An edge computing method based on a lightweight artificial intelligence algorithm

Abstract

This invention relates to an edge computing method based on a lightweight artificial intelligence algorithm, comprising the following steps: S1, edge devices upload their real-time operating data to a fusion terminal, which parses the uploaded data to understand the operating status of the edge devices; S2, a distribution substation has a fusion terminal and various edge devices, and the fusion terminal, based on the current substation operating status and its own and the edge devices' computing resources, selects a resource-rich edge device to execute a lightweight artificial intelligence algorithm suitable for the substation's operating status; S3, the substation fusion terminal verifies the substation operating strategy obtained from the algorithm execution and then controls the edge devices. This method facilitates lightweight artificial intelligence algorithm service response and improves the operating efficiency of the distribution network.

Description

Technical Field

[0001] This invention relates to the field of smart distribution network technology, and specifically to an edge computing method based on a lightweight artificial intelligence algorithm. Background Technology

[0002] With the advancement of the practical application of integrated terminals, the ability to perceive and analyze loads can be significantly improved by combining these terminals with existing distribution area edge equipment, providing a possibility for building a new form of regional autonomy for distribution networks. For a long time, power grid construction has focused primarily on equipment management, lacking analysis and management of distribution network operation data. Furthermore, the complex data structure within the power grid, the massive volume of data, and the difficulty in data integration have contributed to the challenges of data analysis and collaborative management of low-voltage distribution areas. However, through the advancement of the practical application of integrated terminals, massive amounts of low-voltage distribution area data can be collected and aggregated into these terminals. Using these integrated terminals as edge computing platforms to analyze, identify, and manage the diverse loads of low-voltage distribution areas has become possible, laying the technical foundation for achieving regional autonomy in distribution networks.

[0003] The application of artificial intelligence algorithms, scenario analysis, management, and resource collaborative control, primarily based on integrated terminals in distribution substations, still faces numerous challenges. Firstly, the current integration of a massive number of power distribution devices, electrical quantity sensors, and status sensors into the Internet of Things (IoT) network presents a challenge. These devices are diverse, with varying communication protocols and configurations, resulting in a vast amount of heterogeneous, multi-source electricity consumption data. The acquisition, transmission, and computation of this data place enormous pressure on communication channels and the main station's storage and computing systems. A single integrated terminal struggles to simultaneously achieve real-time sensing, data processing, and execution of artificial intelligence algorithms for low-voltage distribution substation edge devices. Secondly, traditional artificial intelligence algorithms are often server-side oriented, relying on ample computing power and storage space to ensure accuracy and efficiency. This conflicts with the computing resources of edge devices in distribution substations. The core CPU of the State Grid converged terminal has a clock speed of no less than 700MHz, lower than the 2GHz commonly found in servers; its memory is no less than 512MB, and its FLASH is no less than 4GB, far inferior to cloud servers with tens of gigabytes of memory and thousands of gigabytes of FLASH. Due to the limited computing resources of the converged terminal, traditional artificial intelligence algorithm models are too large to deploy, requiring lightweight algorithms and a compatible edge-to-edge collaborative computing system. Finally, most current converged terminals in distribution substations primarily receive operational data from edge devices for monitoring and management, without controlling lower-level distributed power sources and charging piles. This hinders further refined management of distribution substations and results in limited improvement in operational efficiency. Summary of the Invention

[0004] The purpose of this invention is to provide an edge computing method based on a lightweight artificial intelligence algorithm, which is beneficial for realizing lightweight artificial intelligence algorithm service response and improving the operating efficiency of power distribution network.

[0005] To achieve the above objectives, the technical solution adopted by this invention is: an edge computing method based on a lightweight artificial intelligence algorithm, comprising the following steps:

[0006] S1. The edge devices upload their real-time operating data to the converged terminal. The converged terminal parses the real-time operating data uploaded by the edge devices to understand the operating status of the edge devices.

[0007] S2. The distribution area has a converged terminal and various edge devices. The converged terminal selects a resource-rich edge device to execute a lightweight artificial intelligence algorithm suitable for the operating status of the distribution area based on the current operating status of the distribution area and its own and the computing power resources of the edge devices.

[0008] S3. After verifying the operation strategy of the transformer area obtained by the algorithm, the transformer area fusion terminal implements the control function of the edge device.

[0009] Furthermore, in step S1, the edge device of the distribution radio area uploads real-time operating data to the converged terminal. The data is represented in the form of data values ​​or data blocks. When data blocks are used, the set protocol specifications are followed. The converged terminal parses the real-time operating data uploaded by the edge device to understand the operating status of the edge device.

[0010] Furthermore, in step S2, with the objectives of minimizing the load fluctuation of the distribution transformer and maximizing the absorption of distributed energy in the distribution area, a lightweight artificial intelligence algorithm library is established. Based on this library, a resource collaborative optimization control model for the distribution area is established. This model analyzes the operating status of each terminal device in the current low-voltage distribution area and selects and executes a lightweight artificial intelligence algorithm suitable for the current operating status of the distribution area from the library. The model solves for the operating instructions of controllable resources. The dual objective function of the resource collaborative optimization control model is:

[0011]

[0012]

[0013] In the formula, P t,i P is the load value of the distribution transformer in the transformer area at time i. av P is the average daily load value of the distribution transformer in the area. e,i This is the output of the distributed energy in the transformer substation at time i.

[0014] Further, in step S2, the ledgers or technical parameter files of the converged terminal and each edge device are obtained to acquire the hardware resource information of the converged terminal and each edge device, and a multi-terminal device computing power resource optimization allocation model is constructed, the objective function of which is:

[0015]

[0016] Where, r i represents the current resource usage percentage of the i-th terminal device, and n is the total number of terminal devices in the distribution radio area.

[0017] Furthermore, based on the established distribution area resource collaborative optimization control model and multi-terminal device computing power resource optimization allocation model, the low-voltage distribution area convergence terminal, combined with its own and each edge device's computing power resource status and the distribution area's operating status, selects a suitable lightweight artificial intelligence algorithm from a lightweight artificial intelligence algorithm library and executes it on the resource-rich edge devices. It then verifies the distribution area operation strategy obtained from the algorithm execution, determines whether the strategy meets the constraints of power balance and source-load-storage system, and implements the control function of the edge devices according to the strategy.

[0018] Compared with existing technologies, the present invention has the following beneficial effects: The present invention provides an edge computing method based on a lightweight artificial intelligence algorithm. This method overcomes the limitations of a single integrated terminal in a distribution substation. The integrated terminal of a low-voltage distribution substation can combine its own and each edge device's computing power resources and the substation's operating status to select a suitable lightweight artificial intelligence algorithm to execute in resource-rich edge devices. It also verifies the substation operation strategy obtained from the algorithm execution, determines whether the strategy meets the constraints of power balance and source-load-storage system, and implements the control function of the edge devices according to the strategy, thereby realizing the service response of the lightweight artificial intelligence algorithm and improving the operating efficiency of the distribution network. Attached Figure Description

[0019] Figure 1 This is a flowchart illustrating the method implementation of an embodiment of the present invention;

[0020] Figure 2 This describes the changes in resource usage and line loss rate of the integrated terminal and one edge device in the transformer area according to an embodiment of the present invention.

[0021] Figure 3 This describes the changes in resource usage and line loss rate of a single converged terminal in a distribution area in this embodiment of the invention. Detailed Implementation

[0022] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0023] It should be noted that the following detailed descriptions are exemplary and intended to provide further explanation of this application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application pertains.

[0024] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the exemplary embodiments according to this application. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0025] like Figure 1 As shown, this embodiment provides an edge computing method based on a lightweight artificial intelligence algorithm, including the following steps:

[0026] S1. Edge devices upload their real-time operating data to the converged terminal. The converged terminal analyzes the real-time operating data uploaded by edge devices such as optical storage, charging, and IoT switches to understand the operating status of the edge devices.

[0027] S2. The distribution area has a converged terminal and various edge devices. The converged terminal selects an edge device with sufficient resources to execute a lightweight artificial intelligence algorithm suitable for the operating status of the distribution area, based on the current operating status of the distribution area and the computing power resources of itself and the edge devices.

[0028] S3. After verifying the operation strategy of the transformer area obtained by the algorithm, the transformer area fusion terminal realizes the control function of the transformer area's photovoltaic storage and charging system, IoT switch and other edge devices.

[0029] (1) Data interaction mode between integrated terminal and edge device

[0030] To address the presence of converged terminals and multiple edge devices in a distribution transformer area, the area is divided into an edge device layer and a terminal device layer. Edge devices in the distribution transformer area upload real-time operational data to the converged terminal. The terminal device layer then performs the function of collecting and reporting actual operational data from the edge devices. Data can be represented as data values ​​or data blocks. When using data blocks, the corresponding protocol specifications must be followed to ensure complete data organization and parsing. The converged terminal, based on the communication protocols of the edge devices, parses the real-time operational data uploaded by edge devices such as source-load-storage and IoT switches, and integrates this data into the converged terminal. This enables the converged terminal to comprehensively and in real-time perceive the operational status of the edge devices, improving the accuracy and granularity of its perception of the connected devices.

[0031] Based on the established resource collaborative optimization control model and multi-terminal device computing power resource optimization allocation model, the low-voltage distribution substation fusion terminal, combined with its own and each edge device's computing power resource status and the substation's operating status, rationally allocates the computing resources and storage capacity of the fusion terminal and each edge device. It selects a suitable lightweight artificial intelligence algorithm from the library and executes it on the edge devices with sufficient resources. The fusion terminal verifies the substation operation strategy obtained from the algorithm execution, and determines whether the strategy meets the constraints, such as power balance constraints, source-load-storage capacity and output limits. It then determines the final lightweight artificial intelligence algorithm strategy for the current substation operation. Combined with the communication protocol of the substation edge devices, the fusion terminal controls the substation's source-load-storage system, IoT switches, and other edge devices according to the strategy, thereby improving the efficiency of the distribution network operation.

[0032] (2) Allocation of computing resources across multiple terminal devices

[0033] To address the resource allocation of various AI algorithms within a terminal device, based on the multiple computing programs deployed on the terminal, the computational power resource requirements of each program, including spectrum, channels, and cache, are pre-assessed. A multi-program runtime computational resource pool initialization allocation mechanism is then constructed. This mechanism divides the computational resources according to the time and space complexity indices of the programs. Following the computer's internal time-slice round-robin scheduling principle, each program is assigned a time slice proportional to its complexity index, achieving initial allocation of computational resources for the terminal device under multi-program scenarios. After this process, the terminal device's computing resources are virtualized into a "resource pool." When the scene in the distribution area changes, the computational resource pool determines the attributes and correlations of the input data based on the terminal device's internal algorithm model, and the computing resources are re-allocated to each program according to the initialization principle.

[0034] (3) Lightweight AI Algorithm Execution Method

[0035] With the objectives of minimizing load fluctuations in distribution transformers and maximizing the absorption of distributed energy resources in the distribution area, a lightweight artificial intelligence (AI) algorithm library is established. Based on this library, a resource-coordinated optimization control model for the distribution area is built. This model utilizes the data interaction mode of the lightweight AI algorithms to perceive and analyze the operating status of each terminal device within the current low-voltage distribution area. It then selects and executes a suitable lightweight AI algorithm from the library, addressing the specific operating conditions of the distribution area. The model solves for the operational commands of controllable resources such as power sources, loads, and energy storage within the distribution area. The dual objective function of the resource-coordinated optimization control model is as follows:

[0036]

[0037]

[0038] In the formula, P t,i P is the load value of the distribution transformer in the transformer area at time i. av P is the average daily load value of the distribution transformer in the area. e,i This is the output of the distributed energy in the transformer substation at time i.

[0039] To address the disparity in computing resources between converged terminals and various edge devices, we obtain the ledgers or technical parameter files of the converged terminals and various edge devices to understand their hardware resource status. We then construct a multi-terminal device computing resource optimization allocation model with the following objective function:

[0040]

[0041] Where, r i represents the current resource usage percentage of the i-th terminal device, and n is the total number of terminal devices in the distribution radio area.

[0042] Based on the current resource usage of the converged terminal and edge devices, the device to execute the algorithm is determined. Relying on a multi-terminal device computing power resource optimization allocation model, a balance of computing power resources between the converged terminal and edge devices in the distribution area is achieved. This solves the problem of computing power resource shortage caused by a single converged terminal executing too many artificial intelligence algorithms, and improves the response speed of the resource collaborative optimization control algorithm.

[0043] (4) Simulation verification

[0044] Simulation verification was conducted using historical data from 8:05:00 to 8:15 on a certain day for a specific transformer substation. The substation's line loss rate was used as an indicator of operational efficiency. The substation entered a typical operating state at 8:05:04, at which point two artificial intelligence algorithms needed to be executed to improve operational efficiency. When the substation had a converged terminal and one edge device, the resource consumption of the terminal device and the changes in the substation's line loss rate during this period were as follows: Figure 2 As shown; when there is only one converged terminal in the distribution area, the resource consumption of the converged terminal and the changes in the line loss rate of the distribution area are as follows. Figure 3 As shown, with multiple terminal devices, the algorithm executes faster, and the efficiency of the distribution area improves earlier. When faced with situations where more algorithms need to be executed, the algorithm execution can be completed more promptly, the execution strategy results can be obtained, and the operational efficiency of the distribution area can be improved earlier.

[0045] Table 1 shows a comparison of the average delay time for the execution of control strategies between a single converged terminal and multiple terminal devices in a distribution area. In the multi-terminal device mode, the distribution area can complete the distribution of control strategies more promptly, thereby improving the operational efficiency of the distribution area.

[0046] Table 1 Comparison of average delay time for control strategy execution

[0047]

[0048] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0049] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0050] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0051] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0052] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. An edge computing method based on a lightweight artificial intelligence algorithm, characterized in that, Includes the following steps: S1. The edge devices upload their real-time operating data to the converged terminal. The converged terminal parses the real-time operating data uploaded by the edge devices to understand the operating status of the edge devices. S2. The distribution area has a converged terminal and various edge devices. The converged terminal selects a resource-rich edge device to execute a lightweight artificial intelligence algorithm suitable for the operating status of the distribution area based on the current operating status of the distribution area and its own and the computing power resources of the edge devices. S3. After verifying the operation strategy of the distribution area obtained by the algorithm execution, the distribution area fusion terminal implements the control function of the edge device. In step S2, with the objectives of minimizing the load fluctuation of the distribution transformer and maximizing the absorption of distributed energy in the distribution area, a lightweight artificial intelligence algorithm library is established. Based on this library, a resource collaborative optimization control model for the distribution area is built. This model analyzes the operating status of each terminal device in the current low-voltage distribution area and selects a lightweight artificial intelligence algorithm from the library that is suitable for the current operating status of the distribution area. The model solves for the operating instructions of controllable resources. The dual objective function of the resource collaborative optimization control model is: In the formula, P t,i P is the load value of the distribution transformer in the transformer area at time i. av P is the average daily load value of the distribution transformer in the area. e,i It is a distributed energy system in the suburbs. Efforts made at all times; In step S2, the ledgers or technical parameter files of the converged terminal and each edge device are obtained to acquire the hardware resource information of the converged terminal and each edge device. A multi-terminal device computing power resource optimization allocation model is constructed, and its objective function is: wherein r i is the current resource occupation percentage of the i-th terminal device, and n is the total number of terminal devices in the distribution area.

2. The edge computing method based on a lightweight artificial intelligence algorithm according to claim 1, characterized in that, In step S1, the edge device of the distribution radio area uploads real-time operating data to the converged terminal. The data is represented in the form of data values ​​or data blocks. When data blocks are used, the set protocol specifications are followed. The converged terminal parses the real-time operating data uploaded by the edge device to understand the operating status of the edge device.

3. The edge computing method based on a lightweight artificial intelligence algorithm according to claim 1, characterized in that, Based on the established distribution area resource collaborative optimization control model and multi-terminal device computing power resource optimization allocation model, the low-voltage distribution area convergence terminal combines its own and each edge device's computing power resource status and the distribution area's operating status to select a matching lightweight artificial intelligence algorithm from a lightweight artificial intelligence algorithm library. It executes the algorithm on the edge device with sufficient resources, verifies the distribution area operation strategy obtained from the algorithm execution, determines whether the strategy meets the constraints of power balance and source-load-storage system, and implements the control function of the edge device according to the strategy.

Images