An intelligent LED lighting energy-saving optimization method based on AI edge computing

Abstract

The application belongs to the technical field of lighting energy-saving optimization, and discloses an intelligent LED lighting energy-saving optimization method based on AI edge calculation; comprising the following steps: S1: collecting the lighting data of each sports field in the sports venue in real time.The application realizes the change from passive response to active prevention by setting an unexpected lighting coefficient prediction model, significantly improves the safety and intelligent management level of the sports venue, and the traditional unexpected lighting processing relies on manual discovery and manual control, and has the problems of response lag and incomplete coverage.The system can predict in advance that unexpected lighting may be needed in other areas, and actively adjust the lighting strategy.In an emergency, through emergency exit correlation degree analysis, the system can intelligently identify the key nodes on the evacuation path, and preferentially ensure the lighting of these areas in the emergency mode, and also dynamically optimize the lighting path according to the real-time personnel distribution, to guide the personnel to quickly and safely evacuate.

Description

Technical Field

[0001] This invention relates to the field of lighting energy-saving optimization technology, and more specifically, to an intelligent LED lighting energy-saving optimization method based on AI edge computing. Background Technology

[0002] With the popularization of national fitness activities and the development of the sports industry, the number of indoor sports venues such as badminton courts and basketball courts has grown rapidly. As an important part of the operation of the venues, the lighting system is not only related to the sports experience and the fairness of the competition, but also directly affects the operating costs and energy consumption.

[0003] Insufficient emergency and accident lighting response capabilities are a significant issue. Sports venues frequently experience equipment drops and personnel accidents, necessitating temporary enhancements to lighting in specific areas. Existing systems either rely on manual detection and control, resulting in delayed responses, or employ fixed emergency lighting modes, leading to poor flexibility and high energy consumption. This is particularly problematic in large venues, where managers struggle to promptly detect all emergencies, impacting service quality and user experience. Existing intelligent solutions largely depend on centralized cloud processing. While some venues are experimenting with intelligent lighting control with the development of IoT technology, most adopt a "sensor-cloud-controller" architecture. This architecture suffers from high network latency, high bandwidth requirements, and privacy risks. For sports requiring rapid response, such as high-speed badminton, cloud processing delays can lead to untimely control. Summary of the Invention

[0004] To address the problems in the background technology, this invention proposes an intelligent LED lighting energy-saving optimization method based on AI edge computing.

[0005] To achieve the above objectives, the present invention provides the following technical solution: a smart LED lighting energy-saving optimization method based on AI edge computing, comprising the following steps: S1: Real-time collection of lighting data for each sports field within the sports venue, and normalization processing of the lighting data; S2: Generate a lighting status score for each used site based on the processed data; S3: Decision on whether to activate auxiliary lighting program for the used site based on lighting status score and preset lighting status score threshold; S4: Obtain historical accidental lighting coefficient data and build an accidental lighting coefficient prediction model, and use the accidental lighting coefficient prediction model to predict the accidental lighting coefficient of the exclusive lighting area; S5: Determine the ceiling light lighting mode for the exclusive lighting area based on the predicted accidental lighting coefficient and the preset accidental lighting activation threshold.

[0006] Furthermore, the sports venue is equipped with multiple sports fields, and multiple LED ceiling lights are evenly installed above the sports venue. Each sports field is equipped with a corresponding number of LED ceiling lights. These ceiling lights are designated as field ceiling lights. The area outside the sports fields is designated as the path within the venue, and ceiling lights are also installed above the path. Sports fields where people are active are designated as used fields, and sports fields where no people are active are designated as unused fields. The ceiling lights of used fields are in high brightness mode, and the ceiling lights of unused fields are in low brightness mode. The lighting data includes the number of people present in the used fields, ambient illuminance, natural light utilization rate, glare index, and visual task complexity. The presence of personnel is determined by detecting and counting targets using infrared pyroelectric sensor arrays deployed at the site boundary or millimeter-wave radar on top, and then the presence of personnel is normalized. Illuminance sensors installed on both sides of the site or on the netting posts are used to measure the horizontal illuminance, and the average value is taken to eliminate the influence of local shadows to obtain the ambient illuminance. The ambient illuminance is then normalized.

[0007] In the formula, To normalize ambient illuminance, This represents the measured ambient illuminance. This represents the minimum standard illuminance required for this type of sport; The natural light utilization rate U is calculated by comparing the changes in total illuminance when artificial light sources are turned on and off. The glare index is measured using a dedicated glare measurement sensor, and then normalized.

[0008] In the formula, The normalized glare index, The glare index is the measured value. This is the maximum acceptable glare index threshold for this sport; A ball speed monitoring system based on high-speed cameras or radar measures the ball speed, derives the visual task complexity from the ball speed, and then normalizes the visual task complexity.

[0009] In the formula, To normalize the complexity of visual tasks, The measured ball speed This is the typical maximum speed for this sport. and The weighting coefficients are obtained through training based on historical data, and T is the baseline value for the movement type. All sensor data processing and decision-making are completed on the local edge node.

[0010] Furthermore, the process of deciding whether to activate auxiliary lighting procedures for already used areas based on lighting condition scores and preset lighting condition score thresholds includes: Set an appropriate lighting status score threshold based on historical lighting status score data. The historical lighting status score data refers to the data set of lighting status scores of previously used sites. Compare the lighting status scores of the used sites with the lighting status score threshold. When the lighting status score is less than the lighting status score threshold, the auxiliary lighting program is activated. Auxiliary lighting procedures: Let the ceiling lights located around the used site be called surrounding ceiling lights. Auxiliary lighting refers to adjusting the surrounding ceiling lights from low brightness mode to high brightness mode. First, calculate the supplementary lighting efficiency value of each surrounding ceiling light for the used site. Sort them in descending order of supplementary lighting efficiency value to determine the initial number of surrounding ceiling lights to be turned on. Prioritize turning on the surrounding ceiling lights with the highest supplementary lighting efficiency value according to the initial number of lights to be turned on. After turning them on, re-evaluate the lighting status score of the used site to obtain a new lighting status score. Based on the new lighting status score, make fine adjustments to the surrounding ceiling lights.

[0011] Furthermore, the venue's three-dimensional coordinate information, including the position coordinates of each ceiling light, is accessed. = ( , , The boundary coordinates of each sports field are predefined, and the geometric center point of the used fields is calculated. = ( , ); Calculate the supplementary lighting efficiency of the surrounding canopy lights on the already used site. :

[0012] In the formula, The central luminous intensity of the surrounding ceiling lights in rated high brightness mode is obtained from the ceiling light's light distribution curve. The angle between the direction of the ceiling light axis and the vector pointing to the center of the target site is calculated using spatial geometry. m is the beam angle index of the luminaire, which is determined by the optical characteristics of the luminaire; For the j-th surrounding canopy light to the center point of the used site The straight-line distance; The occlusion attenuation factor is calculated by taking into account the occlusion effect of possible obstacles in the venue on the lighting, and is obtained through a pre-established 3D model of the venue and ray tracing simulation. n is the atmospheric attenuation coefficient, which takes into account the impact of air dust on lighting transmission and is dynamically adjusted based on air quality monitoring data. This is a correction term for distance attenuation.

[0013] Furthermore, the surrounding ceiling lights were fine-tuned based on the new lighting condition score: Before opening After the surrounding ceiling lights are installed, wait for Δt time and recalculate the new lighting status score of the used site. If the new lighting condition score is greater than the lighting condition score threshold, the new lighting condition score is subtracted from the lighting condition score threshold to obtain the score margin. A tolerance threshold is set. If the score margin is greater than the tolerance threshold, it indicates that there is over-lighting. The surrounding ceiling lights that have been turned on are arranged in ascending order of supplementary lighting efficiency value. Each surrounding ceiling light is turned off in turn and the lighting condition score is re-evaluated until the latest lighting condition score is still greater than the lighting condition score threshold and the score margin is less than the tolerance threshold. If the new lighting status score is less than the lighting status score threshold, calculate the set of available surrounding ceiling lights that are still in a low brightness state, sort them from high to low according to the supplementary lighting efficiency value, and turn on the remaining surrounding ceiling lights in turn. After turning on one surrounding ceiling light each time, re-evaluate the lighting status score until the latest lighting status score is greater than the lighting status score threshold. If turning on all available surrounding ceiling lights still fails to meet the requirement that the new lighting status score is greater than the lighting status score threshold, an alarm message will be generated, indicating that manual inspection or maintenance may be necessary.

[0014] Furthermore, the process of acquiring historical accidental lighting coefficient data and constructing an accidental lighting coefficient prediction model, and then using this model to predict the accidental lighting coefficient for a specific lighting area, includes: The accidental lighting coefficient refers to the probability coefficient of other dedicated lighting areas needing to turn on the high-brightness mode of the ceiling lights in other dedicated lighting areas due to an accidental event in the already used site; The dedicated lighting area refers to the illumination area corresponding to each ceiling light; Factors affecting the accidental lighting coefficient include: historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, and correlation with emergency exits; By utilizing existing security cameras or dedicated visual sensors in the venue, video analysis can be used to confirm falling events and record the location coordinates of the incidents to obtain the historical falling frequency of other dedicated lighting areas. Calculate all shortest paths from the used site nodes to each key functional node, and count the number of paths that pass through the nodes corresponding to the dedicated lighting area in all shortest paths to obtain the path velocities. Real-time population density is obtained by detecting the outline of heat sources and counting the number of people. By using video analytics, abnormal behavior patterns such as sudden stopping after running at high speed, prolonged collapse of people, and sudden gathering of crowds are detected to determine the frequency of abnormal events. The venue management system provides information on the types of events that can be booked for each venue; Emergency exit relevance refers to the dedicated lighting area traversed by the shortest path from the occupied site to the emergency exit; Obtain historical accidental lighting coefficient data for a single dedicated lighting area in different periods. The historical accidental lighting coefficient data includes the historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, correlation of emergency exits, and historical accidental lighting coefficient of the single dedicated lighting area in different periods. Based on the historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, correlation of emergency exits, and corresponding historical accidental lighting coefficients of the dedicated lighting area in different historical accidental lighting coefficient data, an accidental lighting coefficient prediction set is generated and divided into a training set and a test set. Convolutional neural networks are constructed, with the historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, and correlation of emergency exits from different historical accident lighting coefficient data in the training set as input data, and the corresponding historical accident lighting coefficients in the training set as output data. The convolutional neural network is trained to obtain an initial convolutional neural network. The initial convolutional neural network is validated using a test set. The initial convolutional neural network whose output is less than or equal to the preset test error threshold is used as the unexpected illumination coefficient prediction model. The historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, and correlation with emergency exits for each dedicated lighting area within the monitoring period are input into the accidental lighting coefficient prediction model to obtain the predicted accidental lighting coefficient for each dedicated lighting area.

[0015] Furthermore, the process of determining the ceiling light lighting mode for the dedicated lighting area based on the predicted unexpected lighting coefficient and the preset unexpected lighting activation threshold includes: An appropriate threshold for unexpected lighting coefficients is set based on historical coefficient data, which refers to a dataset of past unexpected lighting coefficients for designated lighting areas. The predicted unexpected lighting coefficient for each designated lighting area is compared with the threshold. If the predicted unexpected lighting coefficient is greater than the threshold, not all ceiling lights in the designated lighting area are turned on; instead, they are prioritized.

[0016] In the formula, , , and These are weighting coefficients, obtained through training based on historical data; Let be the unexpected lighting coefficient for the i-th region; Safety risk levels are categorized as high, medium, and low based on historical accident data. For use of urgency: Determine whether there are people about to enter or currently active in the area; In terms of power consumption: the estimated power consumption required to turn on the lighting in this area; The high-brightness mode of the ceiling lights in the designated lighting areas is turned on sequentially from high to low priority until the power limit is reached.

[0017] The technical effects and advantages of the intelligent LED lighting energy-saving optimization method based on AI edge computing of this invention are as follows: (1) By setting up an accidental lighting coefficient prediction model, the system has realized the transformation from passive response to active prevention, which has significantly improved the safety and intelligent management level of sports venues. Traditional accidental lighting treatment relies on manual discovery and manual control, which has problems such as delayed response and incomplete coverage. This system can predict in advance that other areas may need to be illuminated unexpectedly and actively adjust the lighting strategy. In an emergency, through emergency exit correlation analysis, the system can intelligently identify key nodes on the evacuation route and prioritize lighting in these areas in emergency mode. It will also dynamically optimize the lighting path according to the real-time personnel distribution to guide personnel to evacuate quickly and safely.

[0018] (2) By setting up a lighting status score, the lighting needs of sports venues can be accurately assessed and dynamically responded to. Under the premise of ensuring lighting quality, significant energy-saving effects have been achieved. The comprehensive evaluation system, which consists of five core variables, namely the number of people present, ambient illuminance, natural light utilization rate, glare index and visual task complexity, can more comprehensively reflect the actual lighting needs compared with the traditional single sensor control. This refined control avoids the energy waste of traditional systems. The edge computing architecture of this invention ensures the real-time control. All sensor data processing and decision-making are completed on the local edge node with low response delay, which fully meets the needs of high-speed sports scenarios. Compared with systems that rely on cloud processing, it not only responds faster, but also avoids the risk of control failure caused by network instability. Attached Figure Description

[0019] Figure 1 This is a schematic diagram of the method flow of the present invention. Detailed Implementation

[0020] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0021] Reference Figure 1 A smart LED lighting energy-saving optimization method based on AI edge computing includes the following steps: S1: Real-time collection of lighting data for each sports field within the sports venue, and normalization processing of the lighting data; S2: Generate a lighting status score for each used site based on the processed data; S3: Decision on whether to activate auxiliary lighting program for the used site based on lighting status score and preset lighting status score threshold; S4: Obtain historical accidental lighting coefficient data and build an accidental lighting coefficient prediction model, and use the accidental lighting coefficient prediction model to predict the accidental lighting coefficient of the exclusive lighting area; S5: Determine the ceiling light lighting mode for the exclusive lighting area based on the predicted accidental lighting coefficient and the preset accidental lighting activation threshold.

[0022] It should be further explained that, in the specific implementation process, the sports venue is equipped with multiple sports fields, and multiple LED canopy lights are evenly installed above the sports venue. Each sports field is equipped with a corresponding number of LED canopy lights. These canopy lights are designated as field canopy lights. The area outside the sports fields is designated as the path within the venue, and canopy lights are also installed above the path. Sports fields where people are active are designated as used fields, and sports fields where no people are active are designated as unused fields. The canopy lights of the used fields are in high brightness mode, and the canopy lights of the unused fields are in low brightness mode. The lighting data includes the number of people present in the used fields, ambient illuminance, natural light utilization rate, glare index, and visual task complexity. The presence of personnel is determined by detecting and counting targets using infrared pyroelectric sensor arrays deployed at the site boundary or millimeter-wave radar on top, and then the presence of personnel is normalized. Illuminance sensors installed on both sides of the site or on the netting posts are used to measure the horizontal illuminance, and the average value is taken to eliminate the influence of local shadows to obtain the ambient illuminance. The ambient illuminance is then normalized.

[0023] In the formula, To normalize ambient illuminance, This represents the measured ambient illuminance. The minimum standard illuminance required for this type of exercise is 300 lx. The natural light utilization rate U is calculated by comparing the changes in total illuminance when artificial light sources are turned on and off. The glare index is measured using a dedicated glare measurement sensor, and then normalized.

[0024] In the formula, The normalized glare index, The glare index is the measured value. The maximum acceptable glare index threshold for this sport is 25. A ball speed monitoring system based on high-speed cameras or radar measures the ball speed, derives the visual task complexity from the ball speed, and then normalizes the visual task complexity.

[0025] In the formula, To normalize the complexity of visual tasks, The measured ball speed This represents the typical maximum speed for this sport, specifically 300 kilometers per hour. and The weighting coefficients are obtained through training based on historical data, specifically 0.4 and 0.6 respectively. T is the baseline value for the type of sport, typically set as recreation = 0.2, training = 0.5, and competition = 0.8. All sensor data processing and decision-making are completed on the local edge node.

[0026] It should be further explained that, in the specific implementation process, the process of deciding whether to activate the auxiliary lighting program for the already used site based on the lighting status score and the preset lighting status score threshold includes: Set an appropriate lighting status score threshold based on historical lighting status score data. The historical lighting status score data refers to the data set of lighting status scores of previously used sites. Compare the lighting status scores of the used sites with the lighting status score threshold. When the lighting status score is less than the lighting status score threshold, the auxiliary lighting program is activated. Auxiliary lighting procedures: Let the ceiling lights located around the used site be called surrounding ceiling lights. Auxiliary lighting refers to adjusting the surrounding ceiling lights from low brightness mode to high brightness mode. First, calculate the supplementary lighting efficiency value of each surrounding ceiling light for the used site. Sort them in descending order of supplementary lighting efficiency value to determine the initial number of surrounding ceiling lights to be turned on. Prioritize turning on the surrounding ceiling lights with the highest supplementary lighting efficiency value according to the initial number of lights to be turned on. After turning them on, re-evaluate the lighting status score of the used site to obtain a new lighting status score. Based on the new lighting status score, make fine adjustments to the surrounding ceiling lights.

[0027] It should be further explained that, in the specific implementation process, the three-dimensional coordinate information of the venue, including the position coordinates of each ceiling light, is called upon. = ( , , The boundary coordinates of each sports field are predefined, and the geometric center point of the used fields is calculated. = ( , ); Calculate the supplementary lighting efficiency of the surrounding canopy lights on the already used site. :

[0028] In the formula, The central luminous intensity of the surrounding ceiling lights in rated high brightness mode is obtained from the ceiling light's light distribution curve. The angle between the direction of the ceiling light axis and the vector pointing to the center of the target site is calculated using spatial geometry. m is the beam angle index of the luminaire, which is determined by the optical characteristics of the luminaire. Narrow beam luminaires have a larger m value. For the j-th surrounding canopy light to the center point of the used site The straight-line distance; The occlusion attenuation factor is calculated by considering the occlusion effect of possible obstacles in the venue on the lighting, and is obtained through a pre-established 3D model of the venue and ray tracing simulation. The value ranges from 0 to 1. n is the atmospheric attenuation coefficient, which takes into account the impact of air dust on lighting transmission and is dynamically adjusted based on air quality monitoring data. This is a correction term for distance attenuation; supplemental lighting efficiency This reflects the potential contribution of the j-th path ceiling light to the effective illuminance of the target site. The larger the value, the better the effect of the luminaire on improving the lighting conditions of the target site.

[0029] It should be further explained that, during the specific implementation process, the surrounding ceiling lights were finely adjusted based on the new lighting status score: Before opening After the surrounding ceiling lights are installed, wait for Δt time (usually 2-3 seconds to ensure the lights are working stably) and recalculate the new lighting status score of the used site. If the new lighting condition score is greater than the lighting condition score threshold, the new lighting condition score is subtracted from the lighting condition score threshold to obtain the score margin. A tolerance threshold (e.g., 0.05) is set. If the score margin is greater than the tolerance threshold, it indicates that there is over-lighting. The surrounding ceiling lights that have been turned on are arranged in ascending order of supplementary lighting effectiveness value. Each surrounding ceiling light is turned off in turn and the lighting condition score is re-evaluated until the latest lighting condition score is still greater than the lighting condition score threshold and the score margin is less than the tolerance threshold. If the new lighting status score is less than the lighting status score threshold, calculate the set of available surrounding ceiling lights that are still in a low brightness state, sort them from high to low according to the supplementary lighting efficiency value, and turn on the remaining surrounding ceiling lights in turn. After turning on one surrounding ceiling light each time, re-evaluate the lighting status score until the latest lighting status score is greater than the lighting status score threshold. If turning on all available surrounding ceiling lights still fails to meet the requirement that the new lighting status score is greater than the lighting status score threshold, an alarm message will be generated, indicating that manual inspection or maintenance may be necessary.

[0030] It should be further explained that, in the specific implementation process, the process of acquiring historical accidental lighting coefficient data and constructing an accidental lighting coefficient prediction model, and then using the accidental lighting coefficient prediction model to predict the accidental lighting coefficient of the specific lighting area, includes: The accidental lighting coefficient refers to the probability coefficient of other dedicated lighting areas needing to turn on the high-brightness mode of the ceiling lights in other dedicated lighting areas due to an accidental event in the already used site; The dedicated lighting area refers to the illumination area corresponding to each ceiling light; Factors affecting the accidental lighting coefficient include: historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, and correlation with emergency exits; By utilizing existing security cameras or dedicated visual sensors in the venue, video analysis can be used to confirm falling events and record the location coordinates of the incidents to obtain the historical falling frequency of other dedicated lighting areas. Calculate all shortest paths from the used site nodes to each key functional node, and count the number of paths that pass through the nodes corresponding to the dedicated lighting area in all shortest paths to obtain the path velocities. Real-time population density is obtained by detecting the outline of heat sources and counting the number of people. By using video analytics, abnormal behavior patterns such as sudden stopping after running at high speed, prolonged collapse of people, and sudden gathering of crowds are detected to determine the frequency of abnormal events. The venue management system provides information on the types of events that can be booked for each venue; Emergency exit relevance refers to the dedicated lighting area traversed by the shortest path from the occupied site to the emergency exit; Obtain historical accidental lighting coefficient data for a single dedicated lighting area in different periods. The historical accidental lighting coefficient data includes the historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, correlation of emergency exits, and historical accidental lighting coefficient of the single dedicated lighting area in different periods. Based on the historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, correlation of emergency exits, and corresponding historical accidental lighting coefficients of the dedicated lighting area in different historical accidental lighting coefficient data, an accidental lighting coefficient prediction set is generated and divided into a training set and a test set. Convolutional neural networks are constructed, with the historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, and correlation of emergency exits from different historical accident lighting coefficient data in the training set as input data, and the corresponding historical accident lighting coefficients in the training set as output data. The convolutional neural network is trained to obtain an initial convolutional neural network. The initial convolutional neural network is validated using a test set. The initial convolutional neural network whose output is less than or equal to the preset test error threshold is used as the unexpected illumination coefficient prediction model. The historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, and correlation of emergency exit for each dedicated lighting area within the monitoring period are input into the accidental lighting coefficient prediction model to obtain the predicted accidental lighting coefficient for each dedicated lighting area. In an embodiment of the present invention, the predicted accidental lighting coefficient of all dedicated lighting areas is obtained by an accidental lighting coefficient prediction model. The predicted accidental lighting coefficient is related to historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, activity type influence, and emergency exit correlation. The frequency of historical falls directly affects the magnitude of the predicted accidental lighting coefficient. The higher the frequency of historical falls, the more likely the area is to experience similar events again in the near future, and the greater the predicted accidental lighting coefficient. Therefore, the frequency of historical falls and the accidental lighting coefficient are positively correlated. The magnitude of the path necessity directly affects the magnitude of the predicted accidental lighting coefficient. The greater the path necessity, the more people are likely to pass through this area, and the greater the predicted accidental lighting coefficient. Therefore, the path necessity and the accidental lighting coefficient are positively correlated. The magnitude of real-time population density directly affects the magnitude of the predicted accidental lighting coefficient. The higher the real-time population density, the higher the probability of sudden group lighting needs (such as collective movement or temporary gathering). At the same time, the probability of minor accidents (such as dropping small objects) being noticed and requiring lighting assistance is also greater, resulting in a higher predicted accidental lighting coefficient. Therefore, real-time population density is positively correlated with the accidental lighting coefficient. The frequency of abnormal events directly affects the magnitude of the predicted accidental lighting coefficient. The higher the frequency of abnormal events, the more unstable or high-risk the area is, and the greater the predicted accidental lighting coefficient will be. Therefore, the frequency of abnormal events is positively correlated with the accidental lighting coefficient. Regarding the impact of activity type, the higher the risk coefficient of the activity itself, the greater the predicted accidental lighting coefficient; The magnitude of the emergency exit correlation directly affects the magnitude of the predicted accidental lighting coefficient. The greater the emergency exit correlation, the more the area must be unconditionally brightly illuminated under any emergency mode trigger, and the greater the predicted accidental lighting coefficient. Therefore, the emergency exit correlation and the accidental lighting coefficient are positively correlated.

[0031] It should be further explained that, in the specific implementation process, the process of determining the ceiling light lighting mode for the dedicated lighting area based on the predicted unexpected lighting coefficient and the preset unexpected lighting activation threshold includes: An appropriate threshold for unexpected lighting coefficients is set based on historical coefficient data, which refers to a dataset of past unexpected lighting coefficients for designated lighting areas. The predicted unexpected lighting coefficient for each designated lighting area is compared with the threshold. If the predicted unexpected lighting coefficient is greater than the threshold, not all ceiling lights in the designated lighting area are turned on; instead, they are prioritized.

[0032] In the formula, , , and These are weighting coefficients, obtained through training based on historical data; Let be the unexpected lighting coefficient for the i-th region; Safety risk levels are categorized as high, medium, and low based on historical accident data, with corresponding values ​​of 1.0, 0.6, and 0.3, respectively. For use of urgency: Determine whether there are people about to enter or currently active in the area; In terms of power consumption: the estimated power consumption required to turn on the lighting in this area; The high-brightness mode of the ceiling lights in the designated lighting areas is turned on sequentially from high to low priority until the power limit is reached.

[0033] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0034] In conclusion, the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart LED lighting energy-saving optimization method based on AI edge computing, characterized in that, Includes the following steps: S1: Real-time collection of lighting data for each sports field within the sports venue, and normalization processing of the lighting data; S2: Generate a lighting status score for each used site based on the processed data; S3: Decision on whether to activate auxiliary lighting program for the used site based on lighting status score and preset lighting status score threshold; S4: Obtain historical accidental lighting coefficient data and build an accidental lighting coefficient prediction model, and use the accidental lighting coefficient prediction model to predict the accidental lighting coefficient of the exclusive lighting area; S5: Determine the ceiling light lighting mode for the exclusive lighting area based on the predicted unexpected lighting coefficient and the preset unexpected lighting activation threshold; The process of acquiring historical accidental lighting coefficient data and constructing an accidental lighting coefficient prediction model, and then using this model to predict the accidental lighting coefficient for a specific lighting area, includes: The accidental lighting coefficient refers to the probability coefficient of other dedicated lighting areas needing to turn on the high-brightness mode of the ceiling lights in other dedicated lighting areas due to an accidental event in the already used site; The dedicated lighting area refers to the illumination area corresponding to each ceiling light; Factors affecting the accidental lighting coefficient include: historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, and correlation with emergency exits; By utilizing existing security cameras or dedicated visual sensors in the venue, video analysis can be used to confirm falling events and record the location coordinates of the incidents to obtain the historical falling frequency of other dedicated lighting areas. Calculate all shortest paths from the used site nodes to each key functional node, and count the number of paths that pass through the nodes corresponding to the dedicated lighting area in all shortest paths to obtain the path velocities. Real-time population density is obtained by detecting the outline of heat sources and counting the number of people. By using video analytics, abnormal behavior patterns such as sudden stopping after running at high speed, prolonged collapse of people, and sudden gathering of crowds are detected to determine the frequency of abnormal events. The venue management system provides information on the types of events that can be booked for each venue; Emergency exit relevance refers to the dedicated lighting area traversed by the shortest path from the occupied site to the emergency exit; Obtain historical accidental lighting coefficient data for a single dedicated lighting area in different periods. The historical accidental lighting coefficient data includes the historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, correlation of emergency exits, and historical accidental lighting coefficient of the single dedicated lighting area in different periods. Based on the historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, correlation of emergency exits, and corresponding historical accidental lighting coefficients of the dedicated lighting area in different historical accidental lighting coefficient data, an accidental lighting coefficient prediction set is generated and divided into a training set and a test set. Convolutional neural networks are constructed, with the historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, and correlation of emergency exits from different historical accident lighting coefficient data in the training set as input data, and the corresponding historical accident lighting coefficients in the training set as output data. The convolutional neural network is trained to obtain an initial convolutional neural network. The initial convolutional neural network is validated using a test set. The initial convolutional neural network whose output is less than or equal to the preset test error threshold is used as the unexpected illumination coefficient prediction model. The historical fall frequency, path necessity, real-time personnel density, frequency of abnormal events, impact of activity type, and correlation of emergency exit for each dedicated lighting area within the monitoring period are input into the accidental lighting coefficient prediction model to obtain the predicted accidental lighting coefficient for each dedicated lighting area. The process of determining the ceiling light lighting mode for a dedicated lighting area based on the predicted unexpected lighting coefficient and the preset unexpected lighting activation threshold includes: An appropriate threshold for unexpected lighting coefficients is set based on historical coefficient data, which refers to a dataset of past unexpected lighting coefficients for designated lighting areas. The predicted unexpected lighting coefficient for each designated lighting area is compared with the threshold. If the predicted unexpected lighting coefficient is greater than the threshold, not all ceiling lights in the designated lighting area are turned on; instead, they are prioritized. In the formula, , , and These are weighting coefficients, obtained through training based on historical data; Let be the unexpected lighting coefficient for the i-th region; Safety risk levels are categorized as high, medium, and low based on historical accident data. For use of urgency: Determine whether there are people about to enter or currently active in the area; In terms of power consumption: the estimated power consumption required to turn on the lighting in this area; The high-brightness mode of the ceiling lights in the designated lighting areas is turned on sequentially from high to low priority until the power limit is reached.

2. The intelligent LED lighting energy-saving optimization method based on AI edge computing according to claim 1, characterized in that, The sports venue has multiple sports fields, and multiple LED canopy lights are evenly installed above the sports venue. Each sports field has a corresponding number of LED canopy lights. These canopy lights are designated as field canopy lights. The areas outside the sports fields are designated as paths within the venue, and canopy lights are also installed above the paths. Sports fields where people are active are designated as used fields, and sports fields where no people are active are designated as unused fields. The canopy lights of used fields are in high brightness mode, and the canopy lights of unused fields are in low brightness mode. The lighting data includes the number of people present in used fields, ambient illuminance, natural light utilization rate, glare index, and visual task complexity. The presence of personnel is determined by detecting and counting targets using infrared pyroelectric sensor arrays deployed at the site boundary or millimeter-wave radar on top, and then the presence of personnel is normalized. Illuminance sensors installed on both sides of the site or on the netting posts are used to measure the horizontal illuminance, and the average value is taken to eliminate the influence of local shadows to obtain the ambient illuminance. The ambient illuminance is then normalized. In the formula, To normalize ambient illuminance, This represents the measured ambient illuminance. This represents the minimum standard illuminance required for this type of sport; The natural light utilization rate U is calculated by comparing the changes in total illuminance when artificial light sources are turned on and off. The glare index is measured using a dedicated glare measurement sensor, and then normalized. In the formula, The normalized glare index, The glare index is the measured value. This is the maximum acceptable glare index threshold for this sport; A ball speed monitoring system based on high-speed cameras or radar measures the ball speed, derives the visual task complexity from the ball speed, and then normalizes the visual task complexity. In the formula, To normalize the complexity of visual tasks, The measured ball speed This represents the typical maximum speed for this sport. and The weighting coefficients are obtained through training based on historical data, and T is the baseline value for the movement type. All sensor data processing and decision-making are completed on the local edge node.

3. The intelligent LED lighting energy-saving optimization method based on AI edge computing according to claim 2, characterized in that, The process of deciding whether to activate auxiliary lighting procedures for an already used site based on the lighting condition score and a preset lighting condition score threshold includes: Set an appropriate lighting status score threshold based on historical lighting status score data. The historical lighting status score data refers to the data set of lighting status scores of previously used sites. Compare the lighting status scores of the used sites with the lighting status score threshold. When the lighting status score is less than the lighting status score threshold, the auxiliary lighting program is activated. Auxiliary lighting procedures: Let the ceiling lights located around the used site be called surrounding ceiling lights. Auxiliary lighting refers to adjusting the surrounding ceiling lights from low brightness mode to high brightness mode. First, calculate the supplementary lighting efficiency value of each surrounding ceiling light for the used site. Sort them in descending order of supplementary lighting efficiency value to determine the initial number of surrounding ceiling lights to be turned on. Prioritize turning on the surrounding ceiling lights with the highest supplementary lighting efficiency value according to the initial number of lights to be turned on. After turning them on, re-evaluate the lighting status score of the used site to obtain a new lighting status score. Based on the new lighting status score, make fine adjustments to the surrounding ceiling lights.

4. The intelligent LED lighting energy-saving optimization method based on AI edge computing according to claim 3, characterized in that, Access the venue's three-dimensional coordinate information, including the position coordinates of each ceiling light. =( , , The boundary coordinates of each sports field are predefined, and the geometric center point of the used fields is calculated. =( , ); Calculate the supplementary lighting efficiency of the surrounding canopy lights on the already used site. : In the formula, The central luminous intensity of the surrounding ceiling lights in rated high brightness mode is obtained from the ceiling light's light distribution curve. The angle between the direction of the ceiling light axis and the vector pointing to the center of the target site is calculated using spatial geometry. m is the beam angle index of the luminaire, which is determined by the optical characteristics of the luminaire; For the j-th surrounding canopy light to the center point of the used site The straight-line distance; The occlusion attenuation factor is calculated by taking into account the occlusion effect of possible obstacles in the venue on the lighting, and is obtained through a pre-established 3D model of the venue and ray tracing simulation. n is the atmospheric attenuation coefficient, which takes into account the impact of air dust on lighting transmission and is dynamically adjusted based on air quality monitoring data. This is a correction term for distance attenuation.

5. The intelligent LED lighting energy-saving optimization method based on AI edge computing according to claim 4, characterized in that, The surrounding ceiling lights were fine-tuned based on the new lighting condition rating: Before opening After the surrounding ceiling lights are installed, wait for Δt time and recalculate the new lighting status score of the used site. If the new lighting condition score is greater than the lighting condition score threshold, the new lighting condition score is subtracted from the lighting condition score threshold to obtain the score margin. A tolerance threshold is set. If the score margin is greater than the tolerance threshold, it indicates that there is over-lighting. The surrounding ceiling lights that have been turned on are arranged in ascending order of supplementary lighting efficiency value. Each surrounding ceiling light is turned off in turn and the lighting condition score is re-evaluated until the latest lighting condition score is still greater than the lighting condition score threshold and the score margin is less than the tolerance threshold. If the new lighting status score is less than the lighting status score threshold, calculate the set of available surrounding ceiling lights that are still in a low brightness state, sort them from high to low according to the supplementary lighting efficiency value, and turn on the remaining surrounding ceiling lights in turn. After turning on one surrounding ceiling light each time, re-evaluate the lighting status score until the latest lighting status score is greater than the lighting status score threshold. If turning on all available surrounding ceiling lights still fails to meet the requirement that the new lighting status score is greater than the lighting status score threshold, an alarm message will be generated, indicating that manual inspection or maintenance may be necessary.

Images