CN115460745A

CN115460745A - Street lamp control method, device, equipment and medium based on graph neural network structure

Info

Publication number: CN115460745A
Application number: CN202211202875.0A
Authority: CN
Inventors: 白鹭; 陈春光; 陈壬贤; 赵弘昊; 解伟
Original assignee: State Grid Corp of China SGCC; State Grid Beijing Electric Power Co Ltd
Current assignee: State Grid Corp of China SGCC; State Grid Beijing Electric Power Co Ltd
Priority date: 2022-09-29
Filing date: 2022-09-29
Publication date: 2022-12-09

Abstract

The invention discloses a street lamp control method, a device, equipment and a medium based on a graph neural network structure, which are used for carrying out big data analysis on monitoring data of natural illuminance before turning on and turning off a lamp every day to form a lamp turning-on and lamp turning-off time prediction algorithm suitable for various different weather conditions, dynamically pre-judging an illuminance change curve and a lamp turning-on or lamp turning-off time node in the whole preparation stage before turning on or turning off the lamp, and providing an advanced decision basis for operators on duty. The data condition of a plurality of networked street lamp monitoring points is analyzed through an algorithm, an alarm is given for single or a plurality of point monitoring abnormal data, and possible problems are automatically fed back. The method comprises the steps of analyzing the sudden and rapid falling condition of illumination in the daytime in extreme weather in the flood season through an algorithm, analyzing the change trend, the speed and the like, dynamically predicting the possible trend of the illumination, the possibility of triggering the light-on and the possible light-on time point, giving early warning to an operator on duty in advance, and making temporary light-on preparation.

Description

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

Technical Field

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

Background

The natural illumination level when the road lighting is turned on or off is 30lx for the expressway and the main trunk, and 20lx for the secondary trunk and the branch. At present, the switch of the street lamp is mainly controlled by a monitoring terminal arranged at a power supply point of the street lamp, and the control mode is a time control mode and a remote control mode. An automatic lamp switching timetable every day throughout the year is written in the monitoring terminal in advance. Because the day-night length of the whole year every day changes day by day, the lamp switching time of the street lamp also changes day by day, and the lamp switching time is positioned between the sunrise and sunset time and the morning and evening shadow time. When the weather is clear, the background of the monitoring center is not interfered, and the on-site monitoring terminal automatically switches on and off the lamp on time according to the current day schedule, so that the natural illumination requirement when the lamp is switched on and off can be basically met. When weather such as overcast and rainy days is met, a background of the monitoring center needs to send out an instruction to carry out the operations of turning on the lamp in advance and turning off the lamp in a delayed mode. When severe weather such as strong rainfall in short flood season is encountered, the natural illuminance may also fall below 30lx in the daytime, which affects urban traffic trip, and at this time, the street lamp also needs to be remotely controlled to be temporarily turned on, and the lamp is remotely controlled to be turned off after the natural illuminance rises. In rainy days and other weather, the remote control lamp switching needs to be based on the monitoring data of natural illuminance. Therefore, a large number of illuminance monitoring terminals are distributed in six areas of Beijing City for monitoring the natural illuminance in real time. At present, although a large number of natural illumination street lamp monitoring points are distributed, the decision process from monitoring data to lamp turning-on and lamp turning-off operations is mainly determined by manual judgment. In the method, a monitoring center attendant needs to pay close attention to the weather condition of the day and the change condition of the real-time monitoring data of the natural illuminance, prepare for turning on the lamp in advance, and send a lamp-on instruction when the natural illuminance is close to the lamp-on standard. After receiving early warning of severe weather such as heavy rainfall, the change situation of the natural illumination intensity is closely concerned. When the natural illumination is abnormally reduced (possibly slowly reduced or rapidly dropped within a few minutes) and approaches the light-on standard (30 lx), the light-on preparation is made, and a light-on instruction is sent out after a decision program. And after the natural illumination returns to be above 30lx and continues for a certain time, a light-off instruction is sent out.

The prior art has the defects that the monitoring system is highly dependent on manual observation and judgment of detection data, and the monitoring system is insufficient in support for the lighting-on operation of monitoring operators on duty. In severe weather such as rain and snow, the person on duty cannot predict in advance whether the rapidly changing illumination may trigger the light-on condition (natural illumination 30 lx). The existing lamp switching system is operated integrally, all street lamps are switched on and off at the same time, the influence of regions and surrounding light rays is ignored, and the saving of electric power resources is not facilitated.

Disclosure of Invention

The invention aims to provide a street lamp control method, a street lamp control device, street lamp control equipment and a street lamp control medium based on a graph neural network structure. The problem that a monitoring system is insufficient in supporting the lighting-on operation of monitoring operators on duty due to the fact that detection data are observed and judged manually in the background technology is solved.

In order to achieve the purpose, the invention adopts the following technical scheme:

in a first aspect, a street lamp control method based on a graph neural network structure is provided, which includes:

constructing a graph neural network structure by taking the street lamp monitoring points as nodes; wherein, a single node represents the street lamp of the surrounding area;

selecting a preset point as an origin, obtaining the position relation among all nodes by using coordinates, taking the linear distance among the nodes as the length of the edge among the nodes, and respectively storing corresponding historical data information in each node;

selecting a plurality of initialization nodes from nodes in a graph neural network structure, sampling neighbor nodes of the initialization nodes and neighbors of the neighbor nodes, and aggregating the sampled node information to obtain neighbor node information aggregation;

learning is carried out based on the neighbor node information aggregation, and a feature embedded expression representing the initialization node is obtained;

acquiring real-time illumination monitoring data of each initialization node, inputting the illumination monitoring data into the graph neural network structure, and outputting the current predicted switching time of the street lamp represented by each initialization node by the graph neural network structure;

and generating a switch instruction of the street lamp according to the predicted switch time.

Further, in the step of storing history data information corresponding to each node, the history data information includes: the number of the street lamp monitoring points, the coordinates of the street lamp monitoring points, the actual lamp turning on/off time, the illumination intensity at the actual lamp turning on/off time and the corresponding illumination intensity 30 minutes before and after the actual lamp turning on/off time.

Further, sampling neighbor nodes of the initialization node, sampling neighbors of the neighbor nodes, and aggregating the sampled node information to obtain neighbor node information aggregation, specifically comprising the steps of;

firstly, using one-hot coding to all nodes, and using a string of numbers to represent the nodes;

sampling neighbor nodes of an initialization node, and sampling neighbors of the neighbor nodes;

and carrying out an averaging operation on the initialized node and the sampled neighbor nodes to obtain neighbor node information aggregation.

Further, the step of learning based on the neighbor node information aggregation to obtain the feature embedded expression representing the initialization node specifically includes: and repeatedly carrying out sampling and averaging operation to obtain the characteristic embedded expression representing the node.

Further, after the step of obtaining the feature embedded expression characterizing the initialization node, the method further includes the steps of:

acquiring historical data of actual light on-off time as a label, and constructing a loss function;

by utilizing a back propagation method, the structure of the graph neural network is optimized, and the loss function is minimized.

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 |.

Further, after the step of outputting the current predicted switching time of the street lamp represented by each initialization node by the graph neural network structure, the method further comprises the following steps:

presetting a street lamp switching time comparison table; the street lamp switching time comparison table comprises a street lamp switching time range;

and comparing the predicted switching time with the switching time range, and generating alarm information when the predicted switching time is not in the switching time range.

In a second aspect, a street lamp control device based on a graph neural network structure is provided, which includes:

the graph network construction module is used for constructing a graph neural network structure by taking the street lamp monitoring points as nodes; wherein, a single node represents the street lamp of the surrounding area;

the coordinate module is used for selecting a preset point as an origin, obtaining the position relation among all nodes by using coordinates, taking the straight line distance among all nodes as the length of the edge among all nodes, and storing the historical data information corresponding to each node;

the node sampling module is used for selecting a plurality of initialization nodes from nodes in a graph neural network structure, sampling neighbor nodes of the initialization nodes and neighbors of the neighbor nodes, and aggregating sampled node information to obtain neighbor node information aggregation;

the node information learning module is used for learning based on the neighbor node information aggregation to obtain a characteristic embedded expression representing the initialization node;

the prediction module is used for acquiring real-time illumination monitoring data of each initialization node, inputting the illumination monitoring data into the graph neural network structure, and outputting the current predicted switching time of the street lamp represented by each initialization node by the graph neural network structure;

and the instruction generating module is used for generating a switching instruction of the street lamp according to the predicted switching time.

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

In a fourth aspect, a computer-readable storage medium is provided, where at least one instruction is stored, and when executed by a processor, the at least one instruction implements the above-mentioned street lamp control method based on a graph neural network structure.

The beneficial effects of the invention are as follows:

the method provided by the invention has the advantages that under the condition of fully considering historical monitoring data, the monitoring information of surrounding nodes is aggregated by combining the graph neural network technology, the light on/off time of each node can be predicted, the decision under different weather conditions in different areas can be better handled, the energy is saved, and the decision making of operators on duty can be better assisted.

Drawings

The accompanying drawings, which are incorporated in and constitute a part of this application, illustrate embodiments of the invention and, together with the description, serve to explain the invention and not to limit the invention. In the drawings:

FIG. 1 is an architecture diagram of a street lamp control method based on a graph neural network structure according to the present invention;

FIG. 2 is a diagram of a graph network fabric node in accordance with the present invention;

FIG. 3 is a block diagram of a street lamp control device based on a neural network structure of the present invention;

fig. 4 is a block diagram of an electronic device according to the present invention.

Detailed Description

The present invention will be described in detail below with reference to the embodiments with reference to the attached drawings. It should be noted that the embodiments and features of the embodiments in the present application may be combined with each other without conflict.

The following detailed description is exemplary in nature and is intended to provide further explanation of the invention as claimed. Unless otherwise defined, all technical terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein is for the purpose of describing particular embodiments only and is not intended to be limiting of exemplary embodiments according to the invention.

Example 1

As shown in figure 1, the invention carries out model training according to historical monitoring data of monitors distributed in various regions by a graph neural network structure algorithm, predicts the lamp switching-on and switching-off time of the street lamps in the regions around each street lamp monitoring point according to the illumination intensity at the current moment, provides a lamp switching-on decision basis in advance and supports an operator on duty to carry out remote operation lamp switching-on and switching-off preparation in advance.

(1) The method comprises the following steps of (1) constructing a module based on a graph neural network structure of historical monitoring data, wherein the following steps are completed in the module:

and selecting the monitoring data of the last month, cleaning the monitoring data to obtain the illuminance data of each street lamp monitoring point in each half hour before and after the lamp switching time of the street lamp, the historical reasonable lamp switching time and the coordinate relationship between the street lamp monitoring points.

The format is shown in table 1 below.

Table 1 street lamp monitoring point diagram neural network structure node information table

According to the collected and cleaned monitoring data, a graph neural network structure is constructed, street lamp monitoring points are used as nodes in the graph neural network structure, a Beijing city central point is selected as an original point in the scheme, the position relation among all other nodes is obtained by using coordinates, the linear distance among the nodes is used as the length of edges among the nodes, the actual lamp on/off time and the corresponding illumination intensity corresponding to each node are stored as the attached information of the nodes, and the graph neural network structure is built and finished as shown in FIG. 2.

Because each street lamp monitoring point is represented as a node, the single street lamp monitoring point data can represent the street lamp condition of the area around the street lamp monitoring point, only part of the nodes are shown in fig. 2, and the graph neural network structure can better describe the relationship between the nodes, for example, the distance between the node No. 5 and the node No. 1 is closer than the distance between the node No. 5 and the node No. 4, and the distance is represented as the weight of the edge in the graph.

(2) The graph network is embedded with an expression module, and the following steps are completed in the module:

the graph network embedding expression is subdivided into three steps: sampling nodes in a graph network; aggregating neighbor nodes of the node; and learning according to the aggregated information nodes.

It should be noted that the graph network embedded expression module in this scheme is an information transfer framework, information (structure and attached information) between nodes is transferred between points, through an aggregation function, a node can aggregate information contained in its neighbor nodes, and update information of the current node through an update function, thereby completing an iterative information transfer process, and through multiple iterations, each node can aggregate information of higher neighbor nodes.

1) The method for sampling the information of the network nodes of the graph mainly comprises the steps of giving an initialization node from the graph, sampling neighbor nodes (first-level neighbors) of the initialization node, and re-sampling the first-level neighbors (first-level neighbors of the neighbors). In the figure, a plurality of initialization nodes are set, and the initialization nodes are aggregated respectively.

2) The neighbor node aggregation of a node is mainly performed by performing an averaging operation on a current node and neighbor nodes thereof after one-hot coding is performed on the node (in order to enable the node to be represented by a string of numbers), and the formula is as follows:

where v represents the current node,

the MEAN is a k-level embedded vector of the v node, that is, an average result obtained after averaging (k is usually 2, that is, it is considered that 2-layer neighbor nodes are sampled), MEAN is an averaging operation, and N (v) represents a neighbor node of v.

3) The learning is carried out according to the aggregated information nodes, the method in the step (2) is repeated for k times, and the final result is obtained

I.e. the feature-embedded representation characterizing the node (a matrix of 1 x 128 dimensions is used in the actual scene). Specifically, the result of the first learning is h ¹ _v The second result will be the ratio h ¹ _v Higher order embedded h ² _v The next time this node is expressed more fully than the last time, since h ¹ _v Only information of surrounding neighbor nodes is fused, and h ² _v Neighbor information of neighbors is fused.

(3) Node time prediction module

And (3) after the embedded expression matrix of each node in the step (2) is obtained, outputting the predicted light on-off time through a full-connection network (full Connected). Presetting a street lamp switching time comparison table; the street lamp switching time comparison table comprises a street lamp switching time range; and comparing the predicted switching time with the switching time range, and generating alarm information when the predicted switching time is not in the switching time range.

In the preferred scheme, the actual light switching time in the past month time is used as a label, and a loss function is constructed as follows:

loss = | T _ predicted light-on time-T _ actual light-on time | + | T _ predicted light-off time-T _ actual light-off time

Through Loss constraint and backward propagation of a network, loss is smaller and smaller until convergence is reached, namely T _ predicted on/off time-T _ actual on/off time can be approximately equal, and the on/off time of each node can be predicted by inputting monitoring data of the current day.

According to the scheme, automatic big data analysis is carried out on accumulated illumination monitoring data, a pre-judged illumination change curve is fitted in advance for conditions such as early light-on, late light-off, temporary light-on and the like, dynamic adjustment and gradual convergence are carried out, an interval value of light-on and light-off time points or a relatively accurate predicted value is determined, a light-on decision basis is provided for an operator on duty in advance, and the operator on duty is supported to make remote operation light-on and light-off preparation in advance. Meanwhile, aiming at a plurality of networked monitoring nodes, the problem that each monitoring node is possibly wrong is automatically analyzed, an alarm is actively given out, and the on-duty personnel can be helped to find out abnormal conditions in time.

Example 2

As shown in fig. 3, a street lamp control device based on a graph neural network structure includes:

the graph network construction module is used for constructing a graph neural network structure by taking the street lamp monitoring points as nodes; wherein, a single node represents the street lamps of the surrounding area;

the coordinate module is used for selecting a preset point as an origin, obtaining the position relation among all nodes by using coordinates, taking the linear distance among the nodes as the length of the edge among the nodes, and respectively storing the corresponding historical data information of each node;

Example 3

As shown in fig. 4, the present invention further provides an electronic device 100 for implementing the 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 may be used to store a computer program 103, and the processor 102 implements a street lamp control method based on a graph neural network structure by running or executing the computer program stored in the memory 101 and calling data stored in the memory 101. The memory 101 may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required by at least one function (such as a sound playing function, an image playing function, etc.), and the like; the storage data area may store data (such as audio data) created according to the use of the electronic apparatus 100, and the like. In addition, the memory 101 may 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 memory Card (Flash Card), at least one magnetic disk storage device, a Flash memory device, or other non-volatile solid state storage device.

The at least one Processor 102 may be a Central Processing Unit (CPU), other general purpose Processor, a Digital Signal Processor (DSP), an Application Specific Integrated Circuit (ASIC), a Field Programmable Gate Array (FPGA) or other Programmable logic device, discrete Gate or transistor logic, discrete hardware components, etc. The processor 102 may be a microprocessor or the processor 102 may be any conventional processor or the like, and the processor 102 is the control center of the electronic device 100 and connects the various parts of the entire electronic device 100 using various interfaces and lines.

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

Example 4

The integrated modules/units of the electronic device 100 may be stored in a computer-readable storage medium if implemented in the form of software functional units and sold or used as separate products. Based on such understanding, all or part of the flow in the method according to the embodiments of the present invention may also be implemented by a computer program, which may be stored in a computer-readable storage medium and used by a processor to implement the steps of the above-described embodiments of the method. Wherein the computer program comprises computer program code, which may be in the form of source code, object code, an executable file or some intermediate form, etc. The computer readable medium may include: any entity or device capable of carrying computer program code, recording medium, U.S. disk, removable hard disk, magnetic diskette, optical disk, computer Memory, and Read-Only Memory (ROM).

As will be appreciated by one skilled in the art, embodiments of the present invention may be provided as a method, system, or computer program product. Accordingly, the present invention may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present invention may 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, and the like) having computer-usable program code embodied therein.

The present invention has been described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each flow and/or block of the flowchart illustrations and/or block diagrams, and combinations of flows and/or blocks in the flowchart illustrations and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may 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, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.

These computer program instructions may also be loaded onto 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 flow or flows and/or block diagram block or blocks.

Finally, it should be noted that: although the present invention has been described in detail with reference to the above embodiments, it should be understood by those skilled in the art that: modifications and equivalents may be made to the embodiments of the invention without departing from the spirit and scope of the invention, which is to be covered by the claims.

Claims

1. A street lamp control method based on a graph neural network structure is characterized by comprising the following steps:

constructing a graph neural network structure by taking street lamp monitoring points as nodes; wherein, a single node represents the street lamps of the surrounding area;

2. The method for controlling street lamps based on the graph neural network structure according to claim 1, wherein in the step of storing the historical data information corresponding to each node, the historical data information comprises: the number of the street lamp monitoring points, the coordinates of the street lamp monitoring points, the actual lamp turning on/off time, the illumination intensity at the actual lamp turning on/off time and the corresponding illumination intensity 30 minutes before and after the actual lamp turning on/off time.

3. The street lamp control method based on the graph neural network structure as claimed in claim 1, wherein the steps of sampling the neighbor nodes of the initialization node, sampling the neighbors of the neighbor nodes, and aggregating the sampled node information to obtain the aggregate neighbor node information include;

4. The street lamp control method based on the graph neural network structure as claimed in claim 3, wherein the step of learning based on the neighbor node information aggregation to obtain the feature embedded expression representing the initialization node specifically comprises: and repeatedly carrying out sampling and averaging operation to obtain the characteristic embedded expression representing the node.

5. The method for controlling street lamps based on the neural network structure of the figure as claimed in claim 1, wherein after the step of obtaining the feature embedded expression characterizing the initialization node, the method further comprises the steps of:

6. The method of claim 5, wherein 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.

7. The method for controlling street lamps based on the graph neural network structure as claimed in claim 1, wherein after the step of outputting the current predicted switching time of the street lamps represented by each initialization node, the method further comprises the steps of:

8. A street lamp control device based on a graph neural network structure is characterized by comprising:

the node sampling module is used for selecting a plurality of initialization nodes from nodes in a graph neural network structure, sampling neighbor nodes of the initialization nodes and neighbors of the neighbor nodes, and aggregating the sampled node information to obtain neighbor node information aggregation;

9. An electronic device, comprising a processor and a memory, wherein the processor is configured to execute a computer program stored in the memory to implement the method for controlling a street lamp based on a graph neural network structure according to any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that the computer-readable storage medium stores at least one instruction, which when executed by a processor, implements the graph neural network structure-based street lamp control method according to any one of claims 1 to 7.