Search

CN-122028247-A - LED intelligent street lamp real-time control system based on edge calculation

CN122028247ACN 122028247 ACN122028247 ACN 122028247ACN-122028247-A

Abstract

The invention discloses an LED intelligent street lamp real-time control system based on edge calculation, which comprises an edge data processing module, a spatial cooperative characteristic construction module, an exogenous function self-organizing generation module, a space-time prediction and structure updating module and a dimming control module, wherein the edge data processing module is used for acquiring brightness feedback in a control box and preprocessing multi-source input to obtain a sample window set, the spatial cooperative characteristic construction module is used for generating an adjacent matrix and calculating a spatial lag brightness sequence and a diffusion input sequence, the exogenous function self-organizing generation module is used for constructing GMDH nodes, squaring and parameterizing strong influence coefficients and combining the strong influence coefficients according to a pre-interpretation mode to obtain an exogenous function sequence, the space-time prediction and structure updating module is used for constructing a space-time SARIMAX and writing back an exogenous function structure according to an information criterion, and the dimming control module is used for issuing a dimming command according to a target brightness reference value and returning feedback. The invention realizes the forward-looking dimming of the edge side street lamp, reduces the fluctuation and instruction frequency, and improves the consistency.

Inventors

  • CHEN GUIMEI

Assignees

  • 西安大和照明科技有限公司

Dates

Publication Date
20260512
Application Date
20260309

Claims (8)

  1. 1. An LED wisdom street lamp real-time control system based on edge calculation, which is characterized by comprising: The edge data processing module is used for collecting the street lamp grouping brightness feedback time sequence data and the multi-source input time sequence data in the street lamp centralized control box, preprocessing the brightness feedback time sequence data and the multi-source input time sequence data and generating a sample window set; The space cooperative characteristic construction module is used for generating an adjacent matrix according to the street lamp grouping topology, calculating a space lag brightness sequence for brightness feedback time sequence data in the sample window set, and calculating a space diffusion input sequence for multi-source input time sequence data; The exogenous function self-organizing generation module is used for constructing GMDH binary quadratic polynomial nodes, performing square parameterization on the corresponding coefficients of the strong influence features to generate non-negative coefficients to form a basic exogenous function, and introducing the weak influence features according to the sequence of combination after interpretation to generate a low-order interactive exogenous function to form an exogenous function sequence; The space-time prediction and structure updating module is used for constructing a space-time SARIMAX structure introducing space synergy and exogenous functions, writing a space hysteresis brightness sequence, an exogenous function sequence and a space diffusion input sequence, screening the exogenous function sequence according to an information criterion to form a reserved exogenous function sequence, writing a space-time SARIMAX structure, and forming a target brightness space-time prediction model; and the dimming control module is used for outputting a target brightness reference value according to the target brightness space-time prediction model, generating a dimming instruction, transmitting the dimming instruction to the street lamp driving device, collecting and executing feedback, and transmitting the feedback back to the edge data processing module.
  2. 2. The edge-calculation-based LED intelligent street lamp real-time control system according to claim 1, wherein the modules are realized by the following method: s1, acquiring a street lamp grouping brightness feedback time sequence and a multi-source input time sequence, wherein the multi-source input time sequence consists of strong influence characteristics and weak influence characteristics, and preprocessing the brightness feedback time sequence and the multi-source input time sequence to generate a sample window set; S2, generating an adjacent matrix according to the street lamp grouping topology, calculating a space lag brightness sequence based on the brightness feedback time sequence of the adjacent matrix, and calculating a space diffusion input sequence based on the multi-source input time sequence of the adjacent matrix; S3, constructing a GMDH first layer binary quadratic polynomial node, generating a non-negative coefficient by square parameterization of coefficients related to strong influence features in the polynomial node, and outputting a basic exogenous function sequence; S4, taking the basic exogenous function sequence as input, introducing weak influence characteristics according to a sequence of interpretation and combination to generate low-order interaction polynomial nodes, and outputting the exogenous function sequence; S5, constructing a space-time SARIMAX structure introducing space synergy and exogenous functions, and writing a space-time SARIMAX structure into a space-lag brightness sequence, an exogenous function sequence and a space diffusion input sequence to form a target brightness space-time prediction model; S6, performing parameter estimation on the space-time SARIMAX structure based on the sample window set, screening exogenous function sequences according to an information criterion to form reserved exogenous function sequences, and writing the reserved exogenous function sequences into the space-time SARIMAX structure; and S7, outputting a target brightness reference value based on the space-time SARIMAX structure and the reserved exogenous function sequence, generating a dimming instruction according to the target brightness reference value, issuing the dimming instruction to a street lamp driver, and acquiring and executing feedback to update a sample window set.
  3. 3. The LED intelligent street lamp real-time control system based on edge calculation of claim 2, wherein S2 specifically comprises: Obtaining street lamp grouping topology information according to physical arrangement relation and grouping connection relation of street lamps in roads, wherein the street lamp grouping topology information comprises adjacent relation among street lamp groups, and an adjacency matrix is constructed according to the grouping number; Counting the number of elements with the value of 1 in the corresponding row of the adjacent matrix for each street lamp group to obtain the number of adjacent groups, reading and summing the brightness feedback values of the adjacent groups at the sampling time for each sampling time and each street lamp group in a sample window set, dividing the sum result by the number of the adjacent groups to obtain a space lag brightness value, setting the space lag brightness value as the brightness feedback value of the street lamp group at the sampling time when the number of the adjacent groups is 0, and arranging the space lag brightness values at the sampling time according to time sequence to obtain a space lag brightness sequence; And reading and summing the input characteristic values of adjacent groups at the sampling time for each sampling time, each street lamp group and each input characteristic in the multi-source input time sequence in the sample window set, dividing the sum result by the number of the adjacent groups to obtain a space diffusion input value, setting the space diffusion input value as the input characteristic value of the street lamp group at the sampling time when the number of the adjacent groups is 0, and arranging the space diffusion input values at the sampling time according to time sequence to obtain the space diffusion input sequence.
  4. 4. The LED intelligent street lamp real-time control system based on edge calculation of claim 3, wherein S3 specifically comprises: Extracting a strong influence characteristic sequence from the sample window set, wherein the strong influence characteristic sequence comprises ambient light data and time mark data; Constructing a first layer binary quadratic polynomial node set of the GMDH, selecting two strong influence features as input variables for each binary quadratic polynomial node, and generating a polynomial expression comprising a constant term, a first term, a square term and a cross term; Square parameterization is carried out on a first order term coefficient, a square term coefficient and a cross term coefficient which correspond to strong influence characteristics in a polynomial expression, the corresponding coefficient is expressed as a square form, and a square result is taken as a coefficient value to form a binary quadratic polynomial node without negative coefficient constraint; Inputting two strong influence characteristics of each sampling moment in the sample window set into binary quadratic polynomial nodes, calculating to obtain a node output value sequence, and collecting the output value sequences of the binary quadratic polynomial nodes according to node numbers to form a basic exogenous function sequence.
  5. 5. The LED intelligent street lamp real-time control system based on edge calculation of claim 4, wherein S4 specifically comprises: Selecting a single basic exogenous function as a main input item in the basic exogenous function sequence, introducing weak influence characteristics in a sample window set as auxiliary input items, constructing a low-order polynomial expression only comprising a primary item and a cross item, and generating a low-order interaction polynomial node; The method comprises the steps of obtaining a low-order interaction polynomial node, not introducing square terms and high-order cross terms to the low-order interaction polynomial node, only keeping the interaction relation between a basic exogenous function and weak influence characteristics, inputting the basic exogenous function value and the weak influence characteristic value corresponding to each sampling time in a sample window set into the low-order interaction polynomial node, calculating to obtain a node output value sequence, and combining the output value sequence of each low-order interaction polynomial node with the basic exogenous function sequence to form an exogenous function sequence.
  6. 6. The system for controlling the intelligent LED street lamp in real time based on edge calculation according to claim 5, wherein the exogenous function self-organizing generation process is triggered by adopting a condition of firstly explaining and then combining, and the system specifically comprises the following steps: Outputting a basic exogenous function sequence serving as an interpretation layer, performing monotonic direction judgment on the output sequence corresponding to each basic exogenous function in a sample window set according to a time sequence, and recording the change direction sequence of the basic exogenous function between adjacent sampling moments; executing consistency check on the change direction sequence in a preset continuous sampling window, judging that the basic exogenous function meets the structural consistency condition when the change direction of the basic exogenous function in the continuous sampling window is consistent, and judging that the basic exogenous function does not meet the structural consistency condition when the direction of the basic exogenous function in the continuous sampling window is reversed; performing combination operation on the basic exogenous function meeting the structural consistency condition, taking the basic exogenous function as a main input item, introducing weak influence characteristics as an auxiliary input item, constructing a low-order interaction polynomial node which only comprises a primary item and a cross item, and generating a low-order interaction exogenous function; the basic exogenous function which does not meet the structural consistency condition is forbidden to introduce weak influence characteristics and the original basic exogenous function structure is kept unchanged; and combining the generated low-order interaction exogenous function with the basic exogenous function which does not execute the combination operation to form an exogenous function sequence.
  7. 7. The LED intelligent street lamp real-time control system based on edge calculation of claim 6, wherein S5 specifically comprises: Carrying out alignment analysis on the exogenous function output sequences corresponding to each street lamp group in the sample window set and the exogenous function output sequences corresponding to adjacent street lamp groups according to a time sequence; Executing time offset judgment on the output sequences of the same exogenous function in the current street lamp group and the adjacent street lamp group, wherein the time offset judgment determines the time leading relation or the time lagging relation of the exogenous function in the current street lamp group relative to the adjacent street lamp group by comparing the consistency sequence of the output change directions of the exogenous function at the adjacent sampling time; After determining the time leading relation or the time lagging relation, executing time remapping operation on the exogenous function output sequences corresponding to the current street lamp groups, wherein the time remapping operation carries out time index forward or backward movement on the exogenous function output sequences according to the determined time offset relation, and the exogenous function output sequences of the current street lamp groups are consistent with the exogenous function output sequences of adjacent street lamp groups in time evolution sequence; After the time remapping operation is completed, writing the exogenous function sequence subjected to time alignment processing into a time space SARIMAX structure, writing the street lamp grouping brightness feedback sequence into a time space SARIMAX structure, and executing joint modeling on the exogenous function sequence subjected to time alignment and the brightness feedback sequence to form a space-time prediction model of target brightness.
  8. 8. The LED intelligent street lamp real-time control system based on edge calculation of claim 7, wherein S6 specifically comprises: Performing parameter estimation on the space-time SARIMAX structures written with the exogenous function sequences on the sample window set, and performing structure selection on exogenous functions participating in modeling based on information criteria in the parameter estimation process; determining a reserved exogenous function set according to a comparison result of the information criterion, and writing the reserved exogenous function set back into a space-time SARIMAX structure to be used as an exogenous function structure; and re-executing parameter estimation based on the write-back exogenous function structure to form a space-time SARIMAX model.

Description

LED intelligent street lamp real-time control system based on edge calculation Technical Field The invention relates to the field of intelligent city lighting control, in particular to an LED intelligent street lamp real-time control system based on edge calculation. Background Along with the promotion of smart city construction, the road lighting system gradually develops from a traditional timing control mode to an intelligent and refined control mode. The existing LED street lamp system generally obtains brightness feedback information through a centralized control or remote platform, and performs dimming control by combining parameters such as ambient illumination, time information and the like so as to realize balance between energy-saving operation and basic illumination requirements. In the prior art, a scheme directly generates a dimming command according to the deviation between the currently acquired brightness value and a preset threshold value based on real-time brightness feedback, the scheme relies on instant feedback, does not model the brightness change trend, and easily solves the problems of frequent dimming or control lag under the environment rapid change or multi-path linkage scene. The other scheme introduces a time sequence prediction model to model the historical brightness data of a single street lamp or a single road section, but usually only local time information is utilized, the spatial association relationship between adjacent road sections is not fully considered, and the prediction result is difficult to reflect the integral evolution characteristics of road illumination. In the prior art, input features such as environment, traffic and the like are directly used as exogenous variables to be written into a prediction model, and exogenous variable structures are usually fixed before modeling, so that a mechanism for adaptively adjusting different input feature structures is lacked. When the exogenous input quantity is increased, the complexity of the model structure is increased, the parameter selection process depends on manual experience or simple screening rules, and stable operation is difficult to realize under the condition of edge calculation. Therefore, how to provide an LED intelligent street lamp real-time control system based on edge calculation is a problem to be solved by those skilled in the art. Disclosure of Invention The invention aims to provide an LED intelligent street lamp real-time control system based on edge calculation, which constructs a space-time SARIMAX structure integrating a space cooperative relationship, combines an exogenous function self-organization generation and information criterion driven structure updating mechanism to realize the prospective regulation control of street lamp brightness and has the advantages of high control stability, strong structure self-adaptability and low edge operation burden. According to the embodiment of the invention, the LED intelligent street lamp real-time control system based on edge calculation comprises the following components: The edge data processing module is used for collecting the street lamp grouping brightness feedback time sequence data and the multi-source input time sequence data in the street lamp centralized control box, preprocessing the brightness feedback time sequence data and the multi-source input time sequence data and generating a sample window set; The space cooperative characteristic construction module is used for generating an adjacent matrix according to the street lamp grouping topology, calculating a space lag brightness sequence for brightness feedback time sequence data in the sample window set, and calculating a space diffusion input sequence for multi-source input time sequence data; The exogenous function self-organizing generation module is used for constructing GMDH binary quadratic polynomial nodes, performing square parameterization on the corresponding coefficients of the strong influence features to generate non-negative coefficients to form a basic exogenous function, and introducing the weak influence features according to the sequence of combination after interpretation to generate a low-order interactive exogenous function to form an exogenous function sequence; The space-time prediction and structure updating module is used for constructing a space-time SARIMAX structure introducing space synergy and exogenous functions, writing a space hysteresis brightness sequence, an exogenous function sequence and a space diffusion input sequence, screening the exogenous function sequence according to an information criterion to form a reserved exogenous function sequence, writing a space-time SARIMAX structure, and forming a target brightness space-time prediction model; and the dimming control module is used for outputting a target brightness reference value according to the target brightness space-time prediction model, generating a dimming instruction, transmitting th