Real-time decision-making method and device based on multi-level edge nodes, equipment and medium

Abstract

The invention relates to the field of Internet of Things, and particularly discloses a real-time decision-making method and device based on a multi-level edge node and a medium, the method is applied to a multi-level edge node architecture, and the method comprises the following steps: converting sensor data by using multi-protocol conversion middleware to obtain standardized data; aggregating the standardized data, and synchronizing the task to be decided; segmenting the task to be decided into a plurality of sub-tasks, and issuing the plurality of sub-tasks; and receiving a plurality of calculation results corresponding to the plurality of sub-tasks, and generating a real-time decision scheme based on the plurality of calculation results. By setting the protocol conversion middleware on the edge node layer, the data processing amount of the edge gateway can be reduced, so that the data transmission speed is reduced, and meanwhile, by setting the dynamic federated learning framework and the multi-target collaborative optimization algorithm, model updating and task distribution can be realized in time, so that the calculation accuracy is improved, and the calculation efficiency is improved. And the task processing speed.

Description

Technical Field

[0001] The present application relates to the field of Internet of Things, and specifically to a real-time decision-making method, device, and medium based on multi-level edge nodes. Background Art

[0002] In existing technologies, industrial Internet of Things platforms can collect equipment data from industrial equipment and achieve real-time control of industrial equipment based on the equipment data. However, existing solutions rely too much on cloud processing and have slow model updates, resulting in an excessively long average response time to equipment failures and unable to meet the real-time control requirements of key equipment (such as injection molding machines and CNC machine tools).

[0003] Therefore, there is an urgent need for a method that can improve decision response speed and reduce control delay. Summary of the Invention

[0004] In order to solve the above problems, the present application proposes a real-time decision-making method, device and medium based on multi-level edge nodes, wherein the method is applied to a multi-level edge node architecture, and the multi-level edge node architecture includes: an edge server layer, an edge gateway layer and an edge node layer, the edge node layer is connected to the industrial equipment and is provided with a multi-protocol conversion middleware, and the multi-protocol conversion middleware is used to be compatible with preset types of industrial protocols; the edge gateway layer is provided with a dynamic federated learning framework; the edge server layer stores a multi-objective collaborative optimization algorithm; the real-time decision-making method based on multi-level edge nodes includes: receiving sensor data from industrial equipment and using the multi-protocol conversion middleware to convert the sensor data to obtain standardized data; aggregating the standardized data and synchronizing the tasks to be decided; dividing the tasks to be decided into multiple subtasks through the dynamic federated learning framework and the multi-objective collaborative optimization algorithm, and issuing the multiple subtasks; receiving multiple calculation results corresponding to the multiple subtasks, and generating a real-time decision plan based on the multiple calculation results.

[0005] In one example, aggregating the standardized data and synchronizing the tasks to be decided specifically include: aggregating the standardized data through an edge gateway and eliminating redundant data to obtain uploaded data; determining that the uploaded data meets the task decision conditions, and sending the uploaded data to the edge server.

[0006] In one example, the task to be decided is divided into multiple subtasks through a dynamic federated learning framework and a multi-objective collaborative optimization algorithm, and the multiple subtasks are issued, specifically including: updating the computing model stored in each edge node through the dynamic federated learning framework; receiving the node resource status of each edge node, and issuing the multiple subtasks based on the multi-objective collaborative optimization algorithm and the node resource status.

[0007] In one example, the updating of the computing model stored in each edge node through the dynamic federated learning framework specifically includes: training the computing model through local data stored in the edge node to update the computing model parameters and corresponding computing model version of each edge node; when the computing model version changes, performing a full update through the dynamic federated learning framework to replace the global model; when the model version has not changed, incrementally updating the computing model parameters of each edge node.

[0008] In one example, the multiple subtasks are issued based on the multi-objective collaborative optimization algorithm and the node resource status, specifically including: obtaining the computing model version of each edge node and the network load corresponding to each edge node; determining the decision weight corresponding to the task to be decided, the decision weight including the time weight and the accuracy weight; based on the computing model version, the network load and the node resource status, the multiple subtasks are issued.

[0009] In one example, the method further includes: determining the total amount of computing resources of the edge node and the type of industrial equipment connected to the edge node; and determining the type of computing model corresponding to the edge node based on the total amount of computing resources and the type of industrial equipment.

[0010] In one example, before receiving sensor data from industrial equipment and converting the sensor data using a multi-protocol conversion middleware to obtain standardized data, the method also includes: determining the type of industrial equipment connected to the edge node; and determining the multi-protocol conversion middleware corresponding to the edge node based on the industrial equipment type.

[0011] The present application also provides a real-time decision-making device based on multi-level edge nodes, including: a conversion module, which receives sensor data from industrial equipment and uses a multi-protocol conversion middleware to convert the sensor data to obtain standardized data; an upload module, which aggregates the standardized data and synchronizes the tasks to be decided; a segmentation module, which divides the tasks to be decided into multiple subtasks through a dynamic federated learning framework and a multi-objective collaborative optimization algorithm, and distributes the multiple subtasks; a decision module, which receives multiple calculation results corresponding to the multiple subtasks, and generates a real-time decision plan based on the multiple calculation results.

[0012] The present application also provides a real-time decision-making device based on multi-level edge nodes, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the steps of the method described in any one of the above examples.

[0013] The present application also provides a non-volatile computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to execute the steps of the method described in any one of the above examples.

[0014] The method proposed in this application can bring the following beneficial effects: by setting up a protocol conversion middleware at the edge node layer, the data processing volume of the edge gateway can be reduced, thereby reducing the data transmission speed. At the same time, by setting up a dynamic federated learning framework and a multi-objective collaborative optimization algorithm, the model can be updated and the task can be distributed in a timely manner, thereby improving the calculation accuracy and task processing speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The drawings described herein are used to provide a further understanding of the present application and constitute a part of the present application. The illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute an improper limitation on the present application. In the drawings: Figure 1 Schematic diagram of a real-time decision-making method based on multi-level edge nodes in an embodiment of the present application; Figure 2 This is a structural diagram of a real-time decision-making device based on multi-level edge nodes in an embodiment of the present application; Figure 3 This is a structural diagram of a real-time decision-making device based on multi-level edge nodes in an embodiment of the present application. DETAILED DESCRIPTION

[0016] To make the purpose, technical solutions, and advantages of this application more clear, the technical solutions of this application will be clearly and completely described below in conjunction with the specific embodiments of this application and the corresponding drawings. Obviously, the embodiments described are only part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0017] The following describes in detail the technical solutions provided by various embodiments of the present application in conjunction with the accompanying drawings.

[0018] Figure 1This is a flowchart of a real-time decision-making method based on multi-level edge nodes, provided for one or more embodiments of this specification. This method can be applied to the real-time control of various types of industrial equipment (such as injection molding machines and CNC machine tools). The process can be executed by computing devices in the corresponding field (e.g., servers located in the cloud). Certain input parameters or intermediate results in the process can be manually adjusted to help improve accuracy.

[0019] The implementation of the analysis method involved in the embodiments of the present application can be a terminal device or a server, and this application does not impose any special restrictions on this. For ease of understanding and description, the following embodiments are described in detail using a server as an example. It should be noted that the server can be a single device or a system composed of multiple devices, that is, a distributed server, and this application does not impose any specific restrictions on this.

[0020] like Figure 1 As shown, an embodiment of the present application provides a method, which is applied to a multi-level edge node architecture, where the multi-level edge node architecture includes: an edge server layer, an edge gateway layer, and an edge node layer. Among them, the edge node layer is connected to the industrial equipment and is used to obtain the device data of the industrial equipment. The edge node layer is provided with a multi-protocol conversion middleware, and the multi-protocol conversion middleware here is used to be compatible with preset types of industrial protocols (such as compatible with 12 types of industrial protocols such as Modbus / OPC UA / Profinet). The edge gateway layer is provided with a dynamic federated learning framework; the edge server layer stores a multi-objective collaborative optimization algorithm; the real-time decision-making method based on multi-level edge nodes includes: S101: Receive sensor data from industrial equipment and convert the sensor data using multi-protocol conversion middleware to obtain standardized data.

[0021] First, the server acquires sensor data from sensors deployed near industrial equipment and sends it to the edge node. After receiving the sensor data, the edge node converts it into standardized data through the multi-protocol conversion middleware and sends the standardized data to the edge gateway layer.

[0022] S102: Aggregate the standardized data and synchronize the tasks to be decided.

[0023] After receiving the standardized data, the edge gateway layer aggregates it. Based on the standardized data, it determines whether there are any pending decisions. If so, it synchronizes these tasks to the edge server layer. If not, there is no need to synchronize the data to the edge server layer.

[0024] In one embodiment, when aggregating standardized data, the edge gateway can aggregate the standardized data and remove redundant data to obtain uploaded data. Redundant data here refers to the repeated data portions in the data from different sensors. This reduces the data uploaded from the edge gateway layer to the edge server layer, thereby reducing network load. When it is determined that the uploaded data meets the task decision conditions, the uploaded data is sent to the edge server.

[0025] Here we explain the pending decision tasks: pending decision tasks refer to when sensor data meets certain conditions and it is necessary to determine whether industrial equipment needs to be controlled. At this time, by executing the pending decision tasks to obtain the decision results, the industrial equipment can be controlled according to the decision results, such as adjusting the equipment parameters of the industrial equipment.

[0026] S103: Divide the task to be decided into multiple subtasks through a dynamic federated learning framework and a multi-objective collaborative optimization algorithm, and distribute the multiple subtasks.

[0027] When the edge gateway layer synchronizes the tasks to be decided to the edge server layer, the dynamic federated learning framework of the edge gateway layer and the multi-objective toe optimization algorithm of the edge server layer can be used to split the tasks to be decided, thereby obtaining multiple subtasks, and then issuing the multiple subtasks.

[0028] Specifically, before distributing subtasks, the computational model stored in each edge node needs to be updated through the dynamic federated learning framework. The framework then receives the node resource status of each edge node and distributes the subtasks based on the multi-objective collaborative optimization algorithm and the node resource status. This ensures load balancing across edge nodes and improves the overall computational speed of edge nodes.

[0029] Furthermore, when updating the computational models stored in each edge node through the dynamic federated learning framework, the computational models corresponding to each edge node must first be trained using the local data stored on the edge node to update the computational model parameters and corresponding computational model versions of each edge node. When the computational model version changes, a full update is performed through the dynamic federated learning framework to replace the global model; when the model version remains unchanged, the computational model parameters of each edge node are incrementally updated. During incremental updates, the edge gateway aggregates the parameters of all nodes (such as a weighted average) and submits the results to the edge server. During a full update, the edge server directly receives and broadcasts the new model. During global synchronization, the edge server distributes the updated global model, and the edge nodes download and replace the local model. By switching between full and incremental update modes, frequent communication overhead is avoided.

[0030] The implementation code is as follows: # Federated learning parameter synchronization pseudocode def dynamic_federated_learning(): local_model = train_on_edge_node() if model_version_change: sync_global_model(local_model) else: incremental_update(local_model) In one embodiment, when issuing subtasks, it is necessary to consider the computing model type and version of different edge nodes, as well as the network load corresponding to each edge node, so as to comprehensively consider the computing time and computing accuracy and select the appropriate edge node to issue the subtask. Specifically, the server obtains the computing model version of each edge node and the network load corresponding to each edge node. Then, the decision weight corresponding to the pre-set task to be decided is determined, where the decision weight includes time weight and accuracy weight. Finally, based on the computing model version, network load, and node resource status, multiple subtasks are issued.

[0031] For example, a reward function can be designed in advance, and the computing model version and computing model type corresponding to each edge node can be quantified in terms of the computing power for processing the current type of subtask. This can be taken into consideration along with the network load and node resource status as a factor in the reward function, thereby determining the adaptability between each edge node and the subtask in a quantitative manner.

[0032] S104: Receive multiple calculation results corresponding to the multiple subtasks, and generate a real-time decision solution based on the multiple calculation results.

[0033] After sending multiple subtasks to the edge node, the multiple subtasks can be calculated based on the computing power of the edge node to obtain the calculation results corresponding to the multiple subtasks. The calculation results of the multiple subtasks are then uploaded to the edge server through the edge gateway layer, so that the edge server can integrate the calculation results of the multiple subtasks to obtain a real-time decision-making solution.

[0034] In one embodiment, when building a multi-level edge node architecture, different types of computing models can be installed based on the type of edge node. For example, the type of computing model corresponding to an edge node can be determined based on the total amount of computing resources of the edge node and the type of industrial device connected to the edge node. In one embodiment, when building a multi-level edge node architecture, the multi-protocol conversion middleware installed on each edge node can be determined based on the type of industrial device connected to the edge node.

[0035] Taking the example of predictive maintenance for injection molding machines, vibration and temperature sensors can be deployed in advance to obtain vibration and temperature data from the injection molding machine. At this point, the multi-protocol conversion middleware installed on the edge node connected to the injection molding machine can be replaced with the corresponding protocol conversion middleware. Furthermore, by deploying a fault prediction model, such as an LSTM-based model, on this edge node or other nearby edge nodes, edge computing can be used to predict when the injection molding machine is likely to fail, allowing for proactive maintenance of the machine and reducing equipment downtime and maintenance costs.

[0036] like Figure 2 As shown, the embodiment of the present application also provides a real-time decision-making device based on multi-level edge nodes, including: The conversion module 201 receives sensor data from industrial equipment and converts the sensor data using a multi-protocol conversion middleware to obtain standardized data.

[0037] The upload module 202 aggregates the standardized data and synchronizes the tasks to be decided.

[0038] The segmentation module 203 divides the task to be decided into multiple subtasks through a dynamic federated learning framework and a multi-objective collaborative optimization algorithm, and distributes the multiple subtasks.

[0039] The decision module 204 receives a plurality of calculation results corresponding to the plurality of subtasks and generates a real-time decision solution based on the plurality of calculation results.

[0040] like Figure 3 As shown, an embodiment of the present application further provides a real-time decision-making device based on multi-level edge nodes, comprising: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to: Receive sensor data from industrial equipment and use multi-protocol conversion middleware to convert the sensor data to obtain standardized data; aggregate the standardized data and synchronize the tasks to be decided; divide the tasks to be decided into multiple subtasks through a dynamic federated learning framework and a multi-objective collaborative optimization algorithm, and issue the multiple subtasks; receive multiple calculation results corresponding to the multiple subtasks, and generate real-time decision plans based on the multiple calculation results.

[0041] The embodiment of the present application further provides a non-volatile computer storage medium storing computer-executable instructions, wherein the computer-executable instructions are configured to: Receive sensor data from industrial equipment and use multi-protocol conversion middleware to convert the sensor data to obtain standardized data; aggregate the standardized data and synchronize the tasks to be decided; divide the tasks to be decided into multiple subtasks through a dynamic federated learning framework and a multi-objective collaborative optimization algorithm, and issue the multiple subtasks; receive multiple calculation results corresponding to the multiple subtasks, and generate real-time decision plans based on the multiple calculation results.

[0042] The various embodiments in this application are described in a progressive manner. Similar portions between the various embodiments can be referred to in conjunction with each other. Each embodiment focuses on the differences between the other embodiments. In particular, the device and medium embodiments are generally similar to the method embodiments, so their descriptions are relatively simple. For relevant portions, refer to the descriptions of the method embodiments.

[0043] The devices and media provided in the embodiments of the present application correspond one-to-one to the methods. Therefore, the devices and media also have similar beneficial technical effects to their corresponding methods. Since the beneficial technical effects of the methods have been described in detail above, the beneficial technical effects of the devices and media will not be repeated here.

[0044] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware. Furthermore, the present application may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to magnetic disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0045] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0046] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing device to work in a specific manner, so that the instructions stored in the computer readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0047] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operational steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing the instructions executed on the computer or other programmable device for implementing the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0048] In a typical configuration, a computing device includes one or more processors (CPUs), input / output interfaces, network interfaces, and memory.

[0049] Memory may include non-permanent storage in a computer-readable medium, in the form of random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of a computer-readable medium.

[0050] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can be implemented using any method or technology for information storage. Information can be computer-readable instructions, data structures, program modules, or other data. Examples of computer storage media include, but are not limited to, phase-change RAM (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, compact disc read-only memory (CD-ROM), digital versatile disc (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices, or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer-readable media does not include transitory computer-readable media, such as modulated data signals and carrier waves.

[0051] It should also be noted that the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, commodity, or apparatus that includes a series of elements includes not only those elements but also other elements not explicitly listed, or includes elements inherent to such process, method, commodity, or apparatus. In the absence of further limitations, an element defined by the phrase "comprises a ..." does not exclude the presence of other identical elements in the process, method, commodity, or apparatus that includes the element.

[0052] The foregoing is merely an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various changes and variations. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application should all be included within the scope of the claims of the present application.

Claims

1. A real-time decision-making method based on multi-level edge nodes, characterized in that: Applied to a multi-level edge node architecture, the multi-level edge node architecture includes: an edge server layer, an edge gateway layer, and an edge node layer. The edge node layer is connected to industrial equipment and is provided with a multi-protocol conversion middleware, which is used to be compatible with preset types of industrial protocols; the edge gateway layer is provided with a dynamic federated learning framework; the edge server layer stores a multi-objective collaborative optimization algorithm; the real-time decision-making method based on multi-level edge nodes includes: Receiving sensor data from industrial equipment and converting the sensor data using multi-protocol conversion middleware to obtain standardized data; Aggregate the standardized data and synchronize the tasks to be decided; By using a dynamic federated learning framework and a multi-objective collaborative optimization algorithm, the task to be decided is divided into multiple subtasks, and the multiple subtasks are issued; Receive multiple calculation results corresponding to the multiple subtasks, and generate a real-time decision solution based on the multiple calculation results.

2. The method according to claim 1, characterized in that Aggregating the standardized data and synchronizing the tasks to be decided specifically include: Aggregating the standardized data through an edge gateway and removing redundant data to obtain uploaded data; Determine whether the uploaded data meets the task decision condition, and send the uploaded data to the edge server.

3. The method according to claim 1, characterized in that The dynamic federated learning framework and the multi-objective collaborative optimization algorithm are used to divide the task to be decided into multiple subtasks, and the multiple subtasks are issued, specifically including: Updating the computing model stored in each edge node through the dynamic federated learning framework; Receive the node resource status of each edge node, and issue the multiple subtasks based on the multi-objective collaborative optimization algorithm and the node resource status.

4. The method according to claim 3, characterized in that Updating the computing model stored in each edge node through the dynamic federated learning framework specifically includes: The computing model is trained using local data stored on the edge nodes to update the computing model parameters and corresponding computing model versions of each edge node. When the computing model version changes, a full update is performed through the dynamic federated learning framework to replace the global model; When the model version has not changed, the calculation model parameters of each edge node are incrementally updated.

5. The method according to claim 3, characterized in that The issuing of the plurality of subtasks based on the multi-objective collaborative optimization algorithm and the node resource status specifically includes: Obtain the computing model version of each edge node and the network load corresponding to each edge node; Determine a decision weight corresponding to the task to be decided, wherein the decision weight includes a time weight and an accuracy weight; The multiple subtasks are issued based on the computing model version, the network load, and the node resource status.

6. The method according to claim 1, characterized in that Before dividing the task to be decided into multiple subtasks by using the dynamic federated learning framework and the multi-objective collaborative optimization algorithm and issuing the multiple subtasks, the method further includes: Determining the total amount of computing resources of the edge node and the type of industrial equipment connected to the edge node; Based on the total amount of computing resources and the type of industrial equipment, a computing model type corresponding to the edge node is determined.

7. The method according to claim 1, characterized in that Before receiving sensor data from industrial equipment and converting the sensor data using multi-protocol conversion middleware to obtain standardized data, the method further includes: Determining the type of industrial equipment connected to the edge node; Based on the type of the industrial device, a multi-protocol conversion middleware corresponding to the edge node is determined.

8. A real-time decision-making device based on multi-level edge nodes, characterized in that: include: A conversion module receives sensor data from industrial equipment and converts the sensor data using a multi-protocol conversion middleware to obtain standardized data; The upload module aggregates the standardized data and synchronizes the tasks to be decided; A segmentation module, which divides the task to be decided into multiple subtasks through a dynamic federated learning framework and a multi-objective collaborative optimization algorithm, and distributes the multiple subtasks; The decision module receives a plurality of calculation results corresponding to the plurality of subtasks and generates a real-time decision solution based on the plurality of calculation results.

9. A real-time decision-making device based on multi-level edge nodes, characterized in that: include: at least one processor; and a memory communicatively connected to the at least one processor; wherein, The memory stores instructions that can be executed by the at least one processor, and the instructions are executed by the at least one processor to enable the at least one processor to perform the steps of the method according to any one of claims 1 to 7.

10. A non-volatile computer storage medium storing computer executable instructions, characterized in that: The computer executable instructions are configured to perform the steps of the method according to any one of claims 1 to 7.

Images