stability assessment module, an orchestration load adjustment module, and a control task migration module. This invention, by collecting control task scheduling mapping tables and execution instruction queues, and combining loop control flags and interrupt flags to identify closed-loop characteristics of scheduling cycles, can effectively identify sets of orchestratable tasks. It calculates behavior success rates based on parameters such as node response latency and task completion rates, and performs node stability screening to classify control nodes by stability level. By setting release thresholds to form a scheduling priority list, it determines task migration timing based on task dwell time and I / O completion status, and uses behavior pattern matching to accurately select migration targets, significantly improving the stability of task scheduling, the accuracy of resource allocation, and the adaptability of control logic.

Description

Technical Field

[0001] This invention relates to the field of virtualization orchestration technology, and more particularly to virtualization orchestration systems and methods for industrial control. Background Technology

[0002] Virtualization orchestration technology is an interdisciplinary field at the intersection of computer systems and industrial control systems. Specifically, it involves using virtualization technologies, such as virtual machines, containers, and virtual networks, to abstract and isolate computing, storage, and control resources. Through an orchestration engine, it enables centralized scheduling, automatic deployment, and runtime management of resources in multi-node, multi-task environments. The core objective of this field is to improve the resource utilization efficiency, deployment flexibility, and system maintainability of industrial control systems, while supporting rapid migration of control logic, fault recovery, and elastic expansion. It is suitable for industrial scenarios where multiple control strategies coexist and heterogeneous devices operate collaboratively. When applying virtualization orchestration solutions in industrial control systems, their adaptability to real-time performance, determinism, and security must be considered; therefore, it is necessary to integrate soft real-time scheduling strategies, distributed communication mechanisms, and edge computing capabilities.

[0003] The virtualization orchestration system for industrial control is primarily used to migrate the industrial control processes required by mining equipment to a virtualization-based operating environment. Through unified management and scheduling via the orchestration system, it enables flexible deployment, modular reconfiguration, and fault isolation of control tasks. By virtualizing and encapsulating control functional units, modeling task dependencies, mapping scheduling strategies, and monitoring runtime states, the system can address control logic failures caused by task switching, hardware malfunctions, or environmental disturbances during mining equipment operation, thereby improving system reliability, flexibility, and maintenance efficiency.

[0004] Traditional orchestration systems, when faced with the periodic fluctuations in control tasks and unstable node performance of mining equipment, fail to conduct refined analysis of task execution behavior. They rely solely on task dependency modeling and a unified scheduling mechanism to handle various control tasks, leading to ambiguity in task identification during periodic switching. Node performance evaluation remains static and fails to reflect multi-period behavioral trends, easily causing misjudgments in node allocation. The scheduling strategy lacks a mechanism for quantifying remaining resources, which can easily result in uneven resource load, causing continuous overload of high-frequency scheduling nodes. Furthermore, the lack of judgment criteria such as task dwell time and task behavior pattern matching leads to cooling imbalances and inaccurate target selection during migration execution. During continuous equipment operation, this can easily result in adverse consequences such as task loss of control, logical anomalies, and delayed fault recovery. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a virtualization orchestration system and method for industrial control.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: a virtualization orchestration system for industrial control, the system comprising:

[0007] The task extraction module obtains the virtual control container configuration item in the driving logic of mining equipment, calls the loop control flag and interrupt flag in the task tag field, determines the number of control tasks with closed-loop characteristics in the continuous scheduling cycle, and obtains the set of programmable control tasks.

[0008] The node behavior monitoring module collects the control response delay time, task completion count value and total number of planned tasks of the mining equipment control unit node according to the node number bound to each task in the programmable control task set, and filters stable behavior nodes to obtain stable behavior node filtering results.

[0009] Based on the stable behavior node screening results, the stability assessment module detects the changing trend and extracts the changing value according to the task operation data of the mining equipment, calculates the command response stability score, classifies the node stability level, and generates the virtual node stability grading result.

[0010] The orchestration load adjustment module calls the node level field and task allocation field in the virtual node stability grading result, collects the CPU utilization and memory usage of the scheduled tasks in the virtual control of mining equipment, calculates the remaining resource value, extracts the task number of the target number, and obtains the control task priority adjustment list.

[0011] The present invention is improved in that the programmable control task set includes a scheduling task identifier, a closed-loop execution label, and a scheduling cycle number; the stable behavior node screening result specifically refers to the task behavior success rate value, node marking status, and response duration record; the virtual node stability grading result includes the node grading number, behavior duration level, and stability difference; and the control task priority adjustment list specifically includes the task scheduling number, resource remaining assessment value, and load classification label.

[0012] The present invention is improved in that the task extraction module includes:

[0013] The configuration reading submodule obtains the task scheduling mapping table, task execution sequence, and bound executor instruction queue from the virtual control container configuration item in the driving logic of mining equipment. It determines whether there is a continuous combination of scheduling number field, mapping path field, and bound control instruction field in the task tag field of the task execution sequence, filters task configuration records with scheduling integrity, and generates the number of task structure identification tags.

[0014] The task structure identification submodule collects the executor number, scheduling sequence code and execution feedback number of the task according to the scheduling task record corresponding to the number of task structure identification tags, determines whether the number of feedbacks is consistent with the number of sequence codes, filters the set of task numbers with complete instruction structure, and obtains the number of closed-loop feature tasks.

[0015] The task closure judgment submodule is based on the set of task numbers corresponding to the number of tasks with closure characteristics. It counts the number of times each task number appears in a continuous scheduling cycle and determines whether it is greater than the continuous scheduling cycle judgment value. It then filters the task numbers that meet the judgment value to form a set and obtains a set of programmable control tasks.

[0016] The present invention is improved in that the node behavior monitoring module includes:

[0017] The response data acquisition submodule obtains the control unit node number bound to each task in the programmable control task set, collects the control response time record, the completed task count log and the total number of tasks in the task plan table in the current scheduling cycle within each node, determines whether there are valid task records in the task completion count log, and obtains the control task execution quantity information.

[0018] The behavior success rate calculation submodule is based on the total number of planned tasks corresponding to the task records in the control task execution quantity information, calls the corresponding control response time and calculates the average response delay, combines the number of completed tasks with the number of planned tasks to perform normalization, and calculates and obtains the task behavior success ratio.

[0019] The behavior effectiveness screening submodule compares the behavior success ratio with the behavior persistence assessment threshold to screen a set of node numbers with periodic and continuous successful behaviors, and obtains the stable behavior node screening results.

[0020] The present invention is improved in that the stability evaluation module includes:

[0021] The behavior sequence acquisition submodule obtains the node number involved in the stable behavior node screening result, collects the control behavior data sequence of the corresponding node in multiple scheduling cycles, including the number of consecutive successes, the number of control instruction loss markers and the scheduling response delay time value in each cycle, and aggregates the data according to the node number to obtain the periodic behavior continuity index information.

[0022] The trend extraction calculation submodule calculates and obtains the command response stability score by calling the number of consecutive successes per cycle, the control command loss rate, and the average response delay time of the control behavior sequence corresponding to the node of the periodic behavior continuity index information.

[0023] The node level classification submodule determines the level of each node's score range based on the stability score of the instruction response and the level classification boundary set in the stability evaluation standard, identifies the node with a level label, and obtains the virtual node stability classification result.

[0024] The present invention is improved in that the orchestration load adjustment module includes:

[0025] The node parameter calling submodule obtains the node level field and task allocation field from the virtual node stability grading result, collects the CPU utilization and memory usage data corresponding to the current task, obtains the available CPU capacity and remaining memory capacity of the scheduling node where the task is located, establishes a resource mapping index table between tasks and nodes, and generates a task resource mapping index set.

[0026] The resource comparison calculation submodule extracts the current CPU utilization rate and memory usage of the task and the available CPU capacity and remaining memory capacity of the node based on the resource consumption data and node remaining resource value corresponding to each task record in the task resource mapping index set. It calculates and obtains the task resource remaining rate matching value, determines whether the matching value is greater than the resource scheduling release threshold, filters the number of tasks that meet the conditions, and generates the resource remaining rate matching result.

[0027] The task priority filtering submodule sorts the tasks in the resource remaining rate matching results from high to low according to the matching value, extracts the target number of task numbers as a high priority task candidate set, and marks the task type field and node level field to obtain a control task priority adjustment list.

[0028] The present invention has an improvement, wherein the system further includes:

[0029] The control task migration module adjusts the task information in the control task priority list according to the control task priority, determines whether the task dwell time is higher than the defined quiet period window, filters tasks that meet the scheduling cycle migration conditions, filters the migration target node number according to the task behavior pattern matching results in the most recent cycle, and generates a control task migration target allocation table.

[0030] The control task migration target allocation table includes the migrateable task number, the target node number, and the behavior pattern index.

[0031] The present invention is improved in that the control task migration module includes:

[0032] The task dwell time determination submodule obtains the task information in the control task priority adjustment list, collects the IO request completion time, task dwell time and scheduling cooldown flag of the node corresponding to each task, determines whether the task dwell time is greater than the silent period window baseline value, removes the task number whose scheduling cooldown flag is in the prohibited migration state, and generates a set of migrateable task numbers.

[0033] The behavior pattern matching submodule, based on the control behavior sequence of tasks in the most recent scheduling cycle in the set of transferable task numbers, calls the task control response trajectory, instruction loop feature label and feedback beat interval, determines whether the difference between the current behavior sequence and the corresponding behavior template of the previous cycle is less than the control behavior offset tolerance threshold, obtains the set of task numbers for stable behavior matching, and obtains the stable behavior pattern matching set.

[0034] The migration node filtering submodule stably matches the centralized task numbers according to the behavior pattern, extracts the resource schedulable status, behavior compatibility label and resource scheduling priority field from the list of currently available nodes, filters the node numbers with the same behavior label and the scheduling priority field is lower than the task priority level, establishes the correspondence between task number and target node, and obtains the control task migration target allocation table.

[0035] A virtualization orchestration method for industrial control, wherein the virtualization orchestration method for industrial control is used to implement the aforementioned virtualization orchestration system for industrial control, includes the following steps:

[0036] S1: Obtain the virtual control container configuration item in the driving logic of mining equipment, call the loop control flag and interrupt flag in the task tag field, determine the number of control tasks with closed-loop characteristics in the continuous scheduling cycle, and obtain the set of programmable control tasks.

[0037] S2: Based on the node number bound to each task in the programmable control task set, collect the control response delay time, task completion count value and total number of planned tasks of the mining equipment control unit node, filter stable behavior nodes, and obtain stable behavior node filtering results;

[0038] S3: Based on the stable behavior node screening results, according to the task operation data of the mining equipment, detect the changing trend during the week and extract the changing value, calculate the command response stability score, classify the node stability level, and generate the virtual node stability grading result.

[0039] S4: Call the node level field and task allocation field in the virtual node stability grading result, collect the CPU utilization and memory usage of the scheduled tasks in the virtual control of mining equipment, calculate the remaining resource value, extract the task number of the target number, and obtain the control task priority adjustment list.

[0040] S5: Based on the task information in the control task priority adjustment list, determine whether the task dwell time is higher than the defined quiet period window, filter tasks that meet the scheduling cycle migration conditions, filter the migration target node number based on the task's behavior pattern matching results in the most recent cycle, and generate a control task migration target allocation table.

[0041] Compared with the prior art, the advantages and positive effects of the present invention are as follows:

[0042] In this invention, by collecting the control task scheduling mapping table and the execution instruction queue, and combining the loop control flag and interrupt flag to identify the closed-loop characteristics of the scheduling cycle, it is possible to effectively identify the set of schedulable tasks. The success rate of behavior is calculated based on parameters such as node response latency and task completion ratio, and node stability is screened. Based on the changing trends of the number of consecutive successes, control instruction loss rate and scheduling latency within multiple cycles, and combined with the number of interrupt flags, a stability score is formed to classify the stability level of control nodes. A comparison mechanism between CPU utilization and memory remaining is introduced, and a scheduling priority list is formed by setting a release threshold. Then, the timing of task migration is determined based on task dwell time and IO completion status. The migration target is accurately selected by using behavior pattern matching, which significantly improves the stability of task scheduling, the accuracy of resource allocation and the adaptability of control logic, and enhances the overall system's operational reliability in scenarios with frequent task switching and node performance fluctuations. Attached Figure Description

[0043] Figure 1 This is a system flowchart of the present invention;

[0044] Figure 2 This is a flowchart of the task extraction module of the present invention;

[0045] Figure 3 This is a flowchart of the node behavior monitoring module of the present invention;

[0046] Figure 4 This is a flowchart of the stability evaluation module of the present invention;

[0047] Figure 5 This is a flowchart of the orchestration load adjustment module for the present invention;

[0048] Figure 6 This is a flowchart of the control task migration module of the present invention. Detailed Implementation

[0049] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0050] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0051] Please see Figure 1 The present invention provides a technical solution: a virtualized orchestration system for industrial control, the system including a task extraction module, a node behavior monitoring module, a stability assessment module, an orchestration load adjustment module, and a control task migration module;

[0052] The task extraction module obtains the configuration items of the virtual control container in the driving logic of mining equipment, collects the control task scheduling mapping table, task execution sequence and bound executor instruction queue, calls the loop control flag and interrupt flag in the task tag field, determines the number of control tasks with closed loop characteristics in the continuous scheduling cycle, and obtains the set of programmable control tasks.

[0053] The node behavior monitoring module collects the control response delay time, task completion count, and total number of planned tasks of the mining equipment control unit node based on the node number bound to each task in the programmable control task set. It then compares the differences to screen stable behavior nodes and obtains the stable behavior node screening results.

[0054] Control response latency refers to the time interval between the issuance of an instruction and the completion of execution feedback by a node. The task completion count is based on the actual number of successful executions recorded by the scheduler. The behavior success rate is the ratio of the task completion count to the total number of planned tasks.

[0055] Based on the stable behavior node screening results, the stability assessment module collects the control behavior duration sequence of each node in multiple cycles according to the task operation data of mining equipment, including the number of consecutive successes, control command loss rate, and average scheduling delay. It detects the changing trend during the cycle and extracts the changing value. Combined with the number of interruption markers in the cycle, it calculates the command response stability score, classifies the node stability level, and generates virtual node stability grading results.

[0056] Control behavior duration rate is the ratio of the number of successful executions to the total number of schedulings within a control cycle; control instruction loss rate is the ratio of the number of instructions that failed to respond in the scheduling sequence to the total number of instructions; average scheduling latency is the average scheduling to response time over multiple cycles; node load utilization rate is the ratio of the actual amount of resources occupied to the total amount of allocable resources.

[0057] The orchestration load adjustment module calls the node level field and task allocation field in the virtual node stability grading result, collects the CPU utilization and memory usage of the scheduled tasks in the virtual control of mining equipment, compares them with the available CPU capacity and remaining memory capacity of the scheduled node, calculates the remaining resource value, determines whether the remaining value is higher than the set scheduling release threshold, and obtains the control task priority adjustment list.

[0058] The control task migration module adjusts the task information in the list according to the priority of the control tasks, collects the IO request completion time, task dwell time and scheduling cooldown flag of the corresponding node of the task, determines whether the task dwell time is higher than the defined quiet period window, filters tasks that meet the migration conditions of the scheduling cycle, filters the migration target node number based on the behavior pattern matching results of the task in the most recent cycle, and generates a control task migration target allocation table.

[0059] The IO request completion time is the time taken for the last IO instruction to complete in a single task within the scheduling container; the scheduling cooldown flag indicates whether the pre-migration delay freeze window is met, and is set based on the container scheduling policy.

[0060] The set of programmable control tasks includes scheduling task identifiers, closed-loop execution labels, and scheduling cycle numbers. The stable behavior node screening results specifically refer to task behavior success rate values, node tag status, and response persistence records. The virtual node stability grading results include node grading numbers, behavior persistence levels, and stability differences. The control task priority adjustment list specifically includes task scheduling numbers, resource remaining assessment values, and load classification labels. The control task migration target allocation table includes migrateable task numbers, target node numbers, and behavior pattern indexes.

[0061] Please see Figure 2 The task extraction module includes:

[0062] The configuration reading submodule obtains the task scheduling mapping table, task execution sequence, and bound executor instruction queue from the virtual control container configuration item in the driving logic of mining equipment. It determines whether there is a continuous combination of scheduling number field, mapping path field, and bound control instruction field in the task tag field of the task execution sequence, filters task configuration records with scheduling integrity, and generates the number of task structure identification tags.

[0063] The configuration reading submodule first accesses the configuration items of the virtual control container of the tunneling machine (numbered ZC001) in a certain intelligent coal mining face. Specifically, it accesses the task scheduling mapping table stored at memory address 0x7FFF_E000, which records the correspondence between task tags, scheduling numbers, mapping paths, and bound control commands. For example, one record is {task tag: "tunneling machine front roller cutting", scheduling number: "DJ_A_01", mapping path: " / dev / roller / cut", bound control command: "CMD_CUT_START"}. Simultaneously, the configuration reading submodule reads the task execution sequence file located at address 0x7FFF_F000. This file contains a task list, such as ["tunneling machine front roller cutting", "hydraulic support following", "conveyor belt"]. [Start] Next, the configuration reading submodule searches and matches each task tag in the execution sequence in the task scheduling mapping table. Taking the "Tunneling Machine Front Roller Cutting" task as an example, the submodule successfully found its corresponding scheduling number "DJ_A_01", mapping path " / dev / roller / cut", and binding control command "CMD_CUT_START" in the mapping table. Since these three fields are consecutive and complete, it is determined that the task configuration record has scheduling integrity. Conversely, for the task "Conveyor Belt Start", the "Binding Control Command" field is missing when querying the mapping table, so the record is filtered out. After checking a total of 15 task tags one by one, 12 task configuration records with scheduling integrity are filtered out, and task structure identification tags are generated.

[0064] The task structure recognition submodule collects the executor number, scheduling sequence code and execution feedback number of the task according to the scheduling task record corresponding to the number of task structure recognition tags, determines whether the number of feedbacks is consistent with the number of sequence codes, filters the set of task numbers with complete instruction structure, and obtains the number of closed-loop feature tasks.

[0065] The task structure identification submodule processes the 12 task structure identification tags generated in the previous steps. Taking the task "Tunneling Machine Front Drum Cutting" (task number T001) as an example, the module collects the actuator number bound to it as ACT_R01 and obtains its scheduling sequence code from the task scheduling plan as a sequence of 4 steps [1,2,3,4]. This means that the complete execution of the task contains four consecutive sub-instructions. At the same time, the module queries the execution log database to count the number of execution feedback records associated with task T001 and actuator ACT_R01 in the most recent scheduling cycle. The query result is 4. Subsequently, the module judges whether the number of feedbacks of 4 is consistent with the number of sequence codes of 4. Since 4=4, it is determined that the instruction structure of task T001 is complete. The module repeats this process for all 12 records. Among them, the number of feedbacks of 9 records is consistent with the number of sequence codes. Thus, these 9 task numbers constitute a set of task numbers with complete instruction structures, and the number of closed-loop feature tasks is 9.

[0066] The task closure judgment submodule is based on the set of task numbers corresponding to the number of tasks with closure characteristics. It counts the number of times each task number appears in a continuous scheduling cycle and judges whether it is greater than the continuous scheduling cycle judgment value. It then filters the task numbers that meet the judgment value to form a set and obtains a set of programmable control tasks.

[0067] The task closure-loop judgment submodule receives a closure-loop characteristic task quantity information value of 9 and obtains a set of 9 corresponding task numbers {T001, T002, ..., T009}. To determine the programmability of the tasks, it is necessary to evaluate their stability in continuous scheduling cycles. Here, the setting of the continuous scheduling cycle judgment value is based on the analysis of the continuity requirements of key equipment control tasks (such as tunneling, support, and transportation) in coal mining faces. In an industrial Ethernet fieldbus simulation environment containing 50 nodes, 10 typical control tasks were continuously tested for up to 24 hours. Statistical analysis revealed that in events where control anomalies occurred, 92% of the abnormal tasks had fewer than 3 consecutive scheduling cycles, indicating that tasks that were executed for fewer than 3 consecutive cycles had high instability. Therefore, the continuous scheduling cycle judgment value was set to 3, and the submodule then... The submodule counts the occurrences of each task number in the most recent four consecutive scheduling cycles (cycles C1 to C4). For example, task T001 appears in C1, C2, C3, and C4, with a occurrence count of 4; task T002 appears in C1, C3, and C4, with a occurrence count of 3; and task T005 appears only in C2, with a occurrence count of 1. The submodule determines whether the occurrence count of each task is greater than or equal to the judgment value of 3. The occurrence count of task T001 is 4 > 3, which satisfies the condition; the occurrence count of task T002 is 3 = 3, which satisfies the condition; and the occurrence count of task T005 is 1 < 3, which does not satisfy the condition. After judging each of the nine tasks one by one, a total of five tasks satisfy this judgment condition. The submodule combines the numbers of these five tasks into a new set to obtain the set of programmable control tasks {T001, T002, T003, T004, T006}.

[0068] Please see Figure 3 The node behavior monitoring module includes:

[0069] The response data acquisition submodule obtains the control unit node number bound to each task in the programmable control task set, collects the control response time record, the completed task count log and the total number of tasks in the task plan table in the current scheduling cycle within each node, determines whether there are valid task records in the task completion count log, and obtains the control task execution quantity information.

[0070] The response data acquisition submodule, based on the programmable control task set {T001, T002, T003, T004, T006} obtained in the previous steps, first extracts the control unit node number currently bound to each task. Taking task T001 (tunneling machine front drum cutting) as an example, it is currently running on node N01. The submodule then collects data related to task T001 from the performance log of node N01. The collected control response time record is an array [42, 45, 44], in milliseconds (ms). At the same time, it queries the completed task count log of T001 in the current scheduling cycle from the task completion history, and finds that the number of completed tasks is 9. It also reads the total number of tasks assigned to T001 from the task plan table of the current scheduling cycle, which is 10. Finally, the submodule determines whether the 9 completed task count logs are valid task records, that is, whether the value is greater than 0. Since 9 > 0, it is determined to be valid, thus forming a complete record containing node number, response time, completion count, and total number of plans, and obtaining the control task execution quantity information.

[0071] The behavior success rate calculation submodule, based on the total number of planned tasks corresponding to the task records in the control task execution quantity information, calls the corresponding control response time and calculates the average response latency. It then performs normalization by combining the number of completed tasks with the number of planned tasks, using the following formula:

[0072] ;

[0073] Calculate and obtain the success rate of the task.

[0074] in, This represents the success rate of the task's actions. This represents the number of times a task has been executed within the current period, obtained by analyzing the task feedback log statistics in the scheduling record. The total number of scheduled tasks is obtained by counting the total number of plan numbers in the scheduling instruction table. This is a normalized value for the average control response latency of the current task, obtained by dividing the average response time of the task by the maximum allowable response time for the task category. This is the normalized average response latency value for all tasks within the task category. It is obtained by averaging the latency times of all tasks in this category and then normalizing the result. The normalized response delay maximum value constant is set to 1. The normalized value for task execution feedback frequency is obtained by dividing the number of feedbacks per unit time by the maximum acceptable value for feedback frequency. The behavior continuity assessment threshold is the threshold benchmark for judging the success status of task behavior, which is obtained based on the requirements for the timeliness and continuity of feedback in industrial-grade real-time control tasks.

[0075] The behavior success rate calculation submodule calculates the behavior success rate based on the collected control task execution quantity information, specifically for task T001 executed on node N01, to evaluate its behavior success ratio. The formula used here is:

[0076] ;

[0077] The innovation of this formula lies in the fact that it does not simply use task completion rate to evaluate success, but creatively integrates three dimensions. First, Partially representing the basic task execution completeness, secondly... As a penalty factor, this component quantifies the stability of task response latency; tasks that deviate more from the category average latency will have their success ratio reduced more significantly. The factor also serves as a reward, reflecting the timeliness of task feedback. Tasks with higher feedback frequencies experience a non-linear increase in their success rate. The combination of these three factors allows the evaluation results to not only reflect whether the task was completed but also to reveal its execution quality and real-time performance level. Regarding the parameters in the formula... It is the success rate of the task behavior to be calculated. This represents the number of times the task has been executed within the current period, and its value is 9, obtained from the scheduling records. The total number of scheduled tasks is 10, which is obtained from the scheduling instruction table. To calculate the normalized value of the average control response delay for the current task, the average response time is first calculated as follows: The maximum allowable response time for the task category is set based on the requirements for Real-time Ethernet communication (Real-time Class A) in the IEC 61784-2 standard, and combined with the safety margin for the torque control of the tunneling machine's cutting motor. It is set to 50ms. This value was determined by executing the same command 1000 times on a similar Siemens S7-1500 series PLC under standard load, with a response time of 48.2ms at the 99.9th percentile. This value was rounded up and a safety margin was added. Therefore... , To normalize the average response latency of all tasks within the same category (tunnel boring machine cutting), and setting the average response times of two other tasks of the same category in the system to 46ms and 41ms respectively, the average response time for the category is: After normalization , This is the constant representing the maximum normalized response delay, set to 1. The normalized value for the task execution feedback frequency is set as follows: if the current scheduling cycle length is 1 second, and the task provides feedback 9 times within 1 second, the feedback frequency is 9Hz. The maximum acceptable value for the feedback frequency is set based on the analysis of the mechanical dynamic characteristics of the controlled object (cutting drum). To effectively suppress low-frequency vibrations below 5Hz, according to the Nyquist theorem, the feedback frequency should be at least 10Hz. Therefore, this value is set to 10Hz. Substitute the above parameter values ​​into the formula to perform the calculation:

[0078] ;

[0079] The success rate of the task T001 was 0.2393.

[0080] The behavior effectiveness screening submodule compares the behavior success ratio with the behavior persistence assessment threshold to screen the set of node numbers with periodic and continuous successful behaviors, and obtains the stable behavior node screening results;

[0081] The behavior effectiveness screening submodule compares the calculated behavior success ratio of 0.2393 with the behavior persistence assessment threshold. This threshold is set based on the reliability requirements of industrial-grade real-time control tasks, and is determined by analyzing 10,000 task cycles in a historical database. Logistic regression analysis was performed on the values ​​and the corresponding task's final success or failure status, determining that a value could distinguish between stable and unstable behavior with a 95% confidence level. The dividing point, 0.20, is set as the threshold value. This value can maximize the identification of unstable tasks while controlling the probability of misjudging stable tasks to within 5%. Since the success rate of task T001 is 0.2393, which is greater than the threshold value of 0.20, it is determined that the behavior of node N01 on this task has periodic continuity. After performing the same calculation and judgment on the other 4 tasks in the set, it is set that a total of 3 tasks pass the screening. Their corresponding node numbers are formed into a set, and the stable behavior node screening results {N01, N02, N04} are obtained.

[0082] Please see Figure 4 The stability assessment module includes:

[0083] The behavior sequence acquisition submodule obtains the node numbers involved in the stable behavior node screening results, collects the control behavior data sequence of the corresponding node in multiple scheduling cycles, including the number of consecutive successes, the number of control instruction loss markers and scheduling response delay time values ​​in each cycle, and aggregates the data according to the node number to obtain periodic behavior continuity index information.

[0084] The behavior sequence acquisition submodule, based on the stable behavior node filtering results {N01, N02, N04} obtained in the previous steps, collects the control behavior data sequence of node N01 within the most recent 5 scheduling cycles (C1 to C5). These sequences include the number of consecutive successes within the C1 to C5 cycles, recorded as [10, 9, 10, 8, 10] times, the number of control command loss markers, recorded as [0, 1, 0, 2, 0] times, and the scheduling response delay time value, recorded as [42, 45, 44, 49, 43] milliseconds. Based on the node number N01, the submodule aggregates these discrete data across cycles to form a structured time series dataset, obtaining periodic behavior continuity index information.

[0085] The trend extraction and calculation submodule, based on the control behavior sequence of the corresponding node according to the periodic behavior continuity index information, calls the consecutive success count, control command loss rate, and average response latency for each period, using the following formula:

[0086] ;

[0087] Stability score for instruction response during computation;

[0088] in, This indicates the score for command response stability. This represents the total number of consecutive successes, obtained by summing the number of successful executions within the task period. This represents the total number of cycles, obtained by the scheduling cycle length. This represents the normalized value of the control instruction loss rate, obtained as the ratio of the number of lost instructions to the total number of scheduled instructions. This represents the normalized value of the average response delay time within a period, obtained as the ratio of the average periodic response time to the maximum allowable response time. This is the normalized value of the current cycle delay time. This indicates the number of interrupt flags, which is obtained by recording the number of interrupt events. This represents the total number of tasks. The standard deviation of the normalized value of the control command loss rate is obtained by the standard deviation of the normalized value of the loss rate across cycles. The level division boundary is set based on the frequency distribution density function curve of the stability score in the sample node set. By setting the score density curve at the 1 / 3 and 2 / 3 quantiles as critical points, it is divided into three continuous stability level intervals, representing low stability, medium stability and high stability, respectively.

[0089] The trend extraction calculation submodule uses the periodic behavior continuity index information of node N01 to call its control behavior sequence to calculate the command response stability score. The formula used here is:

[0090] ;

[0091] The brilliance of this formula lies in its use of a "base score minus penalty term" structure to quantify stability, where the base score... This reflects the long-term average performance level, while the penalty term incorporates multiple instability factors, specifically, those in the molecule. This represents the direct impact of packet loss. Real-time jitter representing response time. The disturbance caused by external interruption events, these three factors combined constitute a comprehensive fluctuation, while the denominator contains... This considers whether the unstable behavior of packet loss itself is stable. If the packet loss rate fluctuates greatly... If the value is large, the penalty will be amplified, and the entire penalty term will be determined by the coefficient. Weighting is applied to ensure that the score is on an appropriate order of magnitude compared to the base score, thereby achieving a comprehensive and profound characterization of node stability. Regarding the parameters in the formula... It is the instruction response stability score to be calculated. The total number of consecutive successes is the sum of the successes in each cycle: 10 + 9 + 10 + 8 + 10 = 47. The total number of cycles is 5. To control the normalized value of the instruction loss rate, the total number of instruction losses is first calculated as 0+1+0+2+0=3. Assuming the number of planned tasks per cycle is 10, the total number of instructions is 10×5=50, and the instruction loss rate is 3 / 50=0.06. The maximum allowable instruction loss rate referenced for normalization is based on measured data of the bit error rate of industrial switches in underground high-dust environments (…). The analysis of the magnitude (of the signal) and the controller retransmission mechanism (maximum of 3 times) was set to 10% (i.e., 0.1). This value ensures that communication reliability can still be maintained at over 99.9% under typical harsh operating conditions. , To find the normalized value of the average response delay over a period of time, we first calculate the average response delay over 5 periods. Similarly, using the aforementioned 50ms as the maximum allowable response time for normalization, , Let C5 be the normalized value of the delay time for the current period (C5). The delay of C5 is 43ms. , To define the number of interrupt markers, if one interrupt event is recorded due to a network storm within cycle C4, then I=1. The total number of tasks is N=50. To control the standard deviation of the normalized value of the instruction loss rate, the normalized loss rate for each cycle is first calculated. The cycle loss rates are 0 / 10=0, 1 / 10=0.1, 0 / 10=0, 2 / 10=0.2, and 0 / 10=0. After normalization (dividing by 0.1), the sequence [0,1,0,2,0] is obtained. The standard deviation of this sequence is then calculated, and the average value is... Standard deviation:

[0092] ;

[0093] Substitute the parameters into the formula:

[0094] ;

[0095] The stability score for the instruction response of the node N01 is 9.2793.

[0096] The node level classification submodule judges the node based on the calculated instruction response stability score of 9.2793 according to the preset level classification boundary. This boundary is set by collecting a sample set containing 100 virtual nodes of different types, calculating the stability score S of each node, and plotting the frequency distribution density function curve of these scores. By analyzing the density curve, it was found that the curve has obvious inflection points at the score values ​​of 6.0 and 8.5, which indicates that there is a natural clustering of node stability. Therefore, these two inflection points are set as critical points. Specifically, the interval with a score below 6.0 is classified as "low stability", the interval with a score between 6.0 and 8.5 is classified as "medium stability", and the interval with a score above 8.5 is classified as "high stability". Since the score of node N01 is 9.2793, which is greater than 8.5, the submodule marks the node with the level label "high stability" and obtains the virtual node stability classification result, in which the level of node N01 is "high stability".

[0097] The node level classification submodule determines the level of each node's score range based on the command response stability score and the level classification boundaries set in the stability evaluation standard, labels the nodes with level tags, and obtains the virtual node stability classification results.

[0098] Please see Figure 5 The orchestration load adjustment module includes:

[0099] The node parameter call submodule obtains the node level field and task allocation field from the virtual node stability grading result, collects the CPU utilization and memory usage data corresponding to the current task, obtains the available CPU capacity and remaining memory capacity of the scheduling node where the task is located, establishes a resource mapping index table between tasks and nodes, and generates a task resource mapping index set.

[0100] To dynamically adjust system load, the node parameter call submodule first receives the virtual node stability grading results generated in the previous steps. This includes the "high stability" grade of node N01 and the task allocation field it carries, such as task T001 (tunnel boring machine front drum cutting). Subsequently, the submodule queries the real-time monitoring system to collect data showing that task T001 currently consumes 25% CPU and 150MB of memory. Simultaneously, it collects the resource status of the scheduling node N01 where the task resides, obtaining its available CPU capacity of 2.0 physical cores and remaining memory capacity of 1200MB. Based on this collected data, the submodule establishes a resource mapping index record between the task and the node, in the format {Task Number: T001, CPU Consumption: 25%, Memory Consumption: 150MB, Node: N01, Available CPU: 2.0 cores, Remaining Memory: 1200MB}. Finally, all similar records are aggregated to generate a task resource mapping index set.

[0101] Table 1 Example Table of Task and Node Resource Mapping Index

[0102] ;

[0103] As shown in Table 1, this table lists the resource usage and availability of some tasks and their respective nodes, providing a data foundation for subsequent resource matching calculations.

[0104] The resource comparison and calculation submodule extracts the current CPU utilization and memory usage of the task, as well as the available CPU capacity and remaining memory capacity of the node, based on the resource consumption data and remaining resource value of each task record in the task resource mapping index set, using the following formula:

[0105] ;

[0106] The system calculates and obtains the task resource remaining rate matching value, determines whether the matching value is greater than the resource scheduling and release threshold, filters the task numbers that meet the conditions, and generates the resource remaining rate matching result.

[0107] in, Indicates the resource remaining rate matching value. This represents a normalized value indicating the CPU utilization of a task, obtained by dividing the task's CPU utilization by the maximum core utilization limit allocated by the scheduler. This represents a normalized value of the node's available CPU capacity, obtained by dividing the node's current schedulable cores by the node's maximum number of cores. This represents a normalized value indicating the task's memory usage, obtained by dividing the task's current memory usage by the node's maximum allocated memory. This represents a normalized value indicating the remaining memory capacity of a node, obtained by dividing the node's current available memory by the maximum allocation limit. This represents the scheduling cycle number of the current task, which is obtained by finding the current number in the task scheduling cycle sequence. This represents the average of all task scheduling cycle numbers on the node, obtained by averaging the scheduling cycle numbers within the task queue. The normalized value representing the task execution success rate is obtained as the ratio of the current number of successful feedbacks to the total number of executions. The resource scheduling release threshold is set based on the load allocation limit standard for nodes with high resource contention in the scheduling system, which is derived from the empirical calculation value of the lower boundary of the average release success matching value in the resource saturation critical interval.

[0108] The resource comparison calculation submodule calculates the resource remaining rate matching value for each record in the task resource mapping index set. Taking task T001 as an example, the formula is as follows:

[0109] ;

[0110] The advantage of this formula lies in its ability to integrate three completely different dimensions—resource pressure, scheduling coordination, and execution reliability—through a multiplicative structure. This measure assesses the instantaneous pressure of the task on node resources; a higher ratio indicates greater resource strain. This item assesses the synchronization between the task scheduling cycle and other tasks on the node. The greater the cycle difference, the smaller the value of this item, thus reducing the overall matching degree and reflecting the impact of scheduling disturbances. The formula introduces the historical success rate of task execution as a baseline to ensure that only well-performing tasks have high matching values. This multi-dimensional comprehensive consideration allows the matching value V to more accurately identify those tasks that consume a lot of resources and have a potential impact on system stability, and therefore need to be prioritized for adjustment. For the parameters in the formula, V is the resource remaining rate matching value to be calculated; C is the normalized value of task CPU utilization, with a task CPU utilization of 25% and the scheduler's maximum core utilization limit set at 50%. This value is based on stress testing of all tasks in the system, determining that under conditions where single-core utilization does not exceed 50%, 99% of tasks will not cause observable scheduling delays to other tasks, hence C = 0.25 / 0.5 = 0.5; A is the normalized value of the node's available CPU capacity, node N01 is a quad-core processor with 2.0 cores currently available, hence A = 2.0 / 4 = 0.5; and M is the task memory. The normalized value of memory usage: the task uses 150MB of memory, and node N01 allocates a total memory limit of 2GB (2048MB) for this type of task. This limit is set based on the sum of max_rss of all possible tasks running on this node plus a 15% redundancy, so M = 150 / 2048 = 0.073. U is the normalized value of the node's remaining memory capacity; the node has 1200MB remaining, so U = 1200 / 2048 = 0.586. P is the scheduling cycle number of the current task, set to 20. Q is the average of the scheduling cycle numbers of all tasks on node N01. Assuming there are two other tasks on N01 with cycle numbers 22 and 24 respectively, then Q = (20 + 22 + 24) / 3 = 22. E is the normalized value of the task execution success rate. From the previous steps, we know that the task successfully executed 47 times, with a total of 50 executions, so E = 47 / 50 = 0.94. Substituting the values, we calculate:

[0111] ;

[0112] The calculated resource surplus rate matching value for task T001 is 0.995. This result indicates a relatively tight resource relationship between task T001 and node N01. This matching value is then compared with the resource scheduling and release threshold. This threshold is set based on the analysis of historical load data of nodes with high resource contention in the system. In a simulation experiment involving 30 nodes, the computational task load of each node is gradually increased, and its key performance indicators (such as average response latency) are monitored. When the CPU utilization reaches the 85% saturation critical interval, the V value of each task at this moment is calculated. The 25th percentile of these V values ​​is taken as the lower boundary, which is 0.9. Therefore, the threshold is set to 0.9. Since 0.995 > 0.9, the condition is met, and task T001 is selected, generating the resource surplus rate matching result.

[0113] The task priority filtering submodule sorts the tasks from high to low according to the matching value of the resource remaining rate matching results, extracts the task numbers of the target number as a high priority task candidate set, and marks the task type field and node level field to obtain a list of control task priority adjustment.

[0114] The task priority filtering submodule sorts all tasks that meet the conditions based on the matching value of each task in the resource remaining rate matching results. It is set that in addition to T001 (V=0.995), tasks T002 (V=1.102) and T008 (V=0.953) also meet the conditions. They are sorted from high to low according to the matching value to obtain the sequence [T002,T001,T008]. If the current scheduling plan needs to migrate a task (i.e., the target quantity is 1), the number of the first sorted task T002 is extracted as a high-priority task candidate. At the same time, its task type field (such as "hydraulic support control") and the level field ("high stability") of the current node N01 are marked. Finally, a record containing {task number:T002, task type:hydraulic support control, node level:high stability} is formed to obtain the control task priority adjustment list.

[0115] Please see Figure 6 The task migration control module includes:

[0116] The task dwell time determination submodule obtains task information from the task priority adjustment list, collects the IO request completion time, task dwell time and scheduling cooldown flag of the node corresponding to each task, determines whether the task dwell time is greater than the silent period window baseline value, removes task numbers whose scheduling cooldown flag is in a prohibited migration state, and generates a set of migrateable task numbers.

[0117] The task dwell time determination submodule obtains information on high-priority tasks from the control task priority adjustment list. Taking task T002 as an example, which is a hydraulic support control task, the submodule first collects several status parameters related to the task on its node N01, including the IO request completion time (recorded as 5ms), the dwell time of task T002 since its last deployment to N01 (recorded as 1800 seconds), and the scheduling cooldown flag, whose status is "allowed to migrate". Next, the submodule needs to determine whether the task dwell time is greater than the silent period window baseline value. This baseline value is set to prevent tasks from migrating to different nodes. Frequent inter-point migrations (i.e., "task jitter") were addressed by measuring the cold start time of 500 tasks in the system (including image retrieval, environment initialization, and application loading). The 90th percentile time was calculated to be 900 seconds, and a 33% stable operation observation period (300 seconds) was added. The baseline value was then set to 1200 seconds. Since the dwell time of task T002 was 1800 seconds, which is greater than 1200 seconds, and its scheduling cooldown flag was not disabled, this task was added to a temporary set. After judging all tasks in the list, a set of migrateable task numbers was finally generated.

[0118] The behavior pattern matching submodule, based on the control behavior sequence of tasks in the set of transferable task numbers within the most recent scheduling cycle, calls the task control response trajectory, instruction loop feature label, and feedback beat interval to determine whether the difference between the current behavior sequence and the corresponding behavior template of the previous cycle is less than the control behavior offset tolerance threshold, obtains the set of task numbers for stable behavior matching, and obtains the stable behavior pattern matching set.

[0119] The behavior pattern matching submodule retrieves detailed control behavior sequence data for task T002 from the set of transferable task IDs within its most recent scheduling cycle. This includes control response trajectories composed of consecutive response time points, such as [12ms, 15ms, 13ms, ...], instruction cycle feature labels characterizing its operating mode, such as the "support-lowering-moving" sequence, and its periodic feedback beat interval, averaging 100ms. The submodule compares this current behavior sequence with the behavior template from the previous stable cycle stored in the system. This comparison is not a simple numerical comparison but rather calculates the distance between the two behavior sequences in a high-dimensional feature space, such as calculating the dynamic time warping (DTW) distance between the two response trajectories. The distance and the edit distance of the instruction sequence are used to determine whether the difference is less than the preset control behavior deviation tolerance threshold. The threshold is set based on statistical analysis of the behavior data of a large number of normal operation tasks, calculating the standard deviation of its periodic behavior fluctuations, and applying the three-standard-deviation (3-sigma) principle to each behavior feature to construct a multi-dimensional tolerance boundary. Any behavior that exceeds this boundary is considered an abnormal deviation. The calculated comprehensive difference between the current behavior sequence and the template is set to 0.15, and the tolerance threshold is 0.2. Since 0.15 < 0.2, it indicates that the behavior pattern of task T002 is stable and has not drifted significantly. Therefore, its number is put into the behavior stable matching task number set to obtain the behavior pattern stable matching set.

[0120] The migration node filtering submodule stably matches the centralized task number based on the behavior pattern, extracts the resource schedulable status, behavior compatibility label and resource scheduling priority field from the list of currently available nodes, filters the node number with the same behavior label and the scheduling priority field is lower than the task priority level, establishes the correspondence between task number and target node, and obtains the control task migration target allocation table.

[0121] The migration node filtering submodule receives task T002 from the stable behavior pattern matching set and begins searching for the best migration target in the list of currently available nodes across the entire network. The submodule first extracts several key attributes from each available node in the list (e.g., N03, N05), including resource schedulable status (e.g., "available"), behavior compatibility tags describing the node's capabilities (e.g., [hydraulic control, data acquisition]), and the node's resource scheduling priority field (e.g., "medium"). The filtering process first performs functional matching. Task T002 is a "hydraulic support control" task, which requires nodes to have the "hydraulic control" behavior tag. The tag for node N03 is set to [data acquisition]. According to the data collection and network forwarding, node N05 is labeled as [hydraulic control, coal flow monitoring], so N03 is excluded and N05 becomes a candidate. Next, priority judgment is performed. Task T002 comes from a "highly stable" node, which has a higher migration priority. The resource scheduling priority field of candidate node N05 is "medium", which is lower than the implicit migration priority of the task, thus meeting the screening conditions. Therefore, under the premise that the resources (CPU, memory) of N05 also meet the requirements of task T002, the submodule establishes the migration correspondence between task T002 and target node N05. Finally, all such successful matches are summarized to obtain the control task migration target allocation table.

[0122] The virtualization orchestration method for industrial control includes the following steps:

[0123] S1: Obtain the virtual control container configuration item in the driving logic of mining equipment, call the loop control flag and interrupt flag in the task tag field, determine the number of control tasks with closed-loop characteristics in the continuous scheduling cycle, and obtain the set of programmable control tasks.

[0124] S2: Based on the node number bound to each task in the programmable control task set, collect the control response delay time, task completion count value and total number of planned tasks of the mining equipment control unit node, filter stable behavior nodes, and obtain the stable behavior node filtering results;

[0125] S3: Based on the stable behavior node screening results, according to the task operation data of mining equipment, detect the changing trend during the week and extract the changing value, calculate the command response stability score, classify the node stability level, and generate the virtual node stability grading result.

[0126] S4: Call the node level field and task allocation field in the virtual node stability grading result, collect the CPU utilization and memory usage of the scheduled tasks in the virtual control of mining equipment, calculate the remaining resource value, extract the task number of the target number, and obtain the control task priority adjustment list.

[0127] S5: Adjust the task information in the control task priority list, determine whether the task dwell time is higher than the defined quiet period window, filter tasks that meet the scheduling cycle migration conditions, filter the migration target node number based on the task's behavior pattern matching results in the most recent cycle, and generate a control task migration target allocation table.

[0128] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A virtualization orchestration system for industrial control, characterized in that, The system comprises: The task extraction module obtains the virtual control container configuration item in the mine equipment driving logic, calls the loop control mark and the interrupt mark in the task label field, judges the number of control tasks with closed loop characteristics in the continuous scheduling period, and obtains the schedulable control task set; The node behavior monitoring module collects the control response delay time, task completion count value and planned task total number of the mine equipment control unit node according to the node number bound by each task in the schedulable control task set, filters the stable behavior node, and obtains the stable behavior node filtering result; The stability evaluation module detects the periodical change trend and extracts the change value based on the stable behavior node filtering result and the task running data of the mine equipment, calculates the instruction response stability score, performs node stability level division, and generates the virtual node stability grading result; The arrangement load adjustment module calls the node level field and the task allocation field in the virtual node stability grading result, collects the CPU usage rate and memory usage of the scheduled tasks in the virtual control of the mine equipment, calculates the resource remaining value, extracts the target number of task numbers, and obtains the control task priority adjustment list; The control task migration module judges whether the task residence time is higher than the defined silence period window according to the task information in the control task priority adjustment list, filters the tasks meeting the scheduling period migration condition, selects the migration target node number according to the behavior mode matching result in the recent period of the task, and generates the control task migration target allocation table; The schedulable control task set comprises a scheduling task identifier, a closed loop execution label and a scheduling period number, the stable behavior node filtering result specifically indicates a task behavior success rate value, a node mark state and a response duration record, the virtual node stability grading result comprises a node grading number, a behavior duration rate level and a stability difference value, the control task priority adjustment list specifically comprises a task scheduling number, a resource remaining evaluation value and a load classification label, and the control task migration target allocation table comprises a migratable task number, a target node number and a behavior mode index.

2. The virtualized orchestration system of claim 1, wherein, The task extraction module comprises: The configuration reading submodule obtains the task scheduling mapping table, the task execution sequence and the bound executor instruction queue in the virtual control container configuration item in the mine equipment driving logic, judges whether the task label field in the task execution sequence has the continuous combination of the scheduling number field, the mapping path field and the bound control instruction field, filters the task configuration record with scheduling integrity, and generates the number of task structure identification tags; The task structure identification submodule collects the executor number, the scheduling sequence code and the execution feedback number of the task according to the scheduling task record corresponding to the number of task structure identification tags, judges whether the feedback number and the sequence code number are consistent, filters the task number set with complete instruction structure, and obtains the number of closed loop characteristic tasks; The task closed loop judgment submodule is based on the task number set corresponding to the closed loop characteristic task quantity information value, counts the occurrence times of each task number in the continuous scheduling period, judges whether the occurrence times are greater than the continuous scheduling period determination value, screens the task numbers meeting the determination value to form a set, and obtains the schedulable control task set.

3. The virtualized orchestration system of claim 1, wherein, The node behavior monitoring module comprises: The response data acquisition submodule acquires the control unit node number bound by each task in the schedulable control task set, acquires the control response time record, the completed task count log and the total number of tasks in the task schedule table in the current scheduling period in each node, judges whether the task completion count log has a valid task record number, and obtains the control task execution quantity information; The behavior success rate calculation submodule is based on the total number of planned tasks corresponding to the task record in the control task execution quantity information, calls the corresponding control response time and calculates the average response delay, combines the task completion number and the planned task number to perform normalization, and calculates the task behavior success ratio value; The behavior effectiveness screening submodule compares the behavior success ratio value with the behavior persistence evaluation threshold value, screens the node number set having a periodically persistent successful behavior, and obtains the stable behavior node screening result.

4. The virtualization orchestration system of claim 1, wherein, The stability evaluation module comprises: The behavior sequence acquisition submodule acquires the node numbers involved in the stable behavior node screening result, acquires the control behavior data sequence of the corresponding node in multiple scheduling periods, including the number of consecutive successes in each period, the number of control instruction loss marks and the scheduling response delay time value, aggregates the data according to the node numbers, and obtains the periodic behavior continuity index information; The trend extraction calculation submodule acquires the instruction response stability score according to the control behavior sequence of the corresponding node according to the periodic behavior continuity index information, calls the number of consecutive successes in each period, the control instruction loss rate and the average response delay time, and calculates the instruction response stability score. The node level division submodule acquires the virtual node stability grading result according to the instruction response stability score, judges the level to which the score interval of each node belongs according to the level division boundary set in the stability evaluation standard, labels the node with a level label, and acquires the virtual node stability grading result.

5. The virtualization orchestration system of claim 1, wherein, The arrangement load adjustment module comprises: The node parameter calling submodule acquires the node level field and the task allocation field in the virtual node stability grading result, acquires the CPU usage rate and the memory usage data corresponding to the current task, acquires the available CPU capacity and the remaining memory capacity of the scheduling node of the task, establishes a resource mapping index table of the task and the node, and generates a task resource mapping index set; The resource comparison calculation submodule extracts the current CPU usage rate, the memory usage and the node CPU available capacity and the remaining memory capacity data according to the resource consumption data and the node remaining resource value corresponding to each task record in the task resource mapping index set, calculates the task resource remaining rate matching value, judges whether the matching value is greater than the resource scheduling release threshold, screens the numbers of the tasks meeting the condition, and generates the resource remaining rate matching result; The task priority screening submodule extracts a target number of task numbers as a high-priority task candidate set according to the matching value results of the tasks in the resource residual rate matching results, sorts the matching values from high to low, and marks the task type field and the node level field to obtain a control task priority adjustment list.

6. The virtualization orchestration system of claim 1, wherein, The control task migration module comprises: The task residence judgment submodule acquires task information in the control task priority adjustment list, collects IO request completion time, task residence time and scheduling cooling flag of each task corresponding node, judges whether the task residence time is greater than the silence period window reference value, eliminates the task number in the prohibited migration state of the scheduling cooling flag, and generates a migratable task number set; The behavior pattern matching submodule calls task control response trajectory, instruction cycle characteristic label and feedback beat interval according to the control behavior sequence of the tasks in the migratable task number set in the latest scheduling period, judges whether the difference between the current behavior sequence and the corresponding behavior template in the previous period is less than the control behavior offset tolerance threshold, obtains a behavior stable matching task number set, and obtains a behavior pattern stable matching set; The migration node screening submodule extracts the resource schedulable state, behavior compatible label and resource scheduling priority field in the current available node list according to the task number in the behavior pattern stable matching set, screens the node number with the same behavior label and lower scheduling priority field than the task priority level, establishes the corresponding relationship between the task number and the target node, and obtains a control task migration target allocation table.

7. A method of virtualization orchestration for industrial control, characterized by, The method is used for realizing the industrial control virtualization orchestration system in any one of claims 1-6, comprising the following steps: S1: acquiring a virtual control container configuration item in a mine equipment driving logic, calling a loop control mark and an interrupt mark in a task label field, judging the number of control tasks with closed-loop characteristics in a continuous scheduling period, and obtaining a schedulable control task set; S2: according to the node number bound by each task in the schedulable control task set, collecting the control response delay time, task completion count value and planned task total number of the mine equipment control unit node, screening a stable behavior node, and obtaining a stable behavior node screening result; S3: based on the stable behavior node screening result, detecting the change trend and extracting the change value according to the task running data of the mine equipment in a period, calculating an instruction response stability score, performing node stability level division, and generating a virtual node stability grading result; S4: calling the node level field and the task allocation field in the virtual node stability grading result, collecting the CPU usage and memory usage of the scheduled tasks in the virtual control of the mine equipment, calculating a resource residual value, extracting a target number of task numbers, and obtaining a control task priority adjustment list; S5: according to the task information in the control task priority adjustment list, judging whether the task residence time is higher than the defined silence period window, screening tasks meeting the scheduling period migration condition, screening the migration target node number according to the behavior pattern matching result of the task in the latest period, and generating a control task migration target allocation table; The programmable control task set includes a scheduling task identifier, a closed-loop execution label, and a scheduling cycle number, the stable behavior node screening result specifically indicates a task behavior success rate value, a node marker state, and a response duration record, the virtual node stability classification result includes a node classification number, a behavior duration rate level, and a stability difference value, the control task priority adjustment list specifically is a task scheduling number, a resource remaining evaluation value, and a load classification label, and the control task migration target allocation table includes a migratable task number, a target node number, and a behavior mode index.

Images