Urban lighting equipment cooperative monitoring method and system fused with deep learning

Abstract

The invention is suitable for the technical field of intelligent traffic, and provides a deep learning-fused urban lighting equipment cooperative monitoring method and system, and the method comprises the steps: collecting road images and videos in real time, and outputting the lane number, space coordinates and road congestion degree of an accident vehicle through a position recognition model; delimiting a detection lane and calculating a light wave propagation frequency; generating a control instruction for an upstream street lamp by taking the space coordinates of the accident vehicle as a starting point; generating a light wave propagation speed decreasing correction control instruction by combining the congestion degree and the distance; and controlling the street lamp to generate light waves propagating backwards according to the correction instruction, and performing dynamic adjustment. Accurate accident positioning and dynamic warning adjustment are realized through fusion of deep learning, the timeliness and effectiveness of traffic early warning are improved, and the method is suitable for urban road traffic accident emergency disposal scenes.

Description

Technical Field

[0001] This invention belongs to the field of intelligent transportation technology, and in particular relates to a collaborative monitoring method and system for urban lighting equipment that integrates deep learning. Background Technology

[0002] With the acceleration of urbanization and the continuous growth of urban road traffic flow, the efficiency of traffic accident emergency response and traffic management has become crucial to ensuring road safety. Urban lighting equipment, as an important component of road infrastructure, has evolved from a single lighting function to a multi-functional collaborative approach of "lighting + monitoring + warning," gradually integrating IoT, big data, and artificial intelligence technologies to achieve intelligent upgrades such as remote control and status monitoring. Among these advancements, utilizing lighting equipment to transmit traffic warning information and provide early warnings to upstream vehicles at accident scenes has become an important research direction for improving traffic emergency response capabilities.

[0003] Existing technologies have the following problems: they cannot accurately identify the specific lane and spatial coordinates of the accident vehicle, resulting in inaccurate delineation of the warning range; the assessment of road congestion lacks quantification and dynamic updates, making it difficult to match actual traffic conditions; the frequency, range, and speed of light wave propagation are fixed and cannot be dynamically adjusted according to the degree of congestion and the location of the accident, making the warning signal easy for drivers to ignore or causing visual interference, resulting in poor warning effects; and they lack a dynamic adjustment mechanism based on real-time road conditions, making it difficult to adapt to complex and ever-changing traffic scenarios. Summary of the Invention

[0004] The purpose of this invention is to provide a collaborative monitoring method for urban lighting equipment that integrates deep learning, aiming to solve the technical problems existing in the prior art as identified in the background art.

[0005] This invention is implemented as follows: a collaborative monitoring method for urban lighting equipment integrating deep learning, the method comprising:

[0006] Real-time acquisition of road images and videos, and establishment of a location recognition model. By extracting and analyzing features from the road images and videos, the system outputs the precise location information of the lane where the accident vehicle is located and the current road congestion level information. The precise location information includes the lane number and spatial coordinates.

[0007] Based on the output lane number and spatial coordinates, the lane where the accident vehicle is located and the adjacent lanes are designated as detection lanes, and the light wave propagation frequency is calculated based on the congestion information.

[0008] Starting from the spatial coordinates of the accident vehicle, control commands are generated for all continuous streetlights in the upstream lane of the accident lane.

[0009] Based on the degree of congestion and the spatial coordinates of the accident vehicle, the distance between the upstream road section and the accident vehicle is calculated. Combined with the control command and the spatial coordinates of the accident vehicle, a corrective control command with decreasing light wave propagation speed is generated.

[0010] Based on the correction control command, all continuous streetlights in the upstream section of the road are controlled to generate backward-propagating light waves. At the same time, the correction control command is dynamically adjusted in combination with real-time collected road images and road videos.

[0011] As a further aspect of the present invention, the output of the precise location information of the lane where the accident vehicle is located and the current road congestion information specifically includes:

[0012] Real-time road images and videos are captured by cameras, and preprocessed and feature extracted to identify accident vehicles.

[0013] A location recognition model is established based on historically collected road images and videos. The preprocessed images are then input into the location recognition model, which outputs the lane number where the accident vehicle is located and the spatial coordinates of the accident vehicle within the lane.

[0014] By analyzing the vehicle density of the upstream section of the road where the accident vehicle is located in continuous video frames using a location recognition model, the current road congestion level can be obtained.

[0015] As a further aspect of the present invention, the calculation of the light wave propagation frequency based on congestion level information specifically includes:

[0016] The lane where the accident vehicle was located and the adjacent lanes are marked as detection lanes based on the lane number of the lane where the accident vehicle was located;

[0017] Based on the vehicle density in the output congestion information, a congestion threshold is set. When the congestion is equal to or higher than the congestion threshold and when it is lower than the congestion threshold, the light wave propagation frequency is calculated respectively.

[0018] Based on vehicle density and street light spacing, the light wave propagation length is calculated, and the endpoint of the light wave propagation is identified.

[0019] As a further embodiment of the present invention, the generation control instructions specifically include:

[0020] Using the spatial coordinates of the accident vehicle within the lane as the starting point of light wave transmission, all streetlights within the range from the starting point to the ending point of light wave transmission are incorporated into the control cluster.

[0021] Based on the light wave propagation frequency, control commands are generated for the streetlights within the control cluster in the form of lighting up and delaying extinguishing.

[0022] As a further aspect of the present invention, the correction control command for the decrease in the propagation speed of generated light waves specifically includes:

[0023] Based on the spatial coordinates of the accident vehicle within the lane, the actual distances between each street light and the accident vehicle in the upstream road segment from the light wave transmission start point to the light wave transmission end point are calculated, and a distance parameter mapping table is established.

[0024] By combining the vehicle density and the distance from the light wave transmission start point to the light wave transmission end point in the congestion information, the decreasing slope is calculated to generate a light wave propagation speed decreasing curve, which gradually decreases from the light wave transmission start point to the light wave propagation end point.

[0025] The light wave control command is fused with the light wave propagation speed decrease curve to generate a correction control command, which includes the specific lighting sequence and duration parameters for each street light node.

[0026] Another object of the present invention is to provide a collaborative monitoring system for urban lighting equipment that integrates deep learning, the system comprising:

[0027] The real-time acquisition module is used to acquire road images and videos in real time and establish a location recognition model. By extracting and analyzing features from the road images and videos, it outputs the precise location information of the lane where the accident vehicle is located and the current road congestion information. The precise location information includes the lane number and spatial coordinates.

[0028] The propagation frequency calculation module is used to delineate the lane where the accident vehicle is located and the adjacent lanes as detection lanes based on the output lane number and spatial coordinates, and to calculate the light wave propagation frequency based on congestion information.

[0029] The control command generation module is used to generate control commands for all continuous streetlights in the upstream lane of the accident lane, starting from the spatial coordinates of the accident vehicle.

[0030] The corrected control command calculation module is used to calculate the distance between the upstream road segment and the accident vehicle based on the degree of congestion and the spatial coordinates of the accident vehicle, and generate a corrected control command with decreasing light wave propagation speed by combining the control command and the spatial coordinates of the accident vehicle.

[0031] The light wave propagation control module is used to control all continuous streetlights in the upstream section of the road to generate backward-propagating light waves according to the correction control command, and at the same time dynamically adjust the correction control command by combining real-time collected road images and road videos.

[0032] As a further embodiment of the present invention, the real-time acquisition module includes:

[0033] The image and video acquisition unit is used to acquire road images and videos in real time through a camera, and to perform preprocessing and feature extraction to identify accident vehicles.

[0034] The location recognition unit is used to build a location recognition model based on historically collected road images and videos, and inputs the pre-processed image into the location recognition model to output the lane number where the accident vehicle is located and the spatial coordinates of the accident vehicle within the lane.

[0035] The congestion level assessment unit is used to analyze the vehicle density of the upstream section of the road where the accident vehicle is located in continuous video frames through a location recognition model, and to obtain the current road congestion level.

[0036] As a further embodiment of the present invention, the propagation frequency calculation module includes:

[0037] The detection lane marking unit is used to mark the lane where the accident vehicle is located and the adjacent lanes as detection lanes based on the lane number of the accident vehicle.

[0038] The light wave frequency calculation unit is used to set a congestion threshold based on the vehicle density in the output congestion level information, and to calculate the light wave propagation frequency when the congestion is equal to or higher than the congestion threshold and when the congestion is lower than the congestion threshold, respectively.

[0039] The endpoint identification unit is used to calculate the light wave propagation length and identify the endpoint of the light wave propagation based on vehicle density and street light spacing.

[0040] As a further embodiment of the present invention, the control command generation module includes:

[0041] The control cluster incorporation unit is used to incorporate all streetlights within the range from the starting point to the ending point of the light wave transmission, using the spatial coordinates of the accident vehicle within the lane as the starting point of the light wave transmission.

[0042] The instruction generation unit is used to generate control instructions for the streetlights in the control cluster in the form of turning on and delaying off, based on the propagation frequency of the light wave.

[0043] As a further embodiment of the present invention, the modified control command calculation module includes:

[0044] The distance parameter mapping table establishment unit is used to calculate the actual distance between each street light in the upstream road segment from the light wave transmission start point to the light wave transmission end point and establish a distance parameter mapping table based on the spatial coordinates of the accident vehicle in the lane.

[0045] The decreasing curve calculation unit is used to combine the vehicle density and the distance from the light wave transmission start point to the light wave transmission end point in the congestion information to calculate the decreasing slope and generate a light wave propagation speed decreasing curve, which gradually decreases from the light wave transmission start point to the light wave propagation end point.

[0046] The correction instruction generation unit is used to fuse the light wave control instruction with the light wave propagation speed decrease curve to generate a correction control instruction, which includes the specific lighting sequence and duration parameters of each street light node.

[0047] The beneficial effects of this invention are:

[0048] This invention significantly improves the accuracy and adaptability of collaborative monitoring of urban lighting equipment by integrating deep learning technology. Based on a location recognition model, it can accurately output the lane number, spatial coordinates, and road congestion level of the accident vehicle, providing a reliable basis for subsequent warning control. By defining detection lanes and dynamically calculating the light wave propagation frequency and transmission length, it ensures that the warning range focuses on the accident-affected area, avoiding resource waste. The generated correction control commands, combined with distance and congestion level, reduce the light wave propagation speed, making the warning signal more consistent with driver perception patterns. Simultaneously, by dynamically adjusting the control commands based on real-time acquired images and videos, it can adapt to different traffic scenarios (such as congestion, smooth traffic, and complex weather). The overall solution achieves accurate, dynamic, and collaborative accident early warning, effectively improving the early warning efficiency for upstream vehicles, reducing the risk of secondary accidents, and balancing warning effectiveness with resource utilization efficiency. Attached Figure Description

[0049] Figure 1 A flowchart illustrating the collaborative monitoring method for urban lighting equipment incorporating deep learning, provided in an embodiment of the present invention;

[0050] Figure 2 A flowchart illustrating the precise location information of the lane where the accident vehicle is located and the current road congestion level, provided in an embodiment of the present invention.

[0051] Figure 3 This is a flowchart for calculating the light wave propagation frequency based on congestion level information, provided in an embodiment of the present invention.

[0052] Figure 4 A flowchart for generating control instructions provided in an embodiment of the present invention;

[0053] Figure 5 A flowchart of a correction control command for reducing the propagation speed of light waves provided in an embodiment of the present invention;

[0054] Figure 6 This is a structural block diagram of a collaborative monitoring system for urban lighting equipment that integrates deep learning, provided in an embodiment of the present invention.

[0055] Figure 7 This is a structural block diagram of the real-time acquisition module provided in an embodiment of the present invention;

[0056] Figure 8This is a structural block diagram of the propagation frequency calculation module provided in an embodiment of the present invention;

[0057] Figure 9 This is a structural block diagram of the control instruction generation module provided in an embodiment of the present invention;

[0058] Figure 10 This is a structural block diagram of the correction control command calculation module provided in an embodiment of the present invention. Detailed Implementation

[0059] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0060] Figure 1 A flowchart of the collaborative monitoring method for urban lighting equipment integrating deep learning provided in an embodiment of the present invention is shown below. Figure 1 As shown, the method includes:

[0061] S100 collects road images and videos in real time and establishes a location recognition model. By extracting and analyzing features from the road images and videos, it outputs the precise location information of the lane where the accident vehicle is located and the current road congestion information. The precise location information includes the lane number and spatial coordinates.

[0062] High-definition cameras deployed along the roadside collect real-time road images and continuous video streams covering the entire road segment. This data contains a wealth of information, including vehicle driving status, lane distribution, and traffic flow. To ensure the accuracy of subsequent analysis, the collected images and videos undergo preprocessing. This includes removing noise caused by environmental interference such as rain, fog, and strong light, enhancing image contrast to highlight key features such as vehicle outlines and lane lines, and then using feature extraction technology to accurately capture vehicle motion (such as whether a collision has occurred or whether the vehicle has deviated from its lane) and environmental features (such as whether hazard lights are on). These features are the core basis for distinguishing between normally driving vehicles and accident vehicles.

[0063] A location recognition model is trained using a large amount of historically accumulated road image and video data (covering different weather conditions, time periods, accident types, and other scenarios). This model, through deep learning algorithms, can automatically learn the typical characteristics of accident vehicles and the spatial distribution patterns of lanes. After inputting preprocessed real-time images into the model, it can quickly output the lane number where the accident vehicle is located, clearly identifying the specific affected lane and its spatial coordinates within the lane, thus achieving precise location of the accident point.

[0064] Meanwhile, the location recognition model also analyzes continuous video frames and calculates the vehicle density of the upstream section of the road where the accident vehicle is located, that is, the number of vehicles per unit area. Combined with preset thresholds, it classifies the congestion level (Level 1: Smooth Traffic, Level 2: Slow Traffic, Level 3: Congested) to fully grasp the current traffic flow status of the road.

[0065] like Figure 2 As shown, the precise location information of the lane where the accident vehicle is located and the current road congestion information specifically include:

[0066] S110 uses cameras to collect real-time road images and videos, performs preprocessing and feature extraction, and identifies accident vehicles, specifically:

[0067] Extract vehicle motion state characteristics (vehicle collision, vehicle lane departure, etc.) and environmental characteristics (hazard lights on);

[0068] The extracted features are input into a pre-trained accident vehicle classification model (training samples include scenarios such as collisions and breakdowns), and the output is the confidence level that the vehicle is an accident vehicle.

[0069] For candidate targets with a confidence level ≥ 0.85, their motion state is analyzed through three consecutive video frames to confirm whether they are accident vehicles.

[0070] S120 establishes a location recognition model based on historically collected road images and videos, and inputs the pre-processed images into the location recognition model to output the lane number where the accident vehicle is located and the spatial coordinates of the accident vehicle within the lane.

[0071] S130 uses a location recognition model to analyze the vehicle density of the upstream section of the road where the accident vehicle is located in continuous video frames to obtain the current road congestion level:

[0072] ;

[0073] in, The congestion levels are categorized into three levels: Level 1 is smooth traffic, Level 2 is slow traffic, and Level 3 is congested. The vehicle density of the upstream section of the road is (unit: vehicles / 100m², calculated by converting the vehicle pixel area in consecutive video frames with the actual area of ​​the road section). This is the first-level threshold; This is a secondary threshold.

[0074] S200, based on the output lane number and spatial coordinates, delineates the lane where the accident vehicle is located and the adjacent lane as detection lanes, and calculates the light wave propagation frequency based on congestion information;

[0075] Based on the lane number of the vehicle involved in the accident, the lane where the vehicle was located and the adjacent lanes are marked as detection lanes. This focuses on the traffic areas that the accident may directly affect. The accident lane itself is the point where the accident occurred, and vehicles cannot pass normally. Vehicles in the adjacent lanes may slow down to avoid or observe the accident, making them areas with a high risk of secondary accidents. Including these two types of lanes in the detection scope can avoid ineffective control of streetlights in unrelated lanes, saving energy consumption and control resources of lighting equipment, and ensuring that warning information is concentrated on the areas that need it most, achieving accurate coverage.

[0076] Based on this, the light wave propagation frequency is calculated using the output congestion level information. The light wave propagation frequency directly affects the driver's perception of the warning information, and the driver's perception needs and attention distribution differ significantly under different congestion conditions. Therefore, a congestion threshold is set. When the congestion level is higher than or equal to the threshold (i.e., high vehicle density and slow traffic), a lower light wave propagation frequency is calculated. This is because at this time, the distance between vehicles is small, and the driver's attention is more focused on the vehicle in front. High-frequency flashing can easily cause visual fatigue, while low-frequency flashing can maintain a continuous warning presence and avoid strong light stimulation that interferes with driving. When the congestion level is lower than the threshold (i.e., low vehicle density and smooth traffic), a higher light wave propagation frequency is calculated. This is because at this time, the vehicle speed is faster, the driver's field of vision is relatively wide, but attention is easily distracted. High-frequency flashing can quickly break through visual blind spots and ensure that the warning signal is detected in time.

[0077] At the same time, this step also combines vehicle density and street light spacing to calculate the light wave transmission length and identify the transmission endpoint. The higher the vehicle density, the denser the upstream vehicles are, and a longer transmission length is needed to warn more vehicles in advance. The street light spacing determines the actual physical coverage area. The combination of the two can ensure that the light wave transmission range is sufficient for early warning without being too long and wasting resources.

[0078] The precise lane numbers output by the location recognition model ensure that the lane delineation is accurate. For example, in complex road sections with multiple lane intersections or blurred lane lines, the model can still accurately distinguish lane boundaries and avoid including irrelevant lanes in the detection range. The congestion level derived from deep learning analysis provides a reliable basis for the dynamic calculation of light wave frequency. Even in environments that affect visual judgment, such as rain, fog, or nighttime, the model can still accurately assess vehicle density through feature extraction from continuous video frames, ensuring that the frequency calculation matches the actual traffic conditions.

[0079] like Figure 3 As shown, the calculation of light wave propagation frequency based on congestion level information specifically includes:

[0080] S210, based on the lane number of the accident vehicle, marks the lane where the accident vehicle is located and the adjacent lanes as detection lanes;

[0081] S220, based on the vehicle density in the output congestion information, a congestion threshold is set. When the congestion is equal to or higher than the congestion threshold and when it is lower than the congestion threshold, the light wave propagation frequency is calculated respectively:

[0082] ;

[0083] in, The frequency of light wave propagation (unit: Hz, i.e., the number of flashes per second); Minimum frequency; For the maximum frequency, This is an adjustment factor used to ensure that frequency changes can be quickly perceived by the driver. The congestion threshold;

[0084] When traffic is congested, vehicles move slowly and drivers focus their attention on the vehicle in front. Low-frequency flashing can avoid visual fatigue caused by strong light and high-frequency stimulation, while maintaining a warning effect through a continuous light source.

[0085] When traffic is smooth, vehicle speed is high (usually >40km / h), and drivers have a wide field of vision but are distracted. High-frequency flashing can quickly break through blind spots and ensure that the flashing signal is detected.

[0086] S230 calculates the light wave propagation length and identifies the light wave propagation endpoint based on vehicle density and street light spacing:

[0087] ;

[0088] in, The length of light wave propagation. This refers to the spacing between streetlights. To maximize the number of streetlights that can be transmitted, The growth coefficient, This is the baseline density.

[0089] S300, starting from the spatial coordinates of the accident vehicle, generates control commands for all continuous streetlights in the upstream lane of the accident lane;

[0090] The starting point for light wave propagation is determined by the spatial coordinates of the accident vehicle. The accuracy of this starting point is ensured by the model's extraction and analysis of deep features from road images and videos, guaranteeing a high degree of consistency with the actual accident location. Based on this, all continuous streetlights in the upstream detection lane (i.e., the accident lane and adjacent lanes) within the range from the starting point to the end point of light wave propagation are included in the control cluster. This range is defined by considering both the directly affected area of ​​the accident and the propagation length calculated based on vehicle density and streetlight spacing. The aim is to ensure that the initial command covers all lighting equipment requiring warning coverage, avoiding omissions of critical nodes.

[0091] When generating control commands, the action rules for the streetlights in the control cluster are formulated in the form of "lighting up and then turning off after a delay" based on the light wave propagation frequency. Each streetlight is lit up at a specific time and then turns off after a delay, following the order from the accident point upstream. This orderly lighting and turning-off rhythm initially simulates the visual effect of light waves propagating backward.

[0092] First, establish the basic signal transmission logic to ensure that upstream vehicles can sense the anomaly ahead through the dynamic changes of streetlights, while the specific details of the propagation speed will be optimized in subsequent steps.

[0093] When an accident occurs on a secondary arterial road in the city, the streetlights within 500 meters upstream are included in the control cluster based on the detection lane range and the light wave frequency under the current congestion. Initial instructions are generated to light up sequentially every 2 seconds, remain lit for 1 second, and then turn off. These instructions clearly define the list of streetlights participating in the control and the basic lighting and turning-off sequence. However, the differences in propagation speed of streetlights at different distances have not yet been considered, leaving room for adjustment in subsequent speed reduction corrections based on distance and congestion level.

[0094] like Figure 4 As shown, the generation control instructions specifically include:

[0095] S310 uses the spatial coordinates of the accident vehicle within the lane as the starting point of the light wave transmission and incorporates all streetlights from the starting point to the ending point of the light wave transmission into the control cluster.

[0096] S320, in conjunction with the light wave propagation frequency, generates control commands for the streetlights in the control cluster in the form of lighting up and delaying extinguishing.

[0097] S400 calculates the distance between the upstream road segment and the accident vehicle based on the degree of congestion and the spatial coordinates of the accident vehicle, and generates a corrective control command that reduces the speed of light propagation by combining the control command and the spatial coordinates of the accident vehicle.

[0098] Based on the spatial coordinates of the accident vehicle, the actual distances between each street light and the accident vehicle in the upstream section of the road from the light wave transmission starting point (accident point) to the transmission endpoint are accurately calculated, and a distance parameter mapping table is established. This mapping table provides a spatial reference for subsequent speed adjustments, ensuring that the control parameters of each street light can be accurately correlated with its distance from the accident point, avoiding signal misalignment caused by distance estimation errors.

[0099] Based on this, by combining congestion information (vehicle density) and light wave propagation length, the speed reduction slope is calculated, thus generating a light wave propagation speed reduction curve. Higher vehicle density indicates denser upstream traffic, requiring faster speed decay to ensure nearby vehicles have sufficient reaction time. Longer propagation lengths result in gentler speed decay, preventing distant vehicles from missing warnings due to rapid signal weakening. This curve shows a gradual decrease from the starting point to the end point. Its design logic aligns with drivers' perceptual habits: at long distances, the light wave propagation speed is slightly faster, quickly transmitting warning signals to distant vehicles for early warning; at close distances, the speed decreases, strengthening the warning through a longer-lasting light signal, reminding vehicles that they are approaching an accident point and need to drive cautiously.

[0100] Finally, the control commands are integrated with this decreasing curve to generate corrective control commands, which include the specific lighting sequence (when to light up) and duration (how long the light stays on) for each street light node. This integration is not a simple superposition, but rather ensures that the action rhythm of each street light precisely matches the speed requirements of its location. For example, street lights closer to the accident site are lit for a longer duration, while street lights further away switch at a slightly faster pace, forming a coherent and hierarchical flow of warning signals.

[0101] like Figure 5 As shown, the correction control command for the decrease in the propagation speed of generated light waves specifically includes:

[0102] S410, based on the spatial coordinates of the accident vehicle in the lane, calculate the actual distance between each street light in the upstream road section from the light wave transmission start point to the light wave transmission end point and establish a distance parameter mapping table;

[0103] S420, combining the vehicle density and the distance from the light wave transmission start point to the light wave transmission end point in the congestion information, calculate the decreasing slope and generate a light wave propagation speed decreasing curve, wherein the light wave propagation speed decreasing curve gradually decreases from the light wave transmission start point to the light wave propagation end point;

[0104] Specifically, the decreasing slope is calculated to generate a curve showing a decrease in the speed of light propagation.

[0105] ;

[0106] ;

[0107] in, The velocity decrease slope is the rate of decrease in velocity per meter of distance.

[0108] The baseline slope coefficient; Distance from the accident point The speed of light propagation at that location; This is the initial speed (corresponding to the speed limit on city roads, used to allow for reaction time). The distance between the upstream streetlight and the accident site;

[0109] slope Positively correlated with density and negatively correlated with transmission length (higher density requires faster deceleration, while longer distances result in smoother deceleration), speed curve This ensures that upstream vehicles gradually perceive deceleration signals from a distance to a closer distance, consistent with the behavioral logic of long-distance warnings and close-range strong alerts in actual driving.

[0110] S430, the light wave control command is fused with the light wave propagation speed decrease curve to generate a correction control command, which includes the specific lighting sequence and duration parameters of each street light node.

[0111] The S500 system, based on corrective control commands, controls all continuous streetlights in the upstream section of the road to generate backward-propagating light waves, while simultaneously dynamically adjusting the corrective control commands in conjunction with real-time collected road images and videos.

[0112] Figure 6 The structural block diagram of the urban lighting equipment collaborative monitoring system integrating deep learning provided in the embodiments of the present invention is as follows: Figure 6 As shown, the system includes:

[0113] The real-time acquisition module 100 is used to acquire road images and videos in real time and establish a location recognition model. By extracting and analyzing features from the road images and videos, it outputs the precise location information of the lane where the accident vehicle is located and the current road congestion information. The precise location information includes the lane number and spatial coordinates.

[0114] The propagation frequency calculation module 200 is used to delineate the lane where the accident vehicle is located and the adjacent lanes as detection lanes based on the output lane number and spatial coordinates, and to calculate the light wave propagation frequency based on congestion information.

[0115] The control command generation module 300 is used to generate control commands based on the spatial coordinates of the accident vehicle and all continuous streetlights in the upstream section of the accident lane.

[0116] The corrected control command calculation module 400 is used to calculate the distance between the upstream road segment and the accident vehicle based on the degree of congestion and the spatial coordinates of the accident vehicle, and generate a corrected control command with decreasing light wave propagation speed by combining the control command and the spatial coordinates of the accident vehicle.

[0117] The light wave propagation control module 500 is used to control all continuous streetlights in the upstream section of the road to generate light waves that propagate backward according to the correction control command, and at the same time dynamically adjust the correction control command by combining real-time collected road images and road videos.

[0118] like Figure 7 As shown, the real-time acquisition module 100 includes:

[0119] The image and video acquisition unit 110 is used to acquire road images and videos in real time through a camera, and to perform preprocessing and feature extraction to identify accident vehicles.

[0120] The location recognition unit 120 is used to build a location recognition model based on historically collected road images and road videos, and input the preprocessed image into the location recognition model to output the lane number where the accident vehicle is located and the spatial coordinates of the accident vehicle in the lane.

[0121] The congestion level assessment unit 130 is used to analyze the vehicle density of the upstream section of the road where the accident vehicle is located in continuous video frames through a location recognition model to obtain the current road congestion level.

[0122] like Figure 8 As shown, the propagation frequency calculation module 200 includes:

[0123] The lane marking unit 210 is used to mark the lane where the accident vehicle is located and the adjacent lanes as detection lanes based on the lane number of the accident vehicle.

[0124] The light wave frequency calculation unit 220 is used to set a congestion threshold based on the vehicle density in the output congestion level information, and to calculate the light wave propagation frequency when the congestion is equal to or higher than the congestion threshold and when the congestion is lower than the congestion threshold, respectively.

[0125] The endpoint identification unit 230 is used to calculate the light wave transmission length and identify the endpoint of the light wave transmission based on vehicle density and street light spacing.

[0126] like Figure 9 As shown, the control command generation module 300 includes:

[0127] The control cluster incorporation unit 310 is used to incorporate all streetlights from the starting point to the ending point of the light wave transmission into the control cluster, using the spatial coordinates of the accident vehicle within the lane as the starting point of the light wave transmission.

[0128] The instruction generation unit 320 is used to generate control instructions for the streetlights in the control cluster in the form of turning on and delaying the turning off, based on the propagation frequency of the light wave.

[0129] like Figure 10As shown, the correction control command calculation module 400 includes:

[0130] The distance parameter mapping table establishment unit 410 is used to calculate the actual distance between each street light in the upstream road section from the light wave transmission start point to the light wave transmission end point and establish a distance parameter mapping table based on the spatial coordinates of the accident vehicle in the lane.

[0131] The decreasing curve calculation unit 420 is used to combine the vehicle density and the distance from the light wave transmission start point to the light wave transmission end point in the congestion information to calculate the decreasing slope and generate a light wave propagation speed decreasing curve, wherein the light wave propagation speed decreasing curve gradually decreases from the light wave transmission start point to the light wave transmission end point.

[0132] The correction instruction generation unit 430 is used to fuse the light wave control instruction with the light wave propagation speed decrease curve to generate a correction control instruction, which includes the specific lighting sequence and duration parameters of each street light node.

[0133] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0134] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

[0135] The above description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A collaborative monitoring method for urban lighting equipment integrating deep learning, characterized in that, The method includes: Real-time acquisition of road images and videos, and establishment of a location recognition model. By extracting and analyzing features from the road images and videos, the system outputs the precise location information of the lane where the accident vehicle is located and the current road congestion level information. The precise location information includes the lane number and spatial coordinates. Based on the output lane number and spatial coordinates, the lane where the accident vehicle is located and the adjacent lanes are designated as detection lanes, and the light wave propagation frequency is calculated based on the congestion information. Starting from the spatial coordinates of the accident vehicle, control commands are generated for all continuous streetlights in the upstream lane of the accident lane. Based on the degree of congestion and the spatial coordinates of the accident vehicle, the distance between the upstream road section and the accident vehicle is calculated. Combined with the control command and the spatial coordinates of the accident vehicle, a corrective control command with decreasing light wave propagation speed is generated. Based on the correction control command, all continuous streetlights in the upstream section of the road are controlled to generate backward-propagating light waves. At the same time, the correction control command is dynamically adjusted in combination with real-time collected road images and road videos.

2. The method according to claim 1, characterized in that, The precise location information of the lane where the accident vehicle was located and the current road congestion information are output, specifically including: Real-time road images and videos are captured by cameras, and preprocessed and feature extracted to identify accident vehicles. A location recognition model is established based on historically collected road images and videos. The preprocessed images are then input into the location recognition model, which outputs the lane number where the accident vehicle is located and the spatial coordinates of the accident vehicle within the lane. By analyzing the vehicle density of the upstream section of the road where the accident vehicle is located in continuous video frames using a location recognition model, the current road congestion level can be obtained.

3. The method according to claim 2, characterized in that, The calculation of light wave propagation frequency based on congestion level information specifically includes: The lane where the accident vehicle was located and the adjacent lanes are marked as detection lanes based on the lane number of the lane where the accident vehicle was located; Based on the vehicle density in the output congestion information, a congestion threshold is set. When the congestion is equal to or higher than the congestion threshold and when it is lower than the congestion threshold, the light wave propagation frequency is calculated respectively. Based on vehicle density and street light spacing, the light wave propagation length is calculated, and the endpoint of the light wave propagation is identified.

4. The method according to claim 3, characterized in that, The generation control instructions specifically include: Using the spatial coordinates of the accident vehicle within the lane as the starting point of light wave transmission, all streetlights within the range from the starting point to the ending point of light wave transmission are incorporated into the control cluster. Based on the light wave propagation frequency, control commands are generated for the streetlights within the control cluster in the form of lighting up and delaying extinguishing.

5. The method according to claim 4, characterized in that, The correction control command for the decrease in the propagation speed of generated light waves specifically includes: Based on the spatial coordinates of the accident vehicle within the lane, the actual distances between each street light and the accident vehicle in the upstream road segment from the light wave transmission start point to the light wave transmission end point are calculated, and a distance parameter mapping table is established. By combining the vehicle density and the distance from the light wave transmission start point to the light wave transmission end point in the congestion information, the decreasing slope is calculated to generate a light wave propagation speed decreasing curve, which gradually decreases from the light wave transmission start point to the light wave propagation end point. The light wave control command is fused with the light wave propagation speed decrease curve to generate a correction control command, which includes the specific lighting sequence and duration parameters for each street light node.

6. A collaborative monitoring system for urban lighting equipment integrating deep learning, characterized in that, The system includes: The real-time acquisition module is used to acquire road images and videos in real time and establish a location recognition model. By extracting and analyzing features from the road images and videos, it outputs the precise location information of the lane where the accident vehicle is located and the current road congestion information. The precise location information includes the lane number and spatial coordinates. The propagation frequency calculation module is used to delineate the lane where the accident vehicle is located and the adjacent lanes as detection lanes based on the output lane number and spatial coordinates, and to calculate the light wave propagation frequency based on congestion information. The control command generation module is used to generate control commands for all continuous streetlights in the upstream lane of the accident lane, starting from the spatial coordinates of the accident vehicle. The corrected control command calculation module is used to calculate the distance between the upstream road segment and the accident vehicle based on the degree of congestion and the spatial coordinates of the accident vehicle, and generate a corrected control command with decreasing light wave propagation speed by combining the control command and the spatial coordinates of the accident vehicle. The light wave propagation control module is used to control all continuous streetlights in the upstream section of the road to generate backward-propagating light waves according to the correction control command, and at the same time dynamically adjust the correction control command by combining real-time collected road images and road videos.

7. The system according to claim 6, characterized in that, The real-time acquisition module includes: The image and video acquisition unit is used to acquire road images and videos in real time through a camera, and to perform preprocessing and feature extraction to identify accident vehicles. The location recognition unit is used to build a location recognition model based on historically collected road images and videos, and inputs the pre-processed image into the location recognition model to output the lane number where the accident vehicle is located and the spatial coordinates of the accident vehicle within the lane. The congestion level assessment unit is used to analyze the vehicle density of the upstream section of the road where the accident vehicle is located in continuous video frames through a location recognition model, and to obtain the current road congestion level.

8. The system according to claim 7, characterized in that, The propagation frequency calculation module includes: The detection lane marking unit is used to mark the lane where the accident vehicle is located and the adjacent lanes as detection lanes based on the lane number of the accident vehicle. The light wave frequency calculation unit is used to set a congestion threshold based on the vehicle density in the output congestion level information, and to calculate the light wave propagation frequency when the congestion is equal to or higher than the congestion threshold and when the congestion is lower than the congestion threshold, respectively. The endpoint identification unit is used to calculate the light wave propagation length and identify the endpoint of the light wave propagation based on vehicle density and street light spacing.

9. The system according to claim 8, characterized in that, The control command generation module includes: The control cluster incorporation unit is used to incorporate all streetlights within the range from the starting point to the ending point of the light wave transmission, using the spatial coordinates of the accident vehicle within the lane as the starting point of the light wave transmission. The instruction generation unit is used to generate control instructions for the streetlights in the control cluster in the form of turning on and delaying off, based on the propagation frequency of the light wave.

10. The system according to claim 9, characterized in that, The correction control command calculation module includes: The distance parameter mapping table establishment unit is used to calculate the actual distance between each street light in the upstream road segment from the light wave transmission start point to the light wave transmission end point and establish a distance parameter mapping table based on the spatial coordinates of the accident vehicle in the lane. The decreasing curve calculation unit is used to combine the vehicle density and the distance from the light wave transmission start point to the light wave transmission end point in the congestion information to calculate the decreasing slope and generate a light wave propagation speed decreasing curve, which gradually decreases from the light wave transmission start point to the light wave propagation end point. The correction instruction generation unit is used to fuse the light wave control instruction with the light wave propagation speed decrease curve to generate a correction control instruction, which includes the specific lighting sequence and duration parameters of each street light node.

Images