An edge-computing-based embedded architecture control system

Abstract

This invention discloses an embedded architecture control system based on edge computing, comprising: a data processing module for acquiring and preprocessing control data from multiple embedded control nodes; a topology construction module for constructing an edge collaborative control topology, deploying control units, and establishing channels; a state analysis module for generating node state characterization results and writing them into a node state cache; a local control module for generating a set of local control instructions and outputting them to the actuators; a collaborative allocation module for generating candidate collaborative evaluation data and determining a task allocation result set based on a contract network protocol and a strict bilateral matching method; a strategy update module for obtaining an updated control parameter set and an updated control strategy set; and an effect evaluation module for generating control result records and collaborative state records. This invention achieves embedded node collaborative control through edge collaborative control and contract network task allocation, offering advantages such as high real-time performance and strong stability.

Description

Technical Field

[0001] This invention relates to the field of embedded control technology, and in particular to an embedded architecture control system based on edge computing. Background Technology

[0002] With the development of industrial automation systems and intelligent devices, embedded control systems are widely used in production control, equipment scheduling, and intelligent monitoring. Existing embedded control systems typically consist of multiple embedded control nodes that use communication networks to acquire equipment operation data, issue control commands, and monitor equipment status. In some systems, edge computing technology is introduced to improve system efficiency, enabling some data processing and control decisions to be completed on nodes closer to the equipment, thereby reducing the central processing load and improving response speed.

[0003] Existing embedded control systems are mostly based on single-node control or simple distributed control, with limited coordination capabilities between control nodes. In multi-node collaborative control scenarios, there is often a lack of a unified collaborative task allocation mechanism and a stable node collaborative decision-making method, resulting in low control task scheduling efficiency. At the same time, there is a lack of a systematic consistency verification and effect evaluation mechanism between node control states and execution results. In complex control environments, the accumulation of control result deviations and imbalance of collaborative states can easily occur, affecting the stability and collaborative control effect of the overall embedded control system.

[0004] Therefore, how to provide an embedded architecture control system based on edge computing is a problem that urgently needs to be solved by those skilled in the art. Summary of the Invention

[0005] One objective of this invention is to propose an embedded architecture control system based on edge computing. This invention achieves embedded node collaborative control through edge collaborative control and contract network task allocation, which has the advantages of high real-time performance and strong stability.

[0006] An embedded architecture control system based on edge computing according to an embodiment of the present invention includes: The data processing module is used to acquire control data from multiple embedded control nodes and preprocess it according to a unified time base to generate a standardized input dataset for each embedded control node. The topology construction module is used to build an edge collaborative control topology based on the basic information of each embedded control node, generate network topology status information, deploy edge computing processing units, control strategy calculation units, execution control interface units and node communication units in each embedded control node, and establish state sharing channels and task collaboration channels between nodes. The state analysis module is used to perform real-time feature extraction and state analysis on the standardized input dataset, generate node state representation results and write them to the corresponding node state cache. The local control module is used to calculate the local control strategy based on the node state representation results in the node state buffer, generate a set of local control instructions, and output them to the corresponding actuators through the execution control interface unit to complete real-time closed-loop control on the edge side. The collaborative allocation module is used to generate candidate collaborative evaluation data by exchanging node status representation results and network topology status information through node communication units, filter candidate collaborative nodes based on the candidate collaborative evaluation data using the contract network protocol, and determine the task allocation result set through a strict bilateral matching method. The strategy update module is used to generate adaptive control strategies based on the standardized input dataset, node state representation results, and task allocation result set, to obtain the updated control parameter set and the updated control strategy set, and load them into the control strategy calculation unit. The effect evaluation module is used to perform consistency verification and control effect evaluation on the control response data returned by the execution control interface unit, and generate control result records and collaborative status records.

[0007] Optionally, the control data acquired by the data processing module includes equipment operation data, environmental perception data, and node communication status data. The preprocessing includes timestamp alignment, outlier removal, missing value completion, unit normalization, and data segmentation.

[0008] Optionally, the topology building module includes: Collect basic information about each embedded control node, including hardware resource information, control object identifier, network connection relationship and control task type; Based on the hardware resource information of each embedded control node, the node resource evaluation result of each embedded control node is calculated. Based on the control object identifier and control task type of each embedded control node, calculate the task adaptation evaluation result for each embedded control node. Based on network connectivity, the communication link status between any two embedded control nodes is analyzed, and the link collaboration evaluation results between nodes are calculated. Based on the node resource evaluation results, task adaptation evaluation results, and link collaboration evaluation results, the node connection priority between each embedded control node is determined, and a node adjacency table, a link status association table, and a task collaboration association table are established according to the node connection priority. Topology connection edges are generated for node pairs that meet the preset collaboration establishment conditions. The edge collaborative control topology is constructed based on the node adjacency table, link state association table and task collaboration association table. The node connection results between each embedded control node are determined, and the corresponding link state information and task collaboration state information are marked on each topology connection edge to obtain the network topology state information. Based on the node connection results and task collaboration association table of the edge collaborative control topology, an edge computing processing unit, a control strategy calculation unit, an execution control interface unit, and a node communication unit are deployed in each embedded control node, and a state sharing channel and a task collaboration channel between nodes are established based on the topology connection edges.

[0009] Optionally, the state analysis module includes: Each embedded control node calls the edge computing processing unit to read the standardized input dataset, and performs synchronous segmentation according to a unified time window to generate a set of data segments corresponding to each time window; Extract device operation characteristics, environmental status characteristics, and communication status characteristics from data segments for each time window; Based on the device operation characteristics and environmental state characteristics, the current operating state of the embedded control node is comprehensively determined to obtain the local state characterization result of the node. Based on the device's operating characteristics and communication status characteristics, the resource occupancy of the embedded control node is evaluated to obtain the node load characterization results. Based on the characteristics of device operation, environmental status, and communication status, the control requirements of embedded control nodes are identified to obtain control requirement characterization results. The node local state representation results, node load representation results, and control requirement representation results are fused together to generate node state representation results, which are then written into the corresponding node state cache in the order of the time window.

[0010] Optionally, the local control module includes: Each embedded control node reads the node state representation result corresponding to the current control cycle from the corresponding node state buffer, and performs association mapping on the node state representation result according to the control object identifier to generate local control input data for the corresponding control object; Based on local control input data, control state variables, load constraint variables and control demand variables are extracted and local control strategies are calculated to generate a set of local control parameters. The control command parameter values ​​are determined based on the local control parameter set and the control object identifier, and the control command parameter values ​​are adapted according to the execution interface type corresponding to the control object to generate a local control command set. The local control instruction set is written into the corresponding execution control interface unit, and the local control instruction set is output to the corresponding execution mechanism according to the instruction output order to drive the corresponding execution mechanism to perform control actions; The interface feedback information returned by the execution control interface unit and the action feedback information returned by the actuator are collected, and the consistency of the output state of the local control command set is verified based on the interface feedback information and action feedback information to obtain the local control response result. Based on the local control response results, the local control parameter set is corrected in a closed loop to generate a corrected local control parameter set. The corrected local control parameter set is then written back to the corresponding node state buffer to form real-time closed-loop control on the edge side of the corresponding controlled object.

[0011] Optionally, the collaborative allocation module includes: Each embedded control node sends its node status representation results and network topology status information to neighboring embedded control nodes through the node communication unit. It receives, correlates, and organizes the node status representation results and network topology status information sent by neighboring embedded control nodes to generate candidate collaborative evaluation data. Based on candidate collaborative evaluation data, the Contract Network protocol is used to evaluate the collaborative capabilities of adjacent embedded control nodes and generate a set of candidate collaborative nodes. Read the set of control tasks to be assigned, and parse the task attributes of each control task according to the control object identifier, control task type, control cycle requirements, resource call requirements and communication connection constraints to generate a set of task matching attributes; Based on the candidate collaborative node set and the task matching attribute set, a strict two-sided matching method is used to match the candidate collaborative node set with the set of control tasks to be assigned, and a task assignment result set is generated. The control tasks are dynamically adjusted based on the task allocation result set. The dynamically adjusted control task allocation status is written into the task collaboration record, and the task allocation result set is sent to the corresponding candidate collaboration node.

[0012] Optionally, the policy update module includes: The standardized input dataset, node state representation results, and task allocation result set are correlated and organized according to the control object identifier, task identifier, and node identifier to obtain the policy update input data. Based on the policy update input data, the changes in running status, task coordination, and resource scheduling are extracted and policy adaptability analysis is performed to obtain the state adjustment quantity, coordination compensation quantity, and resource constraint correction quantity. Based on the state adjustment quantity, the cooperative compensation quantity, and the resource constraint correction quantity, an adaptive control strategy generation process is performed to obtain an updated control parameter set. Based on the updated control parameter set and the task allocation result set, the local control policy is reconstructed to obtain the updated control policy set. A consistency check is performed on the updated control parameter set and the updated control strategy set. The updated control parameter set and the updated control strategy set that pass the consistency check are then loaded into the control strategy calculation unit.

[0013] Optionally, the effect evaluation module includes: The control response data returned by the execution control interface unit is collected and associated with the control object identifier, node identifier and control cycle identifier to obtain the response verification input data. Based on the response verification input data, extract the instruction execution status quantity, the device operation result status quantity, and the node collaborative response status quantity, and perform consistency verification to obtain the consistency verification result; The control effect is evaluated based on the consistency verification results and the response verification input data to obtain the control effect evaluation results. Based on the consistency verification results and control effect evaluation results, control result records are generated. Based on the control result records, the collaborative response state identifiers corresponding to the node collaborative response state quantities and the collaborative deviation identifiers corresponding to the collaborative deviation values ​​are extracted to generate collaborative state records.

[0014] The beneficial effects of this invention are: This invention provides an edge computing-based embedded architecture control system. By constructing an edge collaborative control topology among multiple embedded control nodes and deploying edge computing processing units, control strategy calculation units, execution control interface units, and node communication units within each embedded control node, control data can be processed and analyzed in real time at the node side. By preprocessing the control data from multiple embedded control nodes under a unified time reference to generate a standardized input dataset, and extracting node state features based on this standardized input dataset, node state representation results are generated and written to the node state buffer. This enables each embedded control node to make real-time control decisions based on local state information, thereby reducing dependence on the central control system and improving the real-time response capability and edge-side control efficiency of the embedded control system.

[0015] In multi-node collaborative control, this invention generates candidate collaborative evaluation data by exchanging node state representation results and network topology state information through node communication units. It then employs a contract network protocol to evaluate the collaborative capabilities of adjacent embedded control nodes, screening candidate collaborative nodes. Furthermore, a rigorous bilateral matching method is used to match the set of candidate collaborative nodes with the set of control tasks to be assigned, generating a task allocation result set. This allows control tasks to be dynamically and collaboratively allocated based on node real-time status, resource capacity, and communication connection conditions. Through this mechanism, a stable collaborative task allocation relationship can be formed among embedded control nodes, improving the matching rationality and execution efficiency of multi-node collaborative control tasks, thereby enhancing the collaborative control capabilities of the embedded control system in complex control environments.

[0016] Meanwhile, this invention performs consistency verification and control effect evaluation on the control response data returned by the execution control interface unit, generates control result records and collaborative status records, and performs adaptive control strategy updates by combining the task allocation result set and node status characterization results. This allows control parameters and control strategies to be dynamically adjusted according to node operating status, task collaboration relationships, and control execution effects. By forming a closed-loop control mechanism between node status analysis, local control decision-making, collaborative task allocation, and control effect evaluation, the accumulation of control deviations can be effectively reduced, the adaptability and stability of the control strategy can be improved, thereby enhancing the operational reliability and collaborative control effect of the entire embedded architecture control system. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a schematic diagram of the structure of an embedded architecture control system based on edge computing proposed in this invention; Figure 2 This is a local control flowchart of an embedded architecture control system based on edge computing proposed in this invention. Figure 3 This is a flowchart illustrating the collaborative task allocation process of an embedded architecture control system based on edge computing proposed in this invention. Detailed Implementation

[0018] The present invention will now be described in further detail with reference to the accompanying drawings. These drawings are simplified schematic diagrams, illustrating only the basic structure of the invention, and therefore only show the components relevant to the invention.

[0019] refer to Figures 1-3 An edge computing-based embedded architecture control system includes: The data processing module is used to acquire control data from multiple embedded control nodes and preprocess it according to a unified time base to generate a standardized input dataset for each embedded control node. The topology construction module is used to build an edge collaborative control topology based on the basic information of each embedded control node, generate network topology status information, deploy edge computing processing units, control strategy calculation units, execution control interface units and node communication units in each embedded control node, and establish state sharing channels and task collaboration channels between nodes. The state analysis module is used to perform real-time feature extraction and state analysis on the standardized input dataset, generate node state representation results and write them to the corresponding node state cache. The local control module is used to calculate the local control strategy based on the node state representation results in the node state buffer, generate a set of local control instructions, and output them to the corresponding actuators through the execution control interface unit to complete real-time closed-loop control on the edge side. The collaborative allocation module is used to generate candidate collaborative evaluation data by exchanging node status representation results and network topology status information through node communication units, filter candidate collaborative nodes based on the candidate collaborative evaluation data using the contract network protocol, and determine the task allocation result set through a strict bilateral matching method. The strategy update module is used to generate adaptive control strategies based on the standardized input dataset, node state representation results, and task allocation result set, to obtain the updated control parameter set and the updated control strategy set, and load them into the control strategy calculation unit. The effect evaluation module is used to perform consistency verification and control effect evaluation on the control response data returned by the execution control interface unit, and generate control result records and collaborative status records.

[0020] In this embodiment, the control data acquired by the data processing module includes equipment operation data, environmental perception data, and node communication status data. Preprocessing includes timestamp alignment, outlier removal, missing value completion, unit normalization, and data segmentation.

[0021] In this embodiment, the topology construction module includes: Collect basic information of each embedded control node. The basic information includes hardware resource information, control object identification, network connection relationship and control task type. Among them, hardware resource information includes processor computing power, storage capacity, number of interfaces and power supply status. Network connection relationship includes physical connection method between nodes, communication link number, reachability, communication latency, link bandwidth, packet loss rate and link jitter. Based on the hardware resource information of each embedded control node, the node resource evaluation result of each embedded control node is calculated. The node resource evaluation result is generated by weighting and summarizing the resource value corresponding to the processor computing power, the resource value corresponding to the storage capacity, the resource value corresponding to the number of interfaces, and the stability resource value corresponding to the power supply status. Based on the control object identifier and control task type of each embedded control node, the task adaptation evaluation result of each embedded control node is calculated. The task adaptation evaluation result is generated by combining the degree of matching of control object type, control cycle, execution interface and task load. Based on the network connection relationship, the communication link status between any two embedded control nodes is analyzed, and the link collaboration evaluation result between nodes is calculated. The link collaboration evaluation result is generated by weighting and summing the timeliness evaluation value corresponding to communication delay, the transmission capacity evaluation value corresponding to link bandwidth, the reliability evaluation value corresponding to packet loss rate, and the stability evaluation value corresponding to link jitter. Based on the node resource evaluation results, task adaptation evaluation results, and link collaboration evaluation results, the node connection priority between each embedded control node is determined, and a node adjacency table, a link status association table, and a task collaboration association table are established according to the node connection priority. Topology connection edges are generated for node pairs that meet the preset collaboration establishment conditions. The edge collaborative control topology is constructed based on the node adjacency table, link state association table and task collaboration association table. The node connection results between each embedded control node are determined, and the corresponding link state information and task collaboration state information are marked on each topology connection edge to obtain the network topology state information. Based on the node connection results and task collaboration association table of the edge collaborative control topology, an edge computing processing unit, a control strategy calculation unit, an execution control interface unit, and a node communication unit are deployed in each embedded control node, and a state sharing channel and a task collaboration channel between nodes are established based on the topology connection edges.

[0022] In this embodiment, the state analysis module includes: Each embedded control node calls the edge computing processing unit to read the standardized input dataset, and synchronously segments the device operation data, environmental perception data and node communication status data according to a unified time window to generate a set of data segments corresponding to each time window; For each time window's data segment set, extract device operation characteristics, environmental status characteristics, and communication status characteristics. Among them, device operation characteristics include the average value of operating parameters, the fluctuation range of operating parameters, and the rate of change of operating parameters; environmental status characteristics include the average value of environmental parameters, the trend of environmental parameter changes, and the intensity of environmental disturbances; and communication status characteristics include link latency level, link jitter level, packet loss level, and bandwidth utilization level. Based on the device operation characteristics and environmental state characteristics, the current operation state of the embedded control node is comprehensively determined to obtain the node local state characterization result. The node local state characterization result is generated by combining the operation state quantity corresponding to the device operation characteristics, the environmental influence quantity corresponding to the environmental state characteristics, and the coupling influence quantity between the device operation characteristics and the environmental state characteristics. Based on the device's operating characteristics and communication status characteristics, the resource occupancy of the embedded control node is evaluated to obtain the node load characterization result. The node load characterization result is generated by weighted summarization of processing load level, interface occupancy level, communication bandwidth occupancy level, and task queuing level. The weighted summarization means that the load weight value corresponding to the processing load level, the interface weight value corresponding to the interface occupancy level, the communication weight value corresponding to the communication bandwidth occupancy level, and the queuing weight value corresponding to the task queuing level are weighted and synthesized. The load weight value, interface weight value, communication weight value, and queuing weight value are preset weight parameters. Based on the characteristics of device operation, environmental status and communication status, the control requirements of embedded control nodes are identified to obtain control requirement representation results. The control requirement representation results are generated by combining the degree of control real-time requirement, control accuracy requirement, collaborative adjustment requirement and resource call requirement. The node local state representation results, node load representation results, and control requirement representation results are fused together to generate node state representation results, which are then written into the corresponding node state cache in the order of the time window. The fusion process includes uniform dimension mapping, priority sorting, and weighted association of the node local state representation results, node load representation results, and control requirement representation results according to preset weight coefficients.

[0023] In this embodiment, the local control module includes: Each embedded control node reads the node state representation result corresponding to the current control cycle from the corresponding node state buffer, and performs association mapping on the node state representation result according to the control object identifier to generate local control input data for the corresponding control object; Based on local control input data, control state variables, load constraint variables, and control demand variables are extracted, and a local control strategy is calculated to generate a local control parameter set. The control state variables characterize the degree of operational state deviation of the current controlled object, the load constraint variables characterize the degree of resource occupancy constraint of the current embedded control node, and the control demand variables characterize the degree of real-time adjustment demand of the current controlled object. The calculation of the local control strategy includes calculating the operational state deviation value based on the control state variables and generating a state adjustment variable; determining the resource occupancy limit value based on the load constraint variables and generating a load limiting variable; determining the control response level based on the control demand variables and generating a demand response variable; weighting and fusing the state adjustment variable, load limiting variable, and demand response variable according to preset weighting coefficients to obtain the control adjustment variable; and performing parameter mapping processing on the control adjustment variable based on the controlled object identifier to generate the adjustment parameter, limiting parameter, and response parameter in the local control parameter set. The control command parameter values ​​are determined based on the local control parameter set and the control object identifier. The control command parameter values ​​are then adapted to the execution interface type corresponding to the control object to generate a local control command set. The generation of the local control command set includes parsing the adjustment parameters, limit parameters, and response parameters in the local control parameter set based on the control object identifier to determine the control command parameter values ​​for the corresponding control object. The control command parameter values ​​are then mapped to the execution interface type corresponding to the control object to generate a command category identifier and command parameter values. The generated control commands are sorted according to a preset control priority rule to determine the command output order. Finally, the control object identifier, command category identifier, command parameter values, and command output order are combined and encapsulated to generate the local control command set. The local control instruction set is written into the corresponding execution control interface unit, and the local control instruction set is output to the corresponding execution mechanism according to the instruction output order to drive the corresponding execution mechanism to perform control actions; The system collects interface feedback information returned by the execution control interface unit and action feedback information returned by the actuator, and performs consistency verification on the output state of the local control command set based on the interface feedback information and action feedback information to obtain the local control response result. The consistency verification includes extracting command execution state information based on the interface feedback information and extracting execution result state information based on the action feedback information, matching the command execution state information with the execution result state information, determining whether the command category identifier in the local control command set is consistent with the action category returned by the actuator, determining whether the command parameter value and the actual execution parameter value in the action feedback information meet the preset deviation range, and performing sequential consistency comparison of the execution state of each command according to the command output order, and generating the local control response result based on the matching result. Based on the local control response results, a closed-loop correction is performed on the local control parameter set to generate a corrected local control parameter set. This corrected local control parameter set is then written back to the corresponding node state buffer to form real-time closed-loop control at the edge side of the corresponding controlled object. The closed-loop correction includes extracting the instruction execution deviation value and execution delay value from the local control response results, comparing the instruction execution deviation value with the adjustment parameters in the local control parameter set to generate a parameter correction amount, comparing the execution delay value with the response parameter to generate a response correction amount, and simultaneously adjusting the limiting parameters based on the resource occupancy change information in the local control response results to generate a limiting correction amount. The parameter correction amount, response correction amount, and limiting correction amount are weighted and synthesized according to preset weight coefficients to obtain the corrected control amount. Finally, the corrected control amount is mapped and updated according to the controlled object identifier to update the adjustment parameters, limiting parameters, and response parameters in the local control parameter set, generating the corrected local control parameter set.

[0024] In this embodiment, the collaborative allocation module includes: Each embedded control node sends its node status representation results and network topology status information to neighboring embedded control nodes through the node communication unit. It receives, correlates, and organizes the node status representation results and network topology status information sent by neighboring embedded control nodes to generate candidate collaborative evaluation data. Based on candidate collaborative evaluation data, the Contract Network protocol is used to evaluate the collaborative capabilities of adjacent embedded control nodes and generate a set of candidate collaborative nodes. The generation of the candidate collaborative node set includes: each embedded control node, acting as a task issuing node, extracts the node state characterization results and network topology state information of adjacent embedded control nodes from the candidate collaborative evaluation data, and generates collaborative task announcement information based on the extracted node state characterization results and network topology state information, and sends the task announcement message to adjacent embedded control nodes through the node communication unit; each adjacent embedded control node, acting as a task response node, determines the node's real-time control capability based on the node's local state characterization results in the candidate collaborative evaluation data, determines the node's resource carrying capacity based on the node's load characterization results, and determines the node's control requirements characterization results. The task response capability is evaluated, and collaborative bidding response information is generated and returned to the task issuing node. The task issuing node summarizes and evaluates the received collaborative bidding response information according to the bidding selection mechanism of the contract network protocol. Based on the communication reachability represented by the network topology status information in the candidate collaborative evaluation data, as well as the real-time control capability, resource carrying capacity, and task response capability of the node corresponding to the collaborative bidding response information, the collaborative evaluation value of each adjacent embedded control node is calculated. The nodes are sorted and filtered according to the size of the collaborative evaluation value. The adjacent embedded control nodes that meet the preset collaborative threshold conditions are determined as candidate collaborative nodes, and a set of candidate collaborative nodes is generated by combining them according to the node identifier. Read the set of control tasks to be assigned, and parse the task attributes of each control task according to the control object identifier, control task type, control cycle requirements, resource call requirements and communication connection constraints to generate a set of task matching attributes; Task attribute parsing includes: reading the control object identifier and control task type corresponding to each control task to be assigned from the set of control tasks to be assigned; parsing the control interface type and actuator identifier of the corresponding control object based on the control object identifier to generate control object attribute items; calculating the task execution cycle value according to the control cycle requirements and generating task cycle attribute items; extracting the processing resource requirements, interface call requirements, and communication bandwidth requirements according to the resource call requirements and generating resource requirement attribute items; parsing the communication reachability relationship, link delay constraints, and bandwidth constraints between the embedded control node where the corresponding control object is located and the candidate collaborative nodes according to the communication connection constraints and generating communication constraint attribute items; and associating and combining the control object attribute items, task cycle attribute items, resource requirement attribute items, and communication constraint attribute items according to the task identifier to generate a task matching attribute set. Based on the candidate collaborative node set and the task matching attribute set, a strict two-sided matching method is used to match the candidate collaborative node set with the set of control tasks to be assigned, and a task assignment result set is generated. The generation of the task allocation result set includes: extracting the control object identifier, control cycle requirement, resource call requirement, and communication connection constraint corresponding to each control task to be allocated based on the task matching attribute set; extracting the node local state representation result, node load representation result, control requirement representation result, and network topology state information corresponding to each candidate collaborative node based on the candidate collaborative node set; calculating the fit between each control task to be allocated and each candidate collaborative node, wherein the fit is calculated by combining the matching degree between the control cycle requirement and the node task response capability, the matching degree between the resource call requirement and the node resource carrying capacity, and the matching degree between the communication connection constraint and the node communication reachability to generate a node task matching evaluation value; and generating the control object identifier, control cycle requirement, resource call requirement, and communication connection constraint according to the node task matching evaluation value. The system generates a task preference sequence for the control task and a node preference sequence for the candidate collaborative nodes, and performs iterative matching processing according to a strict bilateral matching rule. Each control task to be assigned sends a matching request to the corresponding candidate collaborative node in sequence according to its task preference sequence. Each candidate collaborative node prioritizes the received matching requests according to its node preference sequence, retains the highest priority matching request, and rejects lower priority matching requests. The rejected control task continues to send a matching request to the next candidate collaborative node according to its task preference sequence. When the iterative matching continues until no new matching requests are available, the stable matching result between the candidate collaborative node and the control task to be assigned is determined based on the final retained node task matching relationship. The result is then associated and encapsulated according to the task identifier and node identifier to generate a task assignment result set. The control tasks are dynamically adjusted based on the task allocation result set. The dynamically adjusted control task allocation status is written into the task collaboration record, and the task allocation result set is sent to the corresponding candidate collaboration nodes so that each embedded control node can perform collaborative control task adjustment according to the task allocation result set. The dynamic adjustment includes dividing the control tasks of this node into locally retained control tasks, collaborative migration control tasks, and collaborative receiving control tasks. The output control permissions, parameter call permissions, and status synchronization permissions of the collaborative migration control tasks are switched and configured according to the task allocation result set. At the same time, a task receiving identifier, a control object identifier, and a collaborative execution identifier are established for the collaborative receiving control tasks.

[0025] In this embodiment, the policy update module includes: The standardized input dataset, node state representation results, and task allocation result set are correlated and organized according to the control object identifier, task identifier, and node identifier to obtain the policy update input data. Based on the policy update input data, the changes in running state, task coordination, and resource scheduling are extracted and policy adaptability analysis is performed to obtain state adjustment, coordination compensation, and resource constraint correction quantities. The running state change quantity characterizes the degree of state offset between the standardized input dataset and the node state characterization results. The task coordination change quantity characterizes the degree of task migration and coordination execution corresponding to the task allocation result set. The resource scheduling change quantity characterizes the degree of resource occupancy change of each embedded control node under the current task allocation relationship. The state adjustment quantity is determined by combining the control offset value and state fluctuation value corresponding to the running state change quantity. The coordination compensation quantity is determined by combining the task migration impact value and coordination execution impact value corresponding to the task coordination change quantity. The resource constraint correction quantity is determined by combining the processing resource correction value, interface resource correction value, and communication resource correction value corresponding to the resource scheduling change quantity. Adaptive control strategy generation processing is performed based on state adjustment quantity, cooperative compensation quantity and resource constraint correction quantity to obtain an updated control parameter set. The adaptive control strategy generation processing includes weighting and fusing the state adjustment quantity, cooperative compensation quantity and resource constraint correction quantity according to preset weight coefficients to obtain the strategy correction quantity, and mapping the strategy correction quantity according to the control object identifier to determine the adjustment parameter, limit parameter and response parameter in the updated control parameter set. Based on the updated control parameter set and the task allocation result set, the local control policy is reconstructed to obtain the updated control policy set. The policy reconstruction includes determining the parameter calling relationship, permission constraint relationship and state synchronization relationship of the corresponding control objects according to the local retention control task, collaborative migration control task and collaborative reception control task in the task allocation result set, and writing the adjustment parameter, limit parameter and response parameter in the updated control parameter set into the corresponding control policy item to obtain the policy parameter item, policy permission item and policy execution item in the updated control policy set. A consistency check is performed on the updated control parameter set and the updated control strategy set. The updated control parameter set and the updated control strategy set that pass the consistency check are loaded into the control strategy calculation unit so that the control strategy calculation unit can adjust the control strategy in real time according to the updated control parameter set and the updated control strategy set. The consistency check includes checking the parameter correspondence between the adjustment parameters, limit parameters and response parameters in the updated control parameter set and the strategy parameter items in the updated control strategy set; checking the permission correspondence between the task allocation relationship in the task allocation result set and the strategy permission items in the updated control strategy set; and checking the resource adaptation between the resource allocation relationship corresponding to the task allocation result set and the strategy execution items in the updated control strategy set.

[0026] In this embodiment, the effect evaluation module includes: The control response data returned by the execution control interface unit is collected and associated with the control object identifier, node identifier and control cycle identifier to obtain response verification input data. The control response data includes execution feedback data, equipment operation result data and node collaborative response data. Based on the response verification input data, extract the instruction execution status quantity, device operation result status quantity and node collaborative response status quantity and perform consistency verification to obtain the consistency verification result. The consistency verification includes performing execution correspondence verification between the instruction execution status quantity and the device operation result status quantity, performing collaborative correspondence verification between the device operation result status quantity and the node collaborative response status quantity, and performing status correspondence verification between the node collaborative response status quantity and the task execution relationship corresponding to the task allocation result set. The control effect is evaluated based on the consistency verification results and response verification input data to obtain the control effect evaluation results. The control effect evaluation results are determined by a combination of execution deviation value, state stability value and collaborative deviation value. The execution deviation value is determined based on the deviation relationship between the execution feedback data and the equipment operation result data. The state stability value is determined based on the fluctuation relationship of the equipment operation result data under continuous control cycle. The collaborative deviation value is determined based on the deviation relationship between the node collaborative response data and the collaborative execution relationship corresponding to the task allocation result set. Control result records are generated based on consistency verification results and control effect evaluation results. Then, based on the control result records, collaborative response status identifiers corresponding to node collaborative response status quantities and collaborative deviation identifiers corresponding to collaborative deviation values ​​are extracted to generate collaborative status records. The control result records include control object identifiers, node identifiers, control cycle identifiers, consistency verification results, control effect evaluation results, and execution status identifiers. The collaborative status records include node identifiers, task identifiers, collaborative response status identifiers, collaborative deviation identifiers, and collaborative time identifiers.

[0027] Example 1: To verify the feasibility of this invention in practice, it was applied to a multi-device embedded control scenario in an industrial automated production workshop. Multiple embedded control nodes are deployed in this workshop to control the operating status of different production equipment, including conveyor devices, mechanical actuators, and environmental monitoring devices. These embedded control nodes are interconnected via an industrial communication network and continuously collect equipment operating data, environmental sensing data, and node communication status data. In conventional control methods, each node mainly relies on local control logic or is uniformly scheduled by a central control system. When the operating status of the equipment changes or control tasks need to be adjusted among multiple nodes, the lack of an effective coordination mechanism between nodes easily leads to uneven distribution of control tasks, large fluctuations in node load, and delayed control response, thereby affecting the stability of the overall control system and the efficiency of collaborative control.

[0028] In this scenario, device operation data, environmental perception data, and node communication status data are first collected by each embedded control node. These data are then preprocessed according to a unified time base to obtain standardized input datasets for each embedded control node. Subsequently, an edge collaborative control topology is constructed based on the hardware resource information, network connectivity, and control task type of each embedded control node. Edge computing processing units, control strategy calculation units, execution control interface units, and node communication units are deployed within each node, enabling each node to perform data analysis and control strategy calculation locally. Each node performs real-time feature extraction and status analysis on the standardized input dataset, obtaining node status representation results and writing them to the node status cache. While executing local control strategies, nodes exchange node status representation results and network topology status information through node communication units, evaluate the collaborative capabilities of adjacent nodes, and publish collaborative task announcements and receive collaborative bidding response information according to the contract network protocol. Through comprehensive analysis of node real-time control capabilities, node resource carrying capacity, and node task response capabilities, candidate collaborative nodes are screened. Strict bilateral matching processing is then performed based on the control object identifier, control cycle requirements, resource call requirements, and communication connection constraints of the set of control tasks to be assigned, thereby forming a stable task allocation relationship. Each embedded control node dynamically adjusts the control tasks according to the task allocation relationship. At the same time, it performs consistency verification and control effect evaluation on the control response data returned by the execution control interface unit, performs closed-loop correction on the local control parameter set and writes it back to the node state buffer to maintain continuous and stable edge-side closed-loop control.

[0029] In a specific production workshop environment, this invention was deployed in a multi-node embedded control system for continuous operation verification. During operation, each embedded control node continuously performed node status analysis, collaborative capability assessment, and task matching processing across multiple production periods. Real-time information sharing and collaborative task allocation between nodes were achieved through an edge collaborative control topology. Under conditions of changing equipment operating status and fluctuating task load, the system could dynamically adjust control tasks through a contract network protocol and a strict bilateral matching mechanism. Control parameters were continuously corrected through consistency verification of control response data and evaluation of control effects, maintaining a stable collaborative relationship between embedded control nodes. In actual operation, the control strategies of each node could be adjusted in real time according to changes in node status and task allocation relationships. The collaborative execution process between nodes remained stable, and the control task allocation relationship was always in a coordinated state. This verified the feasibility of this invention in a multi-node embedded control environment and its effective role in improving the system's collaborative control capabilities and operational stability.

[0030] Table 1. Performance Comparison of the Invention and Traditional Embedded Architecture Control Methods

[0031] As can be clearly seen from Table 1, the method of the present invention is superior to the traditional method in many indicators.

[0032] In terms of control response time, the traditional method has a response time of 168ms, while the method of this invention has a response time of 142ms, a reduction of 26ms. This improvement stems from the deployment of an edge computing processing unit within the embedded control node, enabling the device's operating data and environmental perception data to be directly analyzed and calculated for control strategies at the node side. This reduces the process of data transmission to the central control system and back, thereby lowering the overall control latency.

[0033] Regarding task allocation computation time, the traditional method takes 96ms, while the method of this invention takes 74ms, a reduction of 22ms. This improvement stems from the fact that this invention uses a contract network protocol to screen candidate collaborative nodes during the collaborative allocation process, and combines it with a strict bilateral matching method for task allocation. This enables a stable matching relationship to be quickly formed between the control tasks to be allocated and the capabilities of the nodes, avoiding the time consumption caused by multiple iterative calculations in traditional centralized scheduling.

[0034] Regarding node resource utilization, the traditional method achieves 71.3%, while the method of this invention improves it to 79.8%. This improvement stems from the fact that this invention reflects the node load status and control demand status in real time through node status characterization results, enabling tasks to dynamically migrate and reallocate among multiple embedded control nodes. This reduces situations where some node resources are idle while other nodes are overloaded, thereby improving overall resource utilization efficiency.

[0035] Regarding the success rate of collaborative tasks, the traditional method achieves 92.1%, while the method of this invention reaches 96.4%. The improved success rate of collaborative tasks mainly stems from the fact that this invention comprehensively evaluates the real-time control capability, resource carrying capacity, and task response capability of nodes, and then forms a stable task matching relationship through strict bilateral matching, resulting in a higher degree of matching between task execution nodes and task requirements. Therefore, the collaborative task execution process is more stable and reliable.

[0036] Regarding the control deviation rate, the traditional method is 5.8%, while the method of this invention reduces it to 4.3%. This improvement stems from the introduction of a consistency verification and control effect evaluation mechanism for control response data during the control execution phase. By continuously comparing the execution feedback data with the equipment operation result data and performing closed-loop correction on the local control parameter set, the control parameters can be dynamically adjusted according to the execution results, thereby reducing control deviation.

[0037] Regarding network communication load, the traditional method achieves 13.7 Mbps, while the method of this invention achieves 11.5 Mbps. The reason for the reduced communication load is that this invention, through edge collaborative control topology, enables most control calculations to be completed on the node side, exchanging only node state representation results and necessary topology state information, thus reducing the transmission of a large amount of raw device data in the network and thereby reducing communication pressure.

[0038] Regarding the stability of the control cycle, the traditional method achieves 90.6%, while the method of this invention improves it to 94.2%. This improvement in stability stems from the fact that this invention continuously records the node state representation results through a node state buffer and combines this with an adaptive control strategy update mechanism. This allows the control strategy to be adjusted in real time according to changes in node operating states and task coordination relationships, thereby enabling the system to maintain a more stable operating state throughout the continuous control cycle.

[0039] The above are merely preferred embodiments of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

Claims

1. An embedded architecture control system based on edge computing, characterized in that, include: The data processing module is used to acquire control data from multiple embedded control nodes and preprocess it according to a unified time base to generate a standardized input dataset for each embedded control node. The topology construction module is used to build an edge collaborative control topology based on the basic information of each embedded control node, generate network topology status information, deploy edge computing processing units, control strategy calculation units, execution control interface units and node communication units in each embedded control node, and establish state sharing channels and task collaboration channels between nodes. The state analysis module is used to perform real-time feature extraction and state analysis on the standardized input dataset, generate node state representation results and write them to the corresponding node state cache. The local control module is used to calculate the local control strategy based on the node state representation results in the node state buffer, generate a set of local control instructions, and output them to the corresponding actuators through the execution control interface unit to complete real-time closed-loop control on the edge side. The collaborative allocation module is used to generate candidate collaborative evaluation data by exchanging node status representation results and network topology status information through node communication units, filter candidate collaborative nodes based on the candidate collaborative evaluation data using the contract network protocol, and determine the task allocation result set through a strict bilateral matching method. The strategy update module is used to generate adaptive control strategies based on the standardized input dataset, node state representation results, and task allocation result set, to obtain the updated control parameter set and the updated control strategy set, and load them into the control strategy calculation unit. The effect evaluation module is used to perform consistency verification and control effect evaluation on the control response data returned by the execution control interface unit, and generate control result records and collaborative status records.

2. The embedded architecture control system based on edge computing according to claim 1, characterized in that, The control data acquired by the data processing module includes equipment operation data, environmental perception data, and node communication status data. The preprocessing includes timestamp alignment, outlier removal, missing value completion, unit normalization, and data segmentation.

3. The embedded architecture control system based on edge computing according to claim 1, characterized in that, The topology construction module includes: Collect basic information about each embedded control node, including hardware resource information, control object identifier, network connection relationship and control task type; Based on the hardware resource information of each embedded control node, the node resource evaluation result of each embedded control node is calculated. Based on the control object identifier and control task type of each embedded control node, calculate the task adaptation evaluation result for each embedded control node. Based on network connectivity, the communication link status between any two embedded control nodes is analyzed, and the link collaboration evaluation results between nodes are calculated. Based on the node resource evaluation results, task adaptation evaluation results, and link collaboration evaluation results, the node connection priority between each embedded control node is determined, and a node adjacency table, a link status association table, and a task collaboration association table are established according to the node connection priority. Topology connection edges are generated for node pairs that meet the preset collaboration establishment conditions. The edge collaborative control topology is constructed based on the node adjacency table, link state association table and task collaboration association table. The node connection results between each embedded control node are determined, and the corresponding link state information and task collaboration state information are marked on each topology connection edge to obtain the network topology state information. Based on the node connection results and task collaboration association table of the edge collaborative control topology, an edge computing processing unit, a control strategy calculation unit, an execution control interface unit, and a node communication unit are deployed in each embedded control node, and a state sharing channel and a task collaboration channel between nodes are established based on the topology connection edges.

4. The embedded architecture control system based on edge computing according to claim 1, characterized in that, The status analysis module includes: Each embedded control node calls the edge computing processing unit to read the standardized input dataset, and performs synchronous segmentation according to a unified time window to generate a set of data segments corresponding to each time window; Extract device operation characteristics, environmental status characteristics, and communication status characteristics from data segments for each time window; Based on the device operation characteristics and environmental state characteristics, the current operating state of the embedded control node is comprehensively determined to obtain the local state characterization result of the node. Based on the device's operating characteristics and communication status characteristics, the resource occupancy of the embedded control node is evaluated to obtain the node load characterization results. Based on the characteristics of device operation, environmental status, and communication status, the control requirements of embedded control nodes are identified to obtain control requirement characterization results. The node local state representation results, node load representation results, and control requirement representation results are fused together to generate node state representation results, which are then written into the corresponding node state cache in the order of the time window.

5. The embedded architecture control system based on edge computing according to claim 1, characterized in that, The local control module includes: Each embedded control node reads the node state representation result corresponding to the current control cycle from the corresponding node state buffer, and performs association mapping on the node state representation result according to the control object identifier to generate local control input data for the corresponding control object; Based on local control input data, control state variables, load constraint variables and control demand variables are extracted and local control strategies are calculated to generate a set of local control parameters. The control command parameter values ​​are determined based on the local control parameter set and the control object identifier, and the control command parameter values ​​are adapted according to the execution interface type corresponding to the control object to generate a local control command set. The local control instruction set is written into the corresponding execution control interface unit, and the local control instruction set is output to the corresponding execution mechanism according to the instruction output order to drive the corresponding execution mechanism to perform control actions; The interface feedback information returned by the execution control interface unit and the action feedback information returned by the actuator are collected, and the consistency of the output state of the local control command set is verified based on the interface feedback information and action feedback information to obtain the local control response result. Based on the local control response results, the local control parameter set is corrected in a closed loop to generate a corrected local control parameter set. The corrected local control parameter set is then written back to the corresponding node state buffer to form real-time closed-loop control on the edge side of the corresponding controlled object.

6. The embedded architecture control system based on edge computing according to claim 1, characterized in that, The collaborative allocation module includes: Each embedded control node sends its node status representation results and network topology status information to neighboring embedded control nodes through the node communication unit. It receives, correlates, and organizes the node status representation results and network topology status information sent by neighboring embedded control nodes to generate candidate collaborative evaluation data. Based on candidate collaborative evaluation data, the Contract Network protocol is used to evaluate the collaborative capabilities of adjacent embedded control nodes and generate a set of candidate collaborative nodes. Read the set of control tasks to be assigned, and parse the task attributes of each control task according to the control object identifier, control task type, control cycle requirements, resource call requirements and communication connection constraints to generate a set of task matching attributes; Based on the candidate collaborative node set and the task matching attribute set, a strict two-sided matching method is used to match the candidate collaborative node set with the set of control tasks to be assigned, and a task assignment result set is generated. The control tasks are dynamically adjusted based on the task allocation result set. The dynamically adjusted control task allocation status is written into the task collaboration record, and the task allocation result set is sent to the corresponding candidate collaboration node.

7. The embedded architecture control system based on edge computing according to claim 1, characterized in that, The policy update module includes: The standardized input dataset, node state representation results, and task allocation result set are correlated and organized according to the control object identifier, task identifier, and node identifier to obtain the policy update input data. Based on the policy update input data, the changes in running status, task coordination, and resource scheduling are extracted and policy adaptability analysis is performed to obtain the state adjustment quantity, coordination compensation quantity, and resource constraint correction quantity. Based on the state adjustment quantity, the cooperative compensation quantity, and the resource constraint correction quantity, an adaptive control strategy generation process is performed to obtain an updated control parameter set. Based on the updated control parameter set and the task allocation result set, the local control policy is reconstructed to obtain the updated control policy set. A consistency check is performed on the updated control parameter set and the updated control strategy set. The updated control parameter set and the updated control strategy set that pass the consistency check are then loaded into the control strategy calculation unit.

8. The embedded architecture control system based on edge computing according to claim 1, characterized in that, The effect evaluation module includes: The control response data returned by the execution control interface unit is collected and associated with the control object identifier, node identifier and control cycle identifier to obtain the response verification input data. Based on the response verification input data, extract the instruction execution status quantity, the device operation result status quantity, and the node collaborative response status quantity, and perform consistency verification to obtain the consistency verification result; The control effect is evaluated based on the consistency verification results and the response verification input data to obtain the control effect evaluation results. Based on the consistency verification results and control effect evaluation results, control result records are generated. Based on the control result records, the collaborative response state identifiers corresponding to the node collaborative response state quantities and the collaborative deviation identifiers corresponding to the collaborative deviation values ​​are extracted to generate collaborative state records.

Images