Street lamp control method and device based on graph neural network structure, equipment and medium

Abstract

The application discloses a street lamp control method and device based on a graph neural network structure, a device and a medium, which performs big data analysis on natural light intensity monitoring data before turning on and off the street lamp every day, forms a street lamp turning on and off time prediction algorithm suitable for various different weather conditions, dynamically predicts the illumination change curve and the street lamp turning on or off time node in the preparation stage before turning on or off the street lamp, and provides an advanced decision basis for the on-duty personnel. Through algorithm analysis of the monitoring data of multiple networked street lamp monitoring points, abnormal data of a single or several points is alarmed, and possible problems are automatically fed back. Through algorithm analysis under extreme weather in the flood season, in the case of sudden and rapid drop of daytime illumination, the change trend, rate and the like are analyzed, the possible trend of the illumination, the possibility of triggering the street lamp turning on and the possible street lamp turning on time point are dynamically predicted, the on-duty personnel is warned in advance, and temporary street lamp turning on preparation is completed.

Description

TECHNICAL FIELD

[0001] The application belongs to the technical field of street lamp control, and particularly relates to a street lamp control method and device based on a graph neural network structure, equipment and a medium. BACKGROUND

[0002] The natural light intensity level when the road lighting is turned on or off is 30lx for expressways and main roads, and 20lx for secondary trunk roads and branch roads. Currently, the turning on and off of the road lighting is mainly controlled by the monitoring terminal installed at the power supply point of the road lighting, and the control mode is time control and remote control. The automatic turning on and off schedule of the road lighting for each day of the year is written in advance in the monitoring terminal. Due to the daily change in the length of day and night throughout the year, the turning on and off time of the road lighting also changes daily, and is located between the sunrise and sunset time and the dawn and dusk time. When the weather is fine, the monitoring terminal at the scene automatically turns on and off the road lighting according to the daily schedule without intervention of the background of the monitoring center, which basically meets the requirement of the natural light intensity when the road lighting is turned on and off. When it is rainy, the monitoring center needs to issue an instruction to turn on the road lighting in advance and to delay the turning off of the road lighting. When it is extremely bad weather such as heavy rain during the flood season, the natural light intensity may also decrease to below 30lx during the day, which affects the urban traffic, and the road lighting needs to be temporarily turned on remotely and turned off remotely after the natural light intensity rises. The remote control of the turning on and off of the road lighting needs to be based on the monitoring data of the natural light intensity. Therefore, a large number of light intensity monitoring terminals are distributed in the six urban districts of Beijing to monitor the natural light intensity in real time. Although a large number of natural light intensity monitoring points are distributed, the decision-making process from the monitoring data to the turning on and off of the road lighting is still mainly based on manual judgment. This mode requires the monitoring center attendant to closely observe the weather condition and the real-time monitoring data of the natural light intensity, and to make preparations for turning on the road lighting in advance, and to issue a turning on instruction when the natural light intensity approaches the standard for turning on the road lighting. After receiving the warning of severe weather such as heavy rain, the monitoring center attendant needs to closely observe the change in the natural light intensity. When the natural light intensity abnormally decreases (which may be a slow decrease or a rapid drop within a few minutes) and approaches the standard for turning on the road lighting (30lx), the monitoring center attendant needs to make preparations for turning on the road lighting, and to issue a turning on instruction after decision-making. When the natural light intensity recovers to above 30lx and lasts for a certain period of time, a turning off instruction is issued.

[0003] The disadvantage of the prior art is that it highly depends on the observation and judgment of the detection data by manual work, and the monitoring system is insufficient in supporting the monitoring attendant in the operation of turning on the road lighting. For severe weather such as rain and snow, the monitoring attendant cannot predict in advance whether the rapidly changing light intensity may trigger the turning on condition (natural light intensity of 30lx). The current turning on and off system belongs to an overall operation, all the road lighting is turned on and off at the same time, and the influence of the region and the surrounding light is ignored, which is not conducive to saving power resources. SUMMARY

[0004] The present application aims to provide a street light control method, device, equipment and medium based on a graph neural network structure. To solve the problem of relying on manual observation and judgment of detection data, and the monitoring system lacks support for the opening operation of the monitoring personnel in the background art.

[0005] In order to achieve the above-mentioned purpose, the present application adopts the following technical solutions:

[0006] In a first aspect, a street light control method based on a graph neural network structure is provided, comprising:

[0007] The graph neural network structure is constructed by taking the street light monitoring points as nodes, wherein a single node represents the street lights in its surrounding area;

[0008] A preset point is selected as the origin, and the position relationship between all nodes is obtained by using coordinates. The straight-line distance between nodes is used as the length of the edge between nodes. Each node stores its corresponding historical data information;

[0009] A plurality of initialization nodes are selected from the nodes in the graph neural network structure, neighbor nodes of the initialization nodes are sampled, and neighbors of the neighbor nodes are sampled. The sampled node information is aggregated to obtain neighbor node information aggregation;

[0010] Learning is performed based on the neighbor node information aggregation to obtain a feature embedding representation representing the initialization node;

[0011] Real-time illumination monitoring data of each initialization node is obtained, and the illumination monitoring data is input into the graph neural network structure. The graph neural network structure outputs the predicted switching time of the street light represented by each initialization node;

[0012] The switching instruction of the street light is generated according to the predicted switching time.

[0013] Further, in the step of each node storing its corresponding historical data information, the historical data information includes: street light monitoring point serial number, street light monitoring point coordinates, actual on / off time, actual on / off time illumination intensity, and 30 minutes before and after the actual on / off time.

[0014] Further, the step of sampling the neighbor nodes of the initialization nodes and sampling the neighbors of the neighbor nodes, and aggregating the sampled node information to obtain neighbor node information aggregation, specifically includes:

[0015] First, all nodes are encoded using one-hot encoding, and the nodes are represented by a string of numbers;

[0016] The neighbor nodes of the initialization nodes are sampled, and the neighbors of the neighbor nodes are sampled.

[0017] The mean operation is performed on the initialized node and the sampled neighbor nodes to obtain neighbor node information aggregation.

[0018] Further, the step of learning based on the neighbor node information aggregation to obtain a feature embedding expression representing the initialized node includes repeatedly performing sampling and mean operation to obtain the feature embedding expression representing the node.

[0019] Further, the step of obtaining the feature embedding expression representing the initialized node further includes the step of:

[0020] The historical data of the actual light-on time is obtained as a label to construct a loss function.

[0021] The graph neural network structure is optimized by using the back propagation method to minimize the loss function.

[0022] Further, the loss function is: Loss = |T predicted light-on time - T actual light-on time | + |T predicted light-off time - T actual light-off time |.

[0023] Further, the step of outputting the current predicted on-off time of each initialized node representing the street light by the graph neural network structure further includes the step of:

[0024] A preset street light on-off time table is provided, wherein the street light on-off time table includes the on-off time range of the street light.

[0025] The predicted on-off time is compared with the on-off time range, and when the predicted on-off time is not within the on-off time range, an alarm information is generated.

[0026] In a second aspect, a street light control device based on a graph neural network structure is provided, which includes:

[0027] A graph network construction module is configured to construct a graph neural network structure by taking street light monitoring points as nodes, wherein a single node represents the street lights in its surrounding area.

[0028] A coordinate module is configured to select a preset point as an origin, obtain the positional relationship between all nodes by using coordinates, and store the historical data information corresponding to each node.

[0029] A node sampling module is configured to select a plurality of initialized nodes from the nodes in the graph neural network structure, sample the neighbor nodes of the initialized nodes, sample the neighbors of the neighbor nodes, aggregate the sampled node information, and obtain neighbor node information aggregation.

[0030] The node information learning module is configured to learn based on the neighbor node information aggregation to obtain a feature embedding expression representing the initialization node.

[0031] The prediction module is configured to obtain real-time illumination monitoring data of each initialization node, input the illumination monitoring data into the graph neural network structure, and output a current predicted switching time of each initialization node represented by the light from the graph neural network structure.

[0032] The instruction generation module is configured to generate a switching instruction of the light according to the predicted switching time.

[0033] In a third aspect, an electronic device is provided, which includes a processor and a memory, and the processor is configured to execute a computer program stored in the memory to implement the above-mentioned light control method based on the graph neural network structure.

[0034] In a fourth aspect, a computer readable storage medium is provided, which stores at least one instruction, and the at least one instruction is executed by a processor to implement the above-mentioned light control method based on the graph neural network structure.

[0035] The present application has the following beneficial effects:

[0036] The method of the present application fully considers historical monitoring data, combines the graph neural network technology, aggregates monitoring information of surrounding nodes, can predict switching time of each node, can make decisions under different regional and weather conditions, and can save energy and help on-duty personnel make decisions. BRIEF DESCRIPTION OF DRAWINGS

[0037] The drawings accompanying the specification of the present application serve to provide a further understanding of the present application, and the illustrative embodiments of the present application and their descriptions serve to explain the present application, and do not constitute an improper limitation on the present application. In the drawings:

[0038] Figure 1 FIG. 1 is a schematic diagram of a light control method based on a graph neural network structure according to the present application;

[0039] Figure 2 FIG. 3 is a node graph of a graph network structure according to the present application;

[0040] Figure 3 FIG. 4 is a structural block diagram of a light control device based on a graph neural network structure according to the present application;

[0041] Figure 4 FIG. 5 is a structural block diagram of an electronic device according to the present application. DETAILED DESCRIPTION

[0042] The application will be described in detail below with reference to the drawings and in conjunction with the embodiments. It should be noted that the embodiments in the present application and the features in the embodiments can be combined with each other without conflict.

[0043] The following detailed description is exemplary and is intended to provide further explanation of the application. Unless otherwise defined, all technical terms used in the present application have the same meanings as generally understood by those skilled in the art to which the present application belongs. The terms used in the present application are only for the purpose of describing the specific embodiments and are not intended to limit the exemplary embodiments according to the present application.

[0044] Embodiment 1

[0045] As Figure 1 shown, the present application trains a model according to historical monitoring data of monitors distributed in various regions by a graph neural network structure algorithm, and predicts the light-on and light-off time of the area around each street light monitoring point according to the current light intensity, thereby providing a light-on decision basis in advance and supporting the on-duty personnel to make preparations for remote operation of light-on and light-off in advance.

[0046] (1) A graph neural network structure building module based on historical monitoring data, in which the following steps are completed:

[0047] The monitoring data of the last month is selected, and the monitoring data is cleaned to obtain the illumination data of each street light monitoring point for half an hour before and after the light-on and light-off time of the street light, the historical reasonable light-on and light-off time, and the coordinate relationship between the street light monitoring points.

[0048] The format is shown in Table 1 below.

[0049] Table 1: Node information table of street light monitoring point graph neural network structure

[0050]

[0051] According to the above collected and cleaned monitoring data, a graph neural network structure is constructed, the street light monitoring points are taken as nodes in the graph neural network structure, the center point of Beijing is selected as the origin in the present scheme, the position relationship between all other nodes is obtained by using coordinates, the straight line distance between the nodes is taken as the length of the edge between the nodes, and the actual on / off time of each node for positive and negative 30 minutes and the corresponding illumination intensity are stored as the attached information of the node. The graph neural network structure is built as shown in Figure 2 .

[0052] Since each street light monitoring point is represented as a node, the data of a single street light monitoring point can represent the situation of the street lights in the surrounding area of the street light monitoring point. Figure 2Only part of the nodes are displayed, the graph neural network structure can better describe the relationship between nodes, such as the distance between node 5 and node 1, which is closer than the distance between node 5 and node 4, and the distance is represented as the weight of the edge in the graph.

[0053] (2) The graph network embedding expression module, in which the following steps are completed:

[0054] The graph network embedding expression is divided into three steps: sampling nodes in the graph network; aggregating the neighbor nodes of the nodes; and learning according to the aggregated information nodes.

[0055] It should be noted that the graph network embedding expression module in the present scheme is an information transmission framework, and the information (structure and attached information) between nodes is transmitted between points and points. Through an aggregation function, a node can aggregate the information contained in its neighbor nodes, and update the information of the current node through an update function to complete an iteration of the information transmission process. After multiple iterations, each node can aggregate the information of higher-order neighbor nodes.

[0056] 1) Sampling the information of the nodes in the graph network, mainly sampling the neighbor nodes (first-order neighbors) of the given initialization nodes in the graph, and sampling the neighbors (first-order neighbors of the neighbors) of the first-order neighbors. Among them, multiple initialization nodes are set in the graph, and the multiple initialization nodes are aggregated respectively.

[0057] 2) The aggregation of the neighbor nodes of the nodes mainly uses one-hot encoding (in order to let the nodes can use a string of numbers to represent) after the nodes, and the mean value operation is performed on the current node and its neighbor nodes, as shown in the following formula:

[0058]

[0059] Among them, v represents the current node, is the k-level embedding vector of v node, that is, the average value result obtained after averaging (k is usually taken as 2, that is, sampling 2 layers of neighbor nodes), MEAN is the mean operation, and N(v) represents the neighbor nodes of v.

[0060] 3) Learning according to the aggregated information nodes mainly repeats the method in step (2) k times to obtain the final That is, the feature embedding expression of the node (in the actual scene, a 1*128-dimensional matrix is used). Specifically, the result of the first learning is h 1 v , the second result will get h 1 v Higher-order embedding h 2 v , the next time will be more sufficient than the previous expression of this node, because h1 v Only the information of several neighbors is fused, while h 2 v The neighbor information of the neighbors is fused.

[0061] (3) Node time prediction module

[0062] After obtaining the embedding expression matrix of each node in step (2), a predicted light-on time can be output through a fully connected network (Fully Connected). A preset light-on / off time table is provided, wherein the light-on / off time table includes a light-on / off time range of the street lamp; the predicted light-on / off time is compared with the light-on / off time range, and when the predicted light-on / off time is not within the light-on / off time range, an alarm information is generated.

[0063] In a preferred scheme, the actual light-on / off time in the past month is taken as a label, and a loss function is constructed as follows:

[0064] Loss = |T predicted light-on time - T actual light-on time | + |T predicted light-off time - T actual light-off time |

[0065] Through the loss constraint and the back propagation of the network, the loss becomes smaller and smaller until it converges, that is, T predicted light-on / off time - T actual light-on / off time can be approximately equal. At this time, the monitoring data of the current day is input, and the light-on / off time of each node can be predicted.

[0066] The scheme automatically analyzes the accumulated illumination monitoring data, and fits the predicted illumination change curve in advance for the situations such as early light-on, delayed light-off, temporary light-on, and dynamically adjusts and gradually converges to determine the interval value or relatively accurate predicted value of the light-on / off time point, thereby providing a light-on decision basis for the on-duty personnel in advance and supporting the on-duty personnel to make preparations for remote operation of light-on / off in advance. At the same time, for a plurality of networked monitoring nodes, the possible problems of each monitoring node can be automatically analyzed and actively alarmed to help the on-duty personnel to discover abnormal situations in time.

[0067] Embodiment 2

[0068] As shown in Figure 3 Fig. 1, a street lamp control device based on a graph neural network structure includes:

[0069] A graph network construction module is configured to construct a graph neural network structure by taking street lamp monitoring points as nodes, wherein a single node represents a street lamp in its surrounding area.

[0070] A coordinate module is configured to select a preset point as an origin, obtain the positional relationship between all nodes by using coordinates, and store the historical data information of each node.

[0071] A node sampling module is configured to select a plurality of initialization nodes from the nodes in the graph neural network structure, sample neighbor nodes of the initialization nodes, sample neighbors of the neighbor nodes, aggregate the sampled node information, and obtain neighbor node information aggregation.

[0072] A node information learning module is configured to learn based on the neighbor node information aggregation to obtain a feature embedding expression representing the initialization node.

[0073] A prediction module is configured to obtain real-time illumination monitoring data of each initialization node, input the illumination monitoring data into the graph neural network structure, and output the predicted switching time of the current street lamp represented by each initialization node by the graph neural network structure.

[0074] An instruction generation module is configured to generate a switching instruction of the street lamp according to the predicted switching time.

[0075] Embodiment 3

[0076] As shown in Figure 4 The present application also provides an electronic device 100 for implementing the above-mentioned street lamp control method based on the graph neural network structure. The electronic device 100 comprises a memory 101, at least one processor 102, a computer program 103 stored in the memory 101 and executable on the at least one processor 102, and at least one communication bus 104. The memory 101 can be used to store the computer program 103, and the processor 102 can implement the steps of the street lamp control method based on the graph neural network structure by running or executing the computer program stored in the memory 101 and calling the data stored in the memory 101. The memory 101 can mainly include a program storage area and a data storage area, wherein the program storage area can store an operating system, at least one application required by a function (such as a sound playing function, an image playing function, etc.), etc.; the data storage area can store data (such as audio data) created according to the use of the electronic device 100, etc. In addition, the memory 101 can include a non-volatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device.

[0077] The at least one processor 102 can be a central processing unit (CPU), and can also be other general-purpose processors, a digital signal processor (DSP), an application specific integrated circuit (ASIC), a field-programmable gate array (FPGA) or other programmable logic device, a discrete gate or transistor logic device, a discrete hardware component, or the like. The processor 102 can be a microprocessor or the processor 102 can also be any conventional processor, etc. The processor 102 is a control center of the electronic device 100, and is connected to various parts of the electronic device 100 through various interfaces and lines.

[0078] The memory 101 in the electronic device 100 stores a plurality of instructions to implement a street lamp control method based on a graph neural network structure, and the processor 102 can execute the plurality of instructions to implement:

[0079] A graph neural network structure is constructed by taking a street lamp monitoring point as a node, wherein a single node represents a street lamp in the surrounding area of the node;

[0080] A preset point is selected as an origin, and the position relationship between all nodes is obtained by using coordinates, and the straight-line distance between nodes is taken as the length of the edge between nodes, and each node respectively stores the corresponding historical data information of the node;

[0081] A plurality of initialization nodes are selected from the nodes in the graph neural network structure, neighbor nodes of the initialization nodes are sampled, and neighbors of the neighbor nodes are sampled, and the sampled node information is aggregated to obtain neighbor node information aggregation;

[0082] Learning is performed based on the neighbor node information aggregation to obtain a feature embedding expression representing the initialization node;

[0083] Real-time illumination monitoring data of each initialization node is obtained, the illumination monitoring data is input into the graph neural network structure, and the graph neural network structure outputs a predicted switching time of the street lamp represented by each initialization node;

[0084] A switching instruction of the street lamp is generated according to the predicted switching time.

[0085] Embodiment 4

[0086] The modules / units integrated in the electronic device 100, if implemented in the form of software function units and sold or used as independent products, can be stored in a computer-readable storage medium. Based on such understanding, all or part of the processes in the above-mentioned embodiment methods can also be completed by a computer program instructing related hardware, and the computer program can be stored in a computer-readable storage medium. When the computer program is executed by a processor, the steps of each method embodiment described above can be implemented. The computer program includes computer program code, which can be in the form of source code, object code, executable files, or some intermediate forms. The computer-readable medium can include any entity or device capable of carrying computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, and read-only memory (ROM).

[0087] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can take the form of an entirely hardware embodiment, an entirely software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can take the form of a computer program product implemented 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.

[0088] The present application is described with reference to flowcharts and / or block diagrams of the method, device (system), and computer program product according to the embodiments of the present application. It should be understood that each flow and / or block in the flowcharts and / or block diagrams, and the combination of flows and / or blocks in the flowcharts and / or block diagrams can be implemented by computer program instructions. These computer program instructions can be provided to a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to produce a machine, so that the instructions executed by the computer or other programmable data processing devices produce a device that implements the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The device that implements the functions specified in one or more flows and / or blocks.

[0089] These computer program instructions can also be stored in a computer-readable storage medium that can guide the computer or other programmable data processing devices to work in a specific way, so that the instructions stored in the computer-readable storage medium produce a product including instruction devices that implement the functions specified in the flowcharts and / or block diagrams. Figure 1 The functions specified in one or more flows and / or blocks Figure 1 The device that implements the functions specified in one or more flows and / or blocks.

[0090] These computer program instructions can also be loaded into a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart block or blocks. Figure 1 one or more flowcharts and / or blocks Figure 1 one or more flowcharts and / or blocks

[0091] Finally, it should be noted that the above-mentioned embodiments are merely used to illustrate the technical solutions of the present application, rather than limit the technical solutions of the present application. Although the present application has been described in detail with reference to the above-mentioned embodiments, those skilled in the art should understand that the specific embodiments of the present application can be modified or replaced equivalently without departing from the spirit and scope of the present application, and any modification or equivalent replacement without departing from the spirit and scope of the present application should be covered in the protection scope of the claims of the present application.

Claims

1. A streetlight control method based on a graph neural network structure, characterized in that, The method comprises the following steps: The street lamp monitoring points are taken as nodes to construct a graph neural network structure, wherein a single node represents the street lamps in the surrounding area of the node, and the straight-line distance between nodes is taken as the weight of the edge; A preset point is selected as the origin, and the coordinates are used to obtain the positional relationship between all nodes, and the straight-line distance between nodes is taken as the length of the edge between nodes. Each node respectively stores the corresponding historical data information of the node; Selecting several initialization nodes from the nodes in the graph neural network structure, using one-hot encoding for all nodes, sampling neighbor nodes of the initialization nodes, and sampling neighbors of the neighbor nodes, aggregating the sampled node information to obtain neighbor node information aggregation, performing a mean aggregation operation on the current node and its k-layer neighbor nodes, the formula is: wherein, v represents the current node, is v the k-level embedding vector of the node, that is, the average result obtained after averaging, MEAN is the mean operation, N(v) represents v the neighbor node of The aggregation operation is repeatedly performed k times to generate a 128-dimensional feature embedding matrix; Based on the neighbor node information aggregation, learning is performed to obtain a feature embedding expression representing the initialization node, specifically including: repeatedly performing sampling and mean value operation to obtain a feature embedding expression representing the node; After obtaining the feature embedding expression representing the initialization node, the historical data of the actual on-off time are obtained as labels to construct a loss function. The graph neural network structure is optimized by using the back propagation method to minimize the loss function. The loss function is: Loss=|T_ predicted on-time - T_ actual on-time|+|T_ predicted off-time - T_ actual off-time|. In the process of constructing the graph neural network, the monitoring data of the last month is selected, and the monitoring data is cleaned to obtain the illumination data of each street lamp monitoring point for each half hour before and after the on-off time of the street lamp, the historical reasonable on-off time, and the coordinate relationship between the street lamp monitoring points; Real-time illumination monitoring data of each initialization node is obtained, and the illumination monitoring data is input into the graph neural network structure. The graph neural network structure outputs the current predicted on-off time of the street lamp represented by each initialization node. A preset street lamp on-off time table is provided, wherein the street lamp on-off time table includes the on-off time range of the street lamp. The predicted on-off time is compared with the on-off time range. When the predicted on-off time is not within the on-off time range, an alarm information is generated. A street lamp on-off instruction is generated according to the predicted on-off time.

2. A street light control device based on a graph neural network structure for implementing the method of claim 1, characterized in that, The method comprises the following steps: The graph network construction module is configured to construct a graph neural network structure by taking street lamp monitoring points as nodes, wherein a single node represents the street lamps in the surrounding area of the node. The coordinate module is configured to select a preset point as the origin, use coordinates to obtain the positional relationship between all nodes, and take the straight-line distance between nodes as the length of the edge between nodes. Each node respectively stores the corresponding historical data information of the node. The node sampling module is configured to select a plurality of initialization nodes from the nodes in the graph neural network structure, sample the neighbor nodes of the initialization nodes, sample the neighbors of the neighbor nodes, aggregate the sampled node information, and obtain neighbor node information aggregation. The node information learning module is configured to learn based on the neighbor node information aggregation to obtain a feature embedding expression representing the initialization node. A prediction module is configured to acquire real-time illumination monitoring data of each initialization node, input the illumination monitoring data into the graph neural network structure, and output a predicted switching time of each initialization node representing a current street lamp. An instruction generation module is configured to generate a switching instruction of the street lamp according to the predicted switching time.

3. An electronic device, comprising: A computer readable storage medium stores at least one instruction, and the at least one instruction is executed by a processor to implement the graph neural network structure-based street lamp control method.

4. A computer-readable storage medium, characterized in that, The computer readable storage medium stores at least one instruction, and the at least one instruction is executed by a processor to implement the graph neural network structure-based street lamp control method.

Images