Trusted execution methods, apparatus, devices, storage media, and products for task chains

Abstract

This application discloses a reliable execution method, apparatus, device, storage medium, and product for task chains. The method includes acquiring raw data from industrial equipment; filtering the raw data through a dynamic input gating layer to obtain task data; generating several atomic tasks through a task scheduling engine based on prior knowledge of the industrial equipment and the task data; and controlling the industrial equipment to execute the several atomic tasks. This application uses a dynamic input gating layer to filter the raw data to obtain effective and reliable task data. Then, it combines prior knowledge providing accurate current state information of the industrial equipment with the task data to perform atomic operations to decompose the industrial task chain into multiple reasonable sub-atomic tasks. Finally, it controls the industrial equipment to execute several atomic tasks, effectively solving problems such as information loss and equipment status misjudgment caused by context window limitations in existing technologies, thus improving the accuracy and reliability of industrial control.

Description

Technical Field

[0001] This application relates to the field of industrial control technology, and in particular to a reliable execution method, apparatus, device, storage medium and product for task chains. Background Technology

[0002] Industrial tasks typically involve multiple interconnected steps that must be executed sequentially. These steps are linked and interdependent, forming a complete task chain. Taking a smart mine scenario as an example, an industrial task chain generally includes scheduling tunneling machines to carry out mining operations, transporting minerals via conveyors, and ensuring the safety of the mine environment through ventilation systems. The tasks are closely linked, and any delay or error in any step can affect the overall effectiveness of the task chain.

[0003] However, when faced with industrial tasks involving multiple interconnected steps, the large amount of contextual information in each step of the task chain can lead to information truncation in large models during multi-step inference due to the limitations of the context window, resulting in the loss of critical information. This can cause misjudgments of equipment status, increasing the failure rate of multi-step planning, and even errors during tool invocation, triggering cascading failures. These potential risks can severely impact the overall stability of the system, and may even lead to production interruptions or safety incidents.

[0004] Therefore, existing technologies still need to be improved and enhanced. Summary of the Invention

[0005] The technical problem to be solved by this application is to provide a reliable execution method, apparatus, device, storage medium and product for task chains, addressing the shortcomings of the prior art.

[0006] To address the aforementioned technical problems, the first aspect of this application provides a trusted execution method for task chains, wherein the trusted execution method for task chains specifically includes: Acquire raw data from industrial equipment; The raw data is filtered through a dynamic input gating layer to obtain task data; The task scheduling engine generates several atomic tasks based on the prior knowledge of the industrial equipment and the task data. Control the industrial equipment to perform several atomic tasks.

[0007] The aforementioned task chain trusted execution method, wherein the original data includes control data of industrial equipment and task requirement data of the target task.

[0008] The aforementioned reliable execution method for task chains, wherein the step of filtering the original data through a dynamic input gating layer to obtain task data specifically includes: The control data in the original data is classified by a classifier in the dynamic input gating layer to obtain effective data; By using the constraint mechanism in the dynamic input gating layer to apply syntactic constraints to the valid data according to the industrial protocol corresponding to the industrial equipment, task data is obtained.

[0009] The aforementioned task chain trusted execution method, wherein the prior knowledge includes historical equipment status data of industrial equipment, industrial knowledge graph, and equipment status snapshots.

[0010] The aforementioned reliable execution method for task chains, wherein generating several atomic tasks through a task scheduling engine based on prior knowledge of the industrial equipment and the task data specifically includes: The task scheduling engine determines the current status data of the industrial equipment based on the prior knowledge of the industrial equipment. The task scheduling engine generates several atomic tasks based on the current state data and the task data.

[0011] The aforementioned reliable execution method for task chains, wherein determining the current state data of the industrial equipment through a task scheduling engine based on prior knowledge of the industrial equipment specifically includes: The historical device status data in the prior knowledge is weighted to obtain reference device status data, wherein the weighting coefficient of the historical device status data is a time decay factor. The reference equipment status data is corrected based on the industrial knowledge graph in the prior knowledge, and the current status data is determined based on the corrected reference equipment status data and the equipment status snapshot in the prior knowledge.

[0012] The aforementioned trusted execution method for task chains, wherein generating several atomic tasks by a task scheduling engine based on the current state data and the task data specifically includes: A candidate decomposition set is constructed based on the task data, wherein each candidate decomposition in the candidate decomposition set is a directed graph consisting of several atomic tasks and their execution order. Construct a resource consumption function with the goal of minimizing resource consumption; The resource consumption function is solved on the candidate decomposition set based on the current state data, with equipment capacity as a constraint, to obtain several atomic tasks.

[0013] The task chain trusted execution method, wherein the resource consumption function includes a dependency conflict indicator function item, which is used to identify circular dependencies and / or resource conflicts between atomic tasks.

[0014] The aforementioned reliable execution method for task chains, wherein the resource consumption function is: , in, This indicates a goal decomposition that includes several atomic tasks. Represents the candidate decomposition set. Represents the resource consumption matrix. Represents the device capability vector. Indicates candidate decomposition, Indicates the conflict adjustment factor. This indicates a dependency conflict indicator function item.

[0015] The trusted execution method for the task chain, wherein after generating several atomic tasks by a task scheduling engine based on prior knowledge of the industrial equipment and the task data, the method further includes: Extract the drift features of atomic tasks and predict the risk probability of atomic tasks based on the drift features; When there is a first target atomic task with a risk probability greater than a preset risk probability threshold, obtain the drift compensation of the first target atomic task; The drift compensation is fed back to the dynamic input gating layer so that the dynamic input gating layer can perform offline updates based on the drift compensation.

[0016] The task chain trusted execution method, wherein the drift feature includes at least one of the following: output variance, latency variation gradient, or tool invocation error rate.

[0017] The aforementioned reliable execution method for task chains, wherein predicting the risk probability of atomic tasks based on the drift features specifically involves: The drift features are input into a trained risk prediction model, which then outputs the risk probability of the original task.

[0018] The task chain trusted execution method further includes, before controlling the industrial equipment to execute several atomic tasks: Perform dual-path redundancy operations on the atomic task to obtain the execution result of each execution path; The confidence level of the atomic task is determined based on the consistency of the execution results of the two execution paths. For a second target atomic task with a confidence level lower than a preset confidence threshold, the execution result of the second target atomic task is determined using a rule engine.

[0019] The aforementioned trusted execution method for task chains further includes: When there is a second target atomic task with a confidence level lower than the preset confidence threshold, the second target atomic task is fed back to the task scheduling engine to provide a dynamic compensation version for the task scheduling engine.

[0020] The task chain trusted execution method, prior to acquiring the raw data of the industrial equipment, further includes: Obtain the candidate resource device set; Construct an objective function that includes measures of resource cost and service quality degradation; Based on the objective function, the target resource device corresponding to each atomic task is selected from the candidate resource device set.

[0021] The aforementioned trusted execution method for task chains, wherein controlling the industrial equipment to execute several atomic tasks specifically includes: The atomic tasks are encoded into industrial control instructions using the target resource equipment; The industrial control instructions are sent to the industrial equipment so that the industrial equipment performs several atomic tasks.

[0022] The trusted execution method for the task chain, wherein after encoding the atomic task into industrial control instructions using the target resource device, the method further includes: Add a cyclic redundancy check code and a device digital signature to the industrial control command.

[0023] A second aspect of this application provides an apparatus for a trusted execution method of a task chain, wherein the apparatus specifically includes: The acquisition module is used to acquire raw data from industrial equipment; The first processing module is used to filter the raw data through a dynamic input gating layer to obtain task data; The second processing module generates several atomic tasks based on the prior knowledge of the industrial equipment and the task data through the task scheduling engine. The control module is used to control the industrial equipment to perform several atomic tasks.

[0024] A third aspect of this application provides a computer-readable storage medium storing one or more programs that can be executed by one or more processors to implement the steps in the task chain trusted execution method as described above.

[0025] A fourth aspect of this application provides an electronic device comprising: a processor and a memory; The memory stores a computer-readable program that can be executed by the processor; When the processor executes the computer-readable program, it implements the steps in any of the above-described trusted execution methods for task chains.

[0026] The fifth aspect of this application provides a computer program product, wherein the computer program product includes a computer program that, when executed by a processor, implements the steps of the task chain trusted execution method as described above.

[0027] Beneficial Effects: Compared with existing technologies, this application provides a reliable execution method, apparatus, device, storage medium, and product for task chains. The method includes acquiring raw data from industrial equipment; filtering the raw data through a dynamic input gating layer to obtain task data; generating several atomic tasks through a task scheduling engine based on prior knowledge of the industrial equipment and the task data; and controlling the industrial equipment to execute the several atomic tasks. This application uses a dynamic input gating layer to filter the raw data to obtain effective and reliable task data. Then, it combines prior knowledge providing accurate current state information of the industrial equipment with the task data to perform atomic operations to decompose the industrial task chain into multiple reasonable sub-atomic tasks. Finally, it controls the industrial equipment to execute several atomic tasks, effectively solving problems such as information loss and equipment status misjudgment caused by context window limitations in existing technologies, thus improving the accuracy and reliability of industrial control. Furthermore, by decomposing the task chain into atomic operations, it avoids the problem of task chain planning collapse caused by the task chain including multiple operation tasks. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the embodiments of this application, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 A flowchart illustrating an application scenario of the trusted execution method for task chains provided in this application embodiment.

[0030] Figure 2 A schematic block diagram of an electronic device provided in an embodiment of this application.

[0031] Figure 3 A flowchart of a task chain trusted execution method provided in an embodiment of this application.

[0032] Figure 4 This is a flowchart illustrating step S20.

[0033] Figure 5 This is a flowchart illustrating step S30.

[0034] Figure 6 This is a flowchart illustrating step S31.

[0035] Figure 7 This is a flowchart illustrating step S32.

[0036] Figure 8 This is a flowchart illustrating the process of constructing the candidate decomposition set.

[0037] Figure 9 A flowchart illustrating the compensation process for drift compensation.

[0038] Figure 10 This is a flowchart illustrating step S40.

[0039] Figure 11 This is a flowchart illustrating the verification process for dual-path redundancy verification.

[0040] Figure 12 This is a flowchart illustrating the process of selecting target resource equipment.

[0041] Figure 13 A schematic diagram of the trusted execution device for the task chain provided in the embodiments of this application. Detailed Implementation

[0042] This application provides a method, apparatus, device, storage medium, and product for trusted execution of a task chain. To make the objectives, technical solutions, and effects of this application clearer and more explicit, the following detailed description, with reference to the accompanying drawings and embodiments, further illustrates this application. It should be understood that the specific embodiments described herein are merely illustrative and are not intended to limit this application.

[0043] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms “a,” “an,” “the,” and “the” used herein may also include the plural forms. It should be further understood that the term “comprising” as used in this application means the presence of the stated features, integers, steps, operations, elements, and / or components, but does not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. It should be understood that when we say an element is “connected” or “coupled” to another element, it can be directly connected or coupled to the other element, or there may be intermediate elements. Furthermore, “connected” or “coupled” as used herein can include wireless connections or wireless coupling. The term “and / or” as used herein includes all or any units and all combinations of one or more associated listed items.

[0044] It will be understood by those skilled in the art that, unless otherwise defined, all terms used herein (including technical and scientific terms) have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains. It should also be understood that terms such as those defined in general dictionaries should be understood to have the same meaning as in the context of the prior art, and should not be interpreted in an idealized or overly formal sense unless specifically defined as herein.

[0045] It should be understood that the sequence number and size of each step in this embodiment do not imply the order of execution. The execution sequence of each process is determined by its function and internal logic, and should not constitute any limitation on the implementation process of this application embodiment.

[0046] Research has revealed that industrial tasks typically involve multiple interconnected steps that must be executed sequentially. These steps are interdependent and form a complete task chain. Taking a smart mine scenario as an example, an industrial task chain generally encompasses scheduling tunneling machines for mining operations, transporting minerals via conveyors, and ensuring mine environmental safety through ventilation systems. These tasks are closely linked, and any delay or error in any step can affect the overall effectiveness of the task chain.

[0047] However, when faced with industrial tasks involving multiple interconnected steps, the large amount of contextual information in each step of the task chain can lead to information truncation in large models during multi-step inference due to the limitations of the context window, resulting in the loss of critical information. This can cause misjudgments of equipment status, increasing the failure rate of multi-step planning, and even errors during tool invocation, triggering cascading failures. These potential risks can severely impact the overall stability of the system, and may even lead to production interruptions or safety incidents.

[0048] To address the aforementioned issues, this application embodiment acquires raw data from industrial equipment; filters the raw data using a dynamic input gating layer to obtain task data; generates several atomic tasks using a task scheduling engine based on prior knowledge of the industrial equipment and the task data; and controls the industrial equipment to execute these atomic tasks. This application uses a dynamic input gating layer to filter the raw data to obtain effective and reliable task data. Then, it combines prior knowledge providing accurate current state information of the industrial equipment with the task data to atomically decompose the industrial task chain, dividing it into multiple reasonable sub-atomic tasks. Finally, it controls the industrial equipment to execute these atomic tasks, effectively solving problems such as information loss and equipment status misjudgment caused by context window limitations in existing technologies, thus improving the accuracy and reliability of industrial control. Furthermore, by atomically decomposing the task chain, it avoids the problem of task chain planning collapse caused by the task chain including multiple operation tasks.

[0049] The application content will be further explained below with reference to the accompanying drawings and the description of the embodiments.

[0050] Please see Figure 1 , Figure 1 This is an application environment diagram of the trusted execution method for task chains provided in the embodiments of this application. For example... Figure 1 As shown in the diagram, the task chain trusted execution method system in this application environment includes a user terminal 110 and a server 120. The user terminal 110 and server 120 are connected via a network. The user terminal 110 can be a desktop user terminal 110 or a mobile user terminal 110, specifically a mobile user terminal 110 can be at least one of a mobile phone, tablet computer, or laptop computer. The server 120 can be implemented as a standalone server 120 or a cluster of multiple servers 120. The user terminal receives raw data from the industrial equipment input by the user. The server 120 obtains the raw data from the user terminal 110, filters the raw data through a dynamic input gating layer to obtain task data, generates several atomic tasks based on prior knowledge of the industrial equipment and the task data through a task scheduling engine, and controls the industrial equipment to execute these atomic tasks.

[0051] Please see Figure 2 , Figure 2 This is a schematic block diagram of the electronic device provided in the embodiments of this application. Figure 2 As shown, the electronic device includes a processor and a memory. The processor 1001 and the memory 1002 are connected by a bus, such as an I2C (Inter-integrated Circuit) bus or a distributed soft bus.

[0052] The memory 1002 may include a non-volatile storage medium and internal memory. The non-volatile storage medium may store an operating system and a computer program. The computer program includes program instructions that, when executed, cause the processor to perform any trusted execution method of a task chain.

[0053] Processor 1001 provides computing and control capabilities to support the operation of the entire electronic device. Processor 1001 can be a Central Processing Unit (CPU), but it can also be other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), field-programmable gate arrays (FPGAs), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor can be a microprocessor or any conventional processor.

[0054] Specifically, the processor 1001 is used to run a computer program stored in memory, and when executing the computer program, it performs the following steps: Acquire raw data from industrial equipment; The raw data is filtered through a dynamic input gating layer to obtain task data; The task scheduling engine generates several atomic tasks based on the prior knowledge of the industrial equipment and the task data. Control the industrial equipment to perform several atomic tasks.

[0055] In some embodiments, the raw data includes control data of industrial equipment and task requirement data for the target task.

[0056] In some embodiments, when the processor 1001 filters the raw data through a dynamic input gating layer to obtain task data, it specifically implements the following: The control data in the original data is classified by a classifier in the dynamic input gating layer to obtain effective data; By using the constraint mechanism in the dynamic input gating layer to apply syntactic constraints to the valid data according to the industrial protocol corresponding to the industrial equipment, task data is obtained.

[0057] In some embodiments, the prior knowledge includes historical equipment status data of industrial equipment, industrial knowledge graphs, and equipment status snapshots.

[0058] In some embodiments, when the processor 1001 generates a plurality of atomic tasks through a task scheduling engine based on prior knowledge of the industrial equipment and the task data, it is specifically used to implement: The task scheduling engine determines the current status data of the industrial equipment based on the prior knowledge of the industrial equipment. The task scheduling engine generates several atomic tasks based on the current state data and the task data.

[0059] In some embodiments, when the processor 1001 determines the current state data of the industrial equipment based on prior knowledge of the industrial equipment through the task scheduling engine, it specifically implements the following: The historical device status data in the prior knowledge is weighted to obtain reference device status data, wherein the weighting coefficient of the historical device status data is a time decay factor. The reference equipment status data is corrected based on the industrial knowledge graph in the prior knowledge, and the current status data is determined based on the corrected reference equipment status data and the equipment status snapshot in the prior knowledge.

[0060] In some embodiments, when the processor 1001 generates several atomic tasks based on the current state data and the task data through the task scheduling engine, it is specifically configured to: In some embodiments, when the processor 1001 generates several atomic tasks based on the current state data and the task data through the task scheduling engine, it is specifically configured to: A candidate decomposition set is constructed based on the task data, wherein each candidate decomposition in the candidate decomposition set is a directed graph consisting of several atomic tasks and their execution order. Construct a resource consumption function with the goal of minimizing resource consumption; The resource consumption function is solved on the candidate decomposition set based on the current state data, with equipment capacity as a constraint, to obtain several atomic tasks.

[0061] In some embodiments, the resource consumption function includes a dependency conflict indicator function term, which is used to identify circular dependencies and / or resource conflicts between atomic tasks.

[0062] In some embodiments, the resource consumption function is: , in, This indicates a goal decomposition that includes several atomic tasks. Represents the candidate decomposition set. Represents the resource consumption matrix. Represents the device capability vector. Indicates candidate decomposition, Indicates the conflict adjustment factor. This indicates a dependency conflict indicator function item.

[0063] In some embodiments, after generating a plurality of atomic tasks by a task scheduling engine based on prior knowledge of the industrial equipment and the task data, the processor 1001 is configured to implement: Extract the drift features of atomic tasks and predict the risk probability of atomic tasks based on the drift features; When there is a first target atomic task with a risk probability greater than a preset risk probability threshold, obtain the drift compensation of the first target atomic task; The drift compensation is fed back to the dynamic input gating layer so that the dynamic input gating layer can perform offline updates based on the drift compensation.

[0064] In some embodiments, the drift characteristics include at least one of the following: output variance, time delay variation gradient, or tool call error rate.

[0065] In some embodiments, when the processor 1001 predicts the risk probability of an atomic task based on the drift features, it is specifically configured to: The drift features are input into a trained risk prediction model, which then outputs the risk probability of the original task.

[0066] In some embodiments, before controlling the industrial equipment to perform several atomic tasks, the processor 1001 is configured to perform: Perform dual-path redundancy operations on the atomic task to obtain the execution result of each execution path; The confidence level of the atomic task is determined based on the consistency of the execution results of the two execution paths. For a second target atomic task with a confidence level lower than a preset confidence threshold, the execution result of the second target atomic task is determined using a rule engine.

[0067] In some embodiments, the processor 1001 is configured to implement: When there is a second target atomic task with a confidence level lower than the preset confidence threshold, the second target atomic task is fed back to the task scheduling engine to provide a dynamic compensation version for the task scheduling engine.

[0068] In some embodiments, before controlling the industrial equipment to perform several atomic tasks, the processor 1001 is configured to perform: Obtain the candidate resource device set; Construct an objective function that includes measures of resource cost and service quality degradation; Based on the objective function, the target resource device corresponding to each atomic task is selected from the candidate resource device set.

[0069] In some embodiments, when the processor 1001 controls the industrial equipment to perform several atomic tasks, it is specifically used to implement: The atomic tasks are encoded into industrial control instructions using the target resource equipment; The industrial control instructions are sent to the industrial equipment so that the industrial equipment performs several atomic tasks.

[0070] In some embodiments, after encoding the atomic task into industrial control instructions using the target resource device, the processor 1001 is configured to implement: Add a cyclic redundancy check code and a device digital signature to the industrial control command.

[0071] Please see Figure 3 , Figure 3 This is a flowchart of the trusted execution method for task chains provided in this embodiment. The trusted execution method includes acquiring raw data from industrial equipment; filtering the raw data through a dynamic input gating layer to obtain task data; generating several atomic tasks through a task scheduling engine based on prior knowledge of the industrial equipment and the task data; and controlling the industrial equipment to execute the several atomic tasks. This application uses a dynamic input gating layer to filter the raw data to obtain effective and reliable task data. Then, it combines prior knowledge providing accurate current state information of the industrial equipment with the task data to perform atomic operation decomposition on the industrial task chain, dividing the industrial task chain into multiple reasonable sub-atomic tasks. Finally, it controls the industrial equipment to execute several atomic tasks, effectively solving problems such as information loss and equipment status misjudgment caused by context window limitations in existing technologies, thus improving the accuracy and reliability of industrial control. Simultaneously, by decomposing the task chain into atomic operations, it avoids the problem of task chain planning collapse caused by the task chain including multiple operation tasks.

[0072] like Figure 3 As shown, the trusted execution method for task chains provided in this application embodiment specifically includes: S10. Obtain raw data from industrial equipment.

[0073] Specifically, industrial equipment can include various types of mechanical and electronic equipment. For example, in a smart mine scenario, industrial equipment can include tunneling machines, conveyors, and ventilation systems; in a smart street light network scenario, industrial equipment can include street light controllers and sensors. Raw data includes control data from the industrial equipment and task requirement data for the target task. Control data can include start / stop commands, operating speeds, etc., of the working equipment. Task requirement data can include time requirements for task completion, expected work results, and operating procedures. Control data from industrial equipment can be collected and transmitted in real time through built-in sensors and monitoring systems. For example, the operating speed data of a tunneling machine can be collected by a speed sensor installed on the machine; the on / off status data of a street light controller can be obtained through a monitoring system. Task requirement data for the target task can be obtained through user input or system presets. For example, in a smart mine scenario, a user can input task requirements such as the tunneling depth of the tunneling machine and the conveying capacity of the conveyor through an operating terminal; in a smart street light network scenario, the system can determine task requirements such as the on / off time and brightness adjustment of the street lights based on preset time and lighting conditions.

[0074] S20. The raw data is filtered through a dynamic input gating layer to obtain task data.

[0075] Specifically, the dynamic input gating layer is a network structure within a pre-built trusted execution base. This trusted execution base includes a dynamic input gating layer and a task scheduling engine. The dynamic input gating layer filters the raw data to obtain task data, while the task scheduling engine generates several atomic tasks based on the prior knowledge of the industrial equipment and the task data. In other words, the trusted execution base is used to decompose the task chain executed by the industrial equipment into atomic tasks based on the raw data. The trusted execution base can employ a large model structure, such as GPT or BERT. Large model structures possess powerful language understanding and processing capabilities, enabling better filtering of raw data and decomposition of task chains. When filtering raw data, the dynamic input gating layer, leveraging the advantages of a large model, can more accurately identify abnormal content, classify control data, and apply grammatical constraints according to industrial protocols.

[0076] The task data is obtained by filtering the original data through a dynamic input gating layer. The data quality of the task data is higher than that of the original data. For example, the task data does not include abnormal data in the original data, data that does not conform to industrial protocols in the original data, or invalid control data in the original data.

[0077] In one embodiment, such as Figure 4 As shown, the process of filtering the raw data through a dynamic input gating layer to obtain task data specifically includes: S21. The control data in the original data is classified by the classifier in the dynamic input gating layer to obtain effective data; S22. By using the constraint in the dynamic input gating layer to apply syntactic constraints to the valid data according to the industrial protocol corresponding to the industrial equipment, task data is obtained.

[0078] In step S21, the control data in the raw data includes control commands for industrial equipment, which may include valid control commands and noise commands. Therefore, after the raw data is input into the dynamic input gating layer, the dynamic input gating layer can use a classifier to classify the control data into valid control commands and noise commands, filtering out noise commands to obtain the valid data portion of the control data.

[0079] Furthermore, due to the different sources of the raw data, the raw data can include data of various modalities, such as text data, voice data, image data, or sensor data. Therefore, the classifier in the dynamic input gating layer is an industrial multimodal separator. This industrial multimodal separator can classify multimodal control data. For example, if the control data in the raw data includes text data, voice data, and sensor data streams, then the industrial multimodal separator can include a text data classifier, a voice data classifier, and a sensor data stream classifier.

[0080] It should be noted that after the raw data is input into the dynamic input gating layer, it can be split into control data and operational data. The control data is then input into the dynamic input gating layer, where it processes the control data and combines it with the operational data to form the task data. Alternatively, all raw data can be input into the dynamic input gating layer, which then processes only the control data. Furthermore, before inputting the raw data into the dynamic input gating layer, pattern recognition and transformation can be performed. For example, for text data, the dynamic input gating layer converts it to text using ASR, preserving metadata such as intonation and pauses; for speech data, it performs encoding standardization and character set verification; and for sensor data streams, it parses the binary protocol and converts it into structured data.

[0081] In step S22, the industrial protocol refers to the communication and operation specifications followed by the industrial equipment in its industrial environment. Different types of industrial equipment may correspond to different industrial protocols, such as Modbus and Profibus. The constraint unit in the dynamic input gating layer performs syntactic constraints on the valid instructions in the valid data according to the specific industrial protocol corresponding to the industrial equipment. This process ensures that valid instructions conform to the standard syntax that the industrial equipment can recognize and execute, avoiding the industrial equipment's inability to perform tasks correctly due to instruction syntax errors. For example, some instructions may lack necessary parameters in their format or the parameter value range may not conform to the protocol requirements; the constraint unit will correct these instructions or filter them out directly.

[0082] When the constraint device applies syntactic constraints to the valid data according to the industrial protocol corresponding to the industrial equipment, it can perform protocol compliance checks on the valid control instructions included in the valid data according to the industrial protocol, then apply security rules to the valid control instructions that pass the protocol compliance check, and finally fill the default values ​​into the control instructions after the application of security rules.

[0083] The compliance testing process for the protocol can be as follows: First, search for the industrial protocol corresponding to the industrial equipment in the industrial equipment protocol database; Secondly, the effective control commands are validated based on industrial protocols, such as whether they conform to industrial protocols such as Modbus, OPCUA, and Profinet. Finally, check whether the parameter range of the valid control commands that have passed verification is within the allowable range of the device.

[0084] The specific process for applying security rules can be as follows: First, check whether the valid control commands are dangerous combinations, such as simultaneously shutting down multiple safety devices; Secondly, enforce operation sequence constraints, such as requiring a safety check before any operation; Finally, add a timestamp and operator identifier.

[0085] The process of filling in the default value is as follows: First, default values ​​are filled for missing parameter items to ensure the integrity and executability of the instruction. The missing parameter item is an optional parameter item. Secondly, standardize the unit of measurement, such as unifying pressure as MPa.

[0086] Furthermore, in practical applications, the dynamic input gating layer can also include an adversarial filtering module. This module filters adversarial samples from the control data passing through the constraint. The adversarial filtering module's operation includes adversarial feature detection, attack type identification, and a sandbox triggering mechanism. Adversarial feature detection identifies adversarial features in the control data, such as abnormal numerical fluctuations or signals that do not conform to normal patterns. Attack type identification determines the attack type after detecting adversarial features, for example, malicious tampering with instructions, injection of false data, or other types of attack instructions. The sandbox triggering mechanism isolates suspicious control data in a sandbox environment after identifying an attack type, ensuring it does not affect the normal operation of industrial equipment. This embodiment of the application uses an adversarial filtering module to process the control data output by the constraint, effectively resisting external malicious attacks and ensuring the security and reliability of industrial equipment control data.

[0087] This application embodiment employs a dynamic input gating layer to provide in-depth defense against adversarial detection on the raw data, encompassing data cleaning and adversarial detection. Furthermore, it utilizes industrial protocols to impose syntactic constraints on the control data, enhancing the quality and security of the task data and providing a solid and reliable data foundation for subsequent task scheduling and industrial equipment control. The filtered and processed task data is then sent to the task scheduling engine for further generation of atomic tasks used to control the industrial equipment.

[0088] S30. Generate several atomic tasks using the task scheduling engine based on the prior knowledge of the industrial equipment and the task data.

[0089] Specifically, the prior knowledge of industrial equipment includes historical equipment status data, an industrial knowledge graph, and equipment status snapshots. Historical equipment status data can include historical operating data, performance parameters, operating procedures, and other information. The industrial knowledge graph represents the industrial knowledge within the industrial domain in which the equipment operates; it integrates various concepts, entities, and relationships within the industrial domain. The industrial knowledge graph allows for the identification of domain-specific terminology for the industrial equipment. Equipment status snapshots provide detailed status information at the current moment, enabling the calibration of real-time industrial data within the specified environment.

[0090] The task scheduling engine combines prior knowledge of industrial equipment with received task data to perform atomic operations on industrial task chains. Specifically, the engine analyzes the instructions and requirements in the task data, while referencing prior knowledge of the industrial equipment, to break down complex industrial task chains into multiple relatively independent and reasonable sub-atomic tasks. When splitting an industrial task chain into atomic tasks, the engine considers several influencing factors. For example, it ensures that the generated atomic tasks are within the executable range of the industrial equipment, taking into account its actual capabilities. For industrial equipment with limited power, tasks exceeding its power capacity will not be generated. Furthermore, the engine considers the sequence and dependencies between tasks to ensure that the execution order of atomic tasks conforms to the logic of industrial production. For instance, in a production process, the raw material preparation task must be completed before subsequent processing tasks can proceed.

[0091] In one embodiment, such as Figure 5 As shown, the generation of several atomic tasks by the task scheduling engine based on the prior knowledge of the industrial equipment and the task data includes: S31. The current state data of the industrial equipment is determined by the task scheduling engine based on the prior knowledge of the industrial equipment. S32. Generate several atomic tasks using the task scheduling engine based on the current state data and the task data.

[0092] In step S31, the current state data is determined based on historical equipment state data from prior knowledge of the industrial equipment. This historical equipment state data can include accumulated operational data, performance parameters, and operating procedures of the industrial equipment over a past period. The task scheduling engine utilizes data mining and machine learning techniques, combining equipment state snapshots and industrial knowledge graphs, to predict the current state of the industrial equipment. For example, analyzing historical operational data reveals the equipment's operating patterns and common failure modes; leveraging the industrial knowledge graph clarifies the equipment's position and role in the industrial field, thus more accurately determining the current state.

[0093] In one embodiment, such as Figure 6 As shown, the step of determining the current state data of the industrial equipment through the task scheduling engine based on the prior knowledge of the industrial equipment specifically includes: S311. Weight the historical device status data in the prior knowledge to obtain reference device status data; S312. Correct the reference equipment status data based on the industrial knowledge graph in the prior knowledge, and determine the current status data according to the corrected reference equipment status data and the equipment status snapshot in the prior knowledge.

[0094] In step S311, the historical device status data includes all historical device status data collected over a past period, i.e., the historical device status data is a sequence of historical device status data. Therefore, all historical device status data in the historical device status data sequence can be weighted to obtain reference device status data. When weighting the historical device status data, a time-series decay factor can be used as the weighting coefficient; that is, the weighting coefficient of historical device status data closer to the current time is larger, and conversely, the weighting coefficient of historical device status data farther from the current time is smaller.

[0095] This application's embodiments employ a time-series decay factor as a weighting coefficient to weight historical equipment status data, thereby more reasonably reflecting the influence of historical data from different times on the current equipment status. This is because the closer the historical equipment status data is to the current moment, the stronger its correlation with the current equipment status, and the greater its reference value for predicting the current equipment status. For example, in an industrial production scenario, if the operating parameters of an industrial machine fluctuate in the first time period before the current moment, then the operating data of that first time period, compared to the operating data of the second time period before it, can more directly reflect the current performance change trend of the machine. By assigning a larger weighting coefficient to the historical equipment status data within the first time period, the role of the historical equipment status data within the first time period can be better reflected, making the referenced equipment status data more accurately match the actual state of the equipment.

[0096] In step S32, because it integrates various concepts, entities, and relationships within the industrial domain, the industrial knowledge graph can anchor the domain terminology of the industrial equipment. Therefore, the task scheduling engine can compare the reference equipment status data with the industrial knowledge graph to identify potential deviations or non-compliance with industry standards in the reference equipment status data. The industrial knowledge graph can then be used to correct the identified data, improving the accuracy of the corrected reference equipment status data. For example, if the industrial knowledge graph records the value range of parameter B for industrial equipment A, and parameter B in the reference equipment status data exceeds this range, then parameter B in the reference equipment status data can be corrected based on its value range. For instance, the value with the smallest difference between its value range and the parameter value in the reference equipment status data can be used as the corrected parameter value, or the corrected parameter value can be obtained through interaction with the user.

[0097] Furthermore, after obtaining the corrected reference device status data, the corrected reference device status data is combined with the device status snapshot to determine the current status data. Specifically, since the device status snapshot records detailed status information of the device at the current moment, it provides a real-time basis for determining the current status data. Therefore, the corrected reference device status data and the status information provided by the device status snapshot can be weighted to obtain the current status data.

[0098] Furthermore, in step S32, the atomic task is the smallest executable unit of the industrial task corresponding to the task data. For example, if the industrial task is a hazardous gas over-limit disposal task, and the task chain of the hazardous gas over-limit disposal task is sensor reading → data verification → ventilation start → personnel evacuation, then the atomic task includes sensor reading, data verification, ventilation start, and personnel evacuation. As another example, if the industrial task is a substation equipment inspection task, and the task chain of the substation equipment inspection task is infrared temperature measurement → partial discharge analysis → anomaly location, then the atomic task includes infrared temperature measurement, partial discharge analysis, and anomaly location.

[0099] Task data is used to define the goals and requirements of industrial tasks, while the current status data of industrial equipment reflects the current actual conditions and operating status of the equipment. The task scheduling engine, based on the current status data and task data, decomposes the industrial task chain into several atomic tasks, constrained by the smallest executable unit of the equipment. The task scheduling engine can be a trained deep learning module, whose inputs are the current status data and task data, whose constraint is the smallest executable unit of the equipment, and whose output is the atomic task. For example, if the task data requires handling hazardous gas exceeding limits, and the current status data shows that sensors, data detectors, and ventilation equipment are all functioning normally, then the task scheduling engine will break down the task chain for handling hazardous gas exceeding limits (i.e., sensor reading → data verification → ventilation start → personnel evacuation) into sensor reading, data verification, ventilation start, and personnel evacuation tasks. In other words, the generated atomic tasks include sensor reading, data verification, ventilation start, and personnel evacuation tasks.

[0100] In one embodiment, such as Figure 7 As shown, step S32, "generating several atomic tasks based on the current state data and the task data using the task scheduling engine," specifically includes: S321. Construct a candidate decomposition set based on the task data; S322. Construct a resource consumption function with the goal of minimizing resource consumption; S323. Using equipment capacity as a constraint, solve the resource consumption function on the candidate decomposition set based on the current state data to obtain several atomic tasks.

[0101] Specifically, the candidate decomposition set is a decomposition database constructed based on the current state data and task data, corresponding to the target tasks. Each candidate decomposition in this set represents a decomposition method for the target task; that is, the candidate decomposition set includes multiple candidate decompositions, and each candidate decomposition includes several atomic tasks and their execution order. To better represent the execution order between atomic tasks, the candidate decomposition can be represented as a directed graph, i.e., a directed graph composed of several atomic tasks and their execution order.

[0102] Furthermore, such as Figure 8 As shown, the specific process of constructing a candidate decomposition set based on the task data can be as follows: M10. Construct a task target map based on task data; M20. Based on the task target graph, enumerate all atomic task combinations in the preset atomic operation library that can realize the target task corresponding to the task data to obtain the candidate decomposition set.

[0103] Specifically, the task target graph is obtained by structuring task data. It includes the final goal to be achieved by the target task, the entities involved (such as equipment, tools, and data), and the sub-goals included in the target task. The preset atomic operation library is a pre-configured set of basic tasks that can be executed by a device.

[0104] After obtaining the preset atomic operation library and task target graph, all atomic task combinations in the preset atomic operation library that can implement the target task corresponding to the task data are obtained through enumeration. Then, a directed graph is generated based on each atomic task combination to obtain a candidate decomposition set. During the construction of the candidate decomposition set, the logical relationships and feasibility between atomic tasks are considered, and combinations that do not conform to the actual situation of industrial production or are unexecutable are eliminated. For example, if an atomic task requires a specific device to complete, but the current industrial equipment does not include that device, then the candidate decomposition containing that atomic task will be excluded. Of course, in practical applications, other methods can also be used to build the candidate decomposition set, such as heuristic algorithms or rule engines.

[0105] Furthermore, this candidate decomposition set is used as a candidate solution set for the resource consumption function. Solving this resource consumption function within the candidate decomposition set yields several atomic tasks. The resource consumption function is constructed with the goal of minimizing resource consumption. It may include a term representing the gap between total resource requirements and the actual capabilities of the equipment. This gap term ensures that resource consumption matches equipment capabilities, thereby pre-estimating the memory / computing power consumption required for each task. In addition, to avoid deadlocks caused by dependencies between multiple industrial devices, the resource consumption function may also include a dependency conflict indicator function term. This dependency conflict indicator function term is used to identify circular dependencies and / or resource conflicts between atomic tasks.

[0106] Based on this, in a specific implementation, the resource consumption function is: , in, This represents the task chain formed by atomic tasks. Represents a task set. Represents the resource consumption matrix. Represents the device capability vector. Represents tasks within a task set. Indicates the conflict adjustment factor. This indicates a dependency conflict indicator function item.

[0107] Furthermore, after obtaining the resource consumption function and candidate decomposition set, the resource consumption function is solved on the candidate decomposition set with device capabilities as a constraint to obtain several atomic tasks. This ensures that the obtained atomic tasks run most efficiently within resource constraints, with clear internal dependencies and no deadlock risk. In solving the source consumption function, optimization algorithms such as genetic algorithms and simulated annealing algorithms can be used to search for the combination of atomic tasks that minimizes the resource consumption function value in the candidate decomposition set.

[0108] In one embodiment, such as Figure 9 As shown, after generating several atomic tasks through the task scheduling engine based on the prior knowledge of the industrial equipment and the task data, the method further includes: H10. Extract the drift features of the atomic task and predict the risk probability of the atomic task based on the drift features; H20. When there is a first target atomic task with a risk probability greater than a preset risk probability threshold, obtain the drift compensation of the first target atomic task. H30. Feedback the drift compensation to the dynamic input gating layer so that the dynamic input gating layer can perform offline updates based on the drift compensation.

[0109] In step H10, the drift characteristics include one or more of the following: output variance, latency gradient, and tool call error rate. These characteristics reflect potential performance fluctuations or deviations from expectations during the execution of the atomic task. Output variance reflects the dispersion of the atomic task's output; a larger output variance indicates poorer output stability and potentially greater execution risk. Latency gradient reflects the trend of atomic task execution time; a larger latency gradient indicates unstable atomic task execution time, potentially affecting the overall progress of the industrial task. Tool call error rate reflects the proportion of errors that occur when the atomic task calls relevant tools; a higher error rate indicates lower atomic task reliability.

[0110] Risk probability reflects the likelihood of risk faced by an atomic task during execution, and it is predicted based on extracted drift features. Specifically, risk prediction models can be constructed using algorithms such as machine learning or deep learning. In other words, the process of predicting the risk probability of an atomic task based on the drift features involves inputting the drift features into a trained risk prediction model, which then outputs the risk probability of the original task.

[0111] In step H20, if the risk probability exceeds a preset risk probability threshold, it indicates that the first target atomic task corresponding to that risk probability has a risk and requires drift compensation for that first target atomic task. Drift compensation refers to parameters or strategies used to correct deviations caused by drift characteristics during the execution of atomic tasks. For example, if the output variance is too large, it can be reduced by adjusting task parameters or optimizing the execution process; if the latency variation gradient is large, the task scheduling strategy can be adjusted to allocate resources reasonably and ensure the stability of task execution time; if the tool call error rate is high, the tool can be checked, maintained, or updated.

[0112] In step H30, after the drift compensation of the first target atomic task, the first target atomic task and its corresponding drift compensation are fed back to the dynamic input gating layer, so that the dynamic input gating layer can dynamically adjust the neural network layer of its own parameters, thereby better adapting to the changes in the atomic task and improving the stability and reliability of task execution.

[0113] S40. Control the industrial equipment to perform several atomic tasks.

[0114] Specifically, after acquiring several atomic tasks, control instructions can be generated for each atomic task, and the corresponding industrial equipment can be controlled through these instructions to execute the atomic task. Of course, when controlling the industrial equipment to execute several atomic tasks, each atomic task will be executed in the order of execution to achieve the target task corresponding to the task data. Furthermore, when controlling the industrial equipment to execute atomic tasks, the industrial equipment required for each atomic task will be determined first, and then the required industrial equipment will be controlled.

[0115] In one embodiment, such as Figure 10 As shown, controlling the industrial equipment to perform several atomic tasks specifically includes: S41. Encode the atomic task into industrial control instructions using the target resource device; S42. Send the industrial control instructions to the industrial equipment so that the industrial equipment performs several atomic tasks.

[0116] Specifically, a target resource device is a device with specific coding capabilities. It can transform atomic tasks into industrial control instructions that can be recognized and executed by industrial equipment, based on the specific requirements of the atomic tasks and the control rules of the industrial equipment. During the coding operation, the target resource device analyzes various parameters of the atomic task, such as the task's execution time, action requirements, and cooperation logic with other atomic tasks, to ensure that the generated industrial control instructions can accurately convey the intent of the atomic task.

[0117] After industrial control commands are generated, they are sent to the corresponding industrial equipment via a suitable communication network, such as industrial Ethernet or CAN bus. To ensure the security of these commands, a Cyclic Redundancy Check (CRC) code and a device digital signature are appended to them before transmission. Upon receiving the command with the CRC code and digital signature, the industrial equipment verifies both. It checks the CRC code's correctness and the digital signature's validity based on pre-stored verification rules and algorithms. If verification fails, the equipment refuses to execute the command and sends an error message to the target resource device, which then regenerates and resends the correct command. If verification succeeds, the equipment parses the command and translates it into specific operational actions. During the execution of atomic tasks, the industrial equipment monitors its own operating status and task execution in real time. For example, it monitors parameters such as temperature, pressure, and rotational speed to ensure the equipment operates within its normal operating range.

[0118] In one embodiment, to further improve the success rate of trusted execution of the task chain corresponding to the target task, before controlling the industrial equipment to execute several atomic tasks, such as... Figure 11 As shown, the method further includes: A10. Perform dual-path redundancy operations on the atomic task to obtain the execution result of each execution path; A20. Determine the confidence level of the atomic task based on the consistency of the execution results of the two execution paths; A30. For a second target atomic task with a confidence level lower than a preset confidence threshold, the execution result of the second target atomic task is determined using a rule engine.

[0119] Specifically, dual-path redundancy operation refers to simultaneously launching two independent execution paths to execute the same atomic task, thereby reducing the possibility of task failure due to a single point of failure. Each of the two independent execution paths records its own execution result, which may include various key information during the execution of the atomic task, such as task completion time, output data, and execution status.

[0120] After performing dual-path redundancy, the confidence level of the atomic task is determined based on the consistency of the execution results of the two execution paths. Specifically, when the execution results of the two execution paths are the same, it indicates that the atomic task performs stably under the two independent paths, has high reliability, and a correspondingly high confidence level. Conversely, if the execution results of the two execution paths are different, it indicates that the execution of the atomic task may be affected by some uncertain factors, and the confidence level is low.

[0121] For a second target atomic task with a confidence level below a pre-set confidence threshold, a rule engine is needed to further determine its execution result. A rule engine is a system that performs rule-based reasoning and decision-making. It analyzes the execution of the second target atomic task based on pre-defined rules and logic. The rule engine stores a series of rules related to industrial tasks and equipment, covering various normal and abnormal situations in task execution. When faced with a second target atomic task, the rule engine performs comprehensive reasoning based on various parameters of the atomic task, data during execution, and pre-defined rules to arrive at a more accurate and reliable execution result.

[0122] The embodiments of this application can effectively improve the success rate of trusted execution of the task chain corresponding to the target task by performing dual-path redundancy operations on atomic tasks, determining the confidence level, and using a rule engine to handle low-confidence tasks, thereby ensuring the stability and reliability of the industrial production process.

[0123] Furthermore, in practical applications, when a second target atomic task with a confidence level lower than a preset confidence threshold exists, this second target atomic task can be fed back to the task scheduling engine for dynamic compensation and real-time updates. In this way, the task scheduling engine can continuously learn and improve by receiving feedback information from low-confidence atomic tasks, gradually enhancing the accuracy and rationality of task scheduling.

[0124] In one embodiment, before acquiring the raw data from the industrial equipment, such as Figure 12 As shown, the method further includes: C10. Obtain the candidate resource device set; C20. Construct an objective function that includes measures of resource cost and service quality degradation. C30. Based on the objective function, select the target resource device corresponding to each atomic task from the candidate resource device set.

[0125] Specifically, the candidate resource device set can include edge computing points and cloud routers. The objective function is used to comprehensively evaluate the resource cost and service quality degradation of the resource devices. Resource cost mainly involves economic factors such as device usage fees and maintenance costs, while the service quality degradation metric is used to consider the device's performance in terms of response time, processing capacity, and data transmission stability.

[0126] When selecting the target resource device for each atomic task from the candidate resource device set based on the objective function, the objective function is solved on the candidate resource device set with the goal of minimizing the comprehensive index formed by resource cost and service quality degradation metric, in order to select the target resource device for each atomic task. Intelligent optimization algorithms such as genetic algorithms and particle swarm optimization can be used to solve the objective function. The comprehensive index can be formed by weighting resource cost and service quality degradation metric.

[0127] This application embodiment comprehensively considers resource cost and service quality degradation measurement, which can avoid the problem of increased economic cost of the entire industrial production process due to excessively high resource cost of target resource equipment, and can also avoid the problem of low execution efficiency of atomic tasks due to severe service quality degradation, thus ensuring the efficient and stable execution of atomic tasks.

[0128] Based on the above-described trusted execution method for task chains, this embodiment provides an apparatus for a trusted execution method for task chains, such as... Figure 13 As shown, the task chain trusted execution method apparatus specifically includes: Module 100 is used to acquire raw data from industrial equipment. The first processing module 200 is used to filter the raw data through a dynamic input gating layer to obtain task data; The second processing module 300 generates several atomic tasks based on the prior knowledge of the industrial equipment and the task data through a task scheduling engine. The control module 400 is used to control the industrial equipment to perform several atomic tasks.

[0129] Based on the above-described trusted execution method for task chains, this embodiment provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, which are used to execute the trusted execution method for task chains described above.

[0130] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system or device, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, and portable compact disk read-only memory (CD-ROM). ROM: CD Read-only memory, optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in connection with an instruction execution system or device. The program code contained on the computer-readable storage medium can be transmitted using any suitable medium, including but not limited to: wires, optical fibers, RF (Radio Frequency), etc., or any suitable combination thereof.

[0131] The aforementioned computer-readable storage medium may be included in an electronic device or may exist independently without being assembled into an electronic device.

[0132] The aforementioned computer-readable storage medium carries one or more programs. When the aforementioned one or more programs are executed by an electronic device, they acquire raw data from the industrial equipment; filter the raw data through a dynamic input gating layer to obtain task data; generate several atomic tasks through a task scheduling engine based on the prior knowledge of the industrial equipment and the task data; and control the industrial equipment to execute the several atomic tasks.

[0133] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, and conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a Local Area Network (LAN) or a Wide Area Network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0134] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0135] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0136] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the above-described trusted execution method of the task chain, thereby solving the technical problem of poor application performance. Compared with the prior art, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the trusted execution method of the task chain provided in the above embodiments, and will not be repeated here.

[0137] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the steps of the task chain trusted execution method described above.

[0138] The computer program product provided in this application can solve the technical problem of poor application performance. Compared with the prior art, the beneficial effects of the computer program product provided in this application are the same as those of the task chain trusted execution method provided in the above embodiments, and will not be repeated here.

[0139] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application.

Claims

1. A method for trusted execution of a task chain, characterized in that, The trusted execution method for the task chain specifically includes: Acquire raw data from industrial equipment; The raw data is filtered through a dynamic input gating layer to obtain task data; The task scheduling engine generates several atomic tasks based on the prior knowledge of the industrial equipment and the task data. Control the industrial equipment to perform several atomic tasks.

2. The task chain trusted execution method according to claim 1, characterized in that, The raw data includes control data of industrial equipment and task requirements data for the target task.

3. The task chain trusted execution method according to claim 1, characterized in that, The process of filtering the raw data through a dynamic input gating layer to obtain task data specifically includes: The control data in the original data is classified by a classifier in the dynamic input gating layer to obtain effective data; By using the constraint mechanism in the dynamic input gating layer to apply syntactic constraints to the valid data according to the industrial protocol corresponding to the industrial equipment, task data is obtained.

4. The task chain trusted execution method according to claim 1, characterized in that, The prior knowledge includes historical equipment status data of industrial equipment, industrial knowledge graphs, and equipment status snapshots.

5. The task chain trusted execution method according to claim 1 or 4, characterized in that, The step of generating several atomic tasks through a task scheduling engine based on prior knowledge of the industrial equipment and the task data specifically includes: The task scheduling engine determines the current status data of the industrial equipment based on the prior knowledge of the industrial equipment. The task scheduling engine generates several atomic tasks based on the current state data and the task data.

6. The task chain trusted execution method according to claim 5, characterized in that, The step of determining the current state data of the industrial equipment through the task scheduling engine based on the prior knowledge of the industrial equipment specifically includes: The historical device status data in the prior knowledge is weighted to obtain reference device status data, wherein the weighting coefficient of the historical device status data is a time decay factor. The reference equipment status data is corrected based on the industrial knowledge graph in the prior knowledge, and the current status data is determined based on the corrected reference equipment status data and the equipment status snapshot in the prior knowledge.

7. The task chain trusted execution method according to claim 5, characterized in that, The step of generating several atomic tasks based on the current state data and the task data through the task scheduling engine specifically includes: A candidate decomposition set is constructed based on the task data, wherein each candidate decomposition in the candidate decomposition set is a directed graph consisting of several atomic tasks and their execution order. Construct a resource consumption function with the goal of minimizing resource consumption; The resource consumption function is solved on the candidate decomposition set based on the current state data, with equipment capacity as a constraint, to obtain several atomic tasks.

8. The task chain trusted execution method according to claim 7, characterized in that, The resource consumption function includes a dependency conflict indicator function, which is used to identify circular dependencies and / or resource conflicts between atomic tasks.

9. The task chain trusted execution method according to claim 8, characterized in that, The resource consumption function is: ， in, This indicates a goal decomposition that includes several atomic tasks. Represents the candidate decomposition set. Represents the resource consumption matrix. Represents the device capability vector. Indicates candidate decomposition, Indicates the conflict adjustment factor. This indicates a dependency conflict indicator function item.

10. The task chain trusted execution method according to claim 1, characterized in that, After generating several atomic tasks through a task scheduling engine based on prior knowledge of the industrial equipment and the task data, the method further includes: Extract the drift features of atomic tasks and predict the risk probability of atomic tasks based on the drift features; When there is a first target atomic task with a risk probability greater than a preset risk probability threshold, obtain the drift compensation of the first target atomic task; The drift compensation is fed back to the dynamic input gating layer so that the dynamic input gating layer can perform offline updates based on the drift compensation.

11. The task chain trusted execution method according to claim 10, characterized in that, The drift characteristics include at least one of the following: output variance, time delay variation gradient, or tool call error rate.

12. The task chain trusted execution method according to claim 10, characterized in that, Specifically, the prediction of the risk probability of the atomic task based on the drift feature is as follows: The drift features are input into a trained risk prediction model, which then outputs the risk probability of the original task.

13. The task chain trusted execution method according to claim 1, characterized in that, Before controlling the industrial equipment to perform several atomic tasks, the method further includes: Perform dual-path redundancy operations on the atomic task to obtain the execution result of each execution path; The confidence level of the atomic task is determined based on the consistency of the execution results of the two execution paths. For a second target atomic task with a confidence level lower than a preset confidence threshold, the execution result of the second target atomic task is determined using a rule engine.

14. The task chain trusted execution method according to claim 13, characterized in that, The method further includes: When there is a second target atomic task with a confidence level lower than the preset confidence threshold, the second target atomic task is fed back to the task scheduling engine to provide a dynamic compensation version for the task scheduling engine.

15. The task chain trusted execution method according to claim 1, characterized in that, Before acquiring the raw data of the industrial equipment, the method further includes: Obtain the candidate resource device set; Construct an objective function that includes measures of resource cost and service quality degradation; Based on the objective function, the target resource device corresponding to each atomic task is selected from the candidate resource device set.

16. The task chain trusted execution method according to claim 15, characterized in that, The control of the industrial equipment to perform several atomic tasks specifically includes: The atomic tasks are encoded into industrial control instructions using the target resource equipment; The industrial control instructions are sent to the industrial equipment so that the industrial equipment performs several atomic tasks.

17. The task chain trusted execution method according to claim 16, characterized in that, After encoding the atomic task into industrial control instructions using the target resource device, the method further includes: Add a cyclic redundancy check code and a device digital signature to the industrial control command.

18. A method and apparatus for trusted execution of a task chain, characterized in that, The aforementioned task chain trusted execution method apparatus specifically includes: The acquisition module is used to acquire raw data from industrial equipment; The first processing module is used to filter the raw data through a dynamic input gating layer to obtain task data; The second processing module generates several atomic tasks based on the prior knowledge of the industrial equipment and the task data through the task scheduling engine. The control module is used to control the industrial equipment to perform several atomic tasks.

19. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores one or more programs, which can be executed by one or more processors to implement the steps in the task chain trusted execution method as described in any one of claims 1 to 17.

20. An electronic device, characterized in that, include: Processor and memory; The memory stores a computer-readable program that can be executed by the processor; When the processor executes the computer-readable program, it implements the steps of the task chain trusted execution method as described in any one of claims 1 to 17.

21. A computer program product, characterized in that, The computer program product includes a computer program that, when executed by a processor, implements the steps of the task chain trusted execution method as described in any one of claims 1 to 17.

Images