A Collision Event Collision Response Method and System Based on Smart Streetlights and Guardrails

Abstract

This application relates to a collaborative response method and system for collision events based on smart streetlights and railings. The method includes acquiring local collision sensing signals through a dynamic sensing module installed on the streetlight, receiving event sensing data packets sent by other streetlights, fusing the event sensing data packets and local collision sensing signals to generate a decision data packet, and broadcasting it to surrounding streetlights. The system includes: a dynamic sensing module, a communication module, a collaborative decision-making module, and a lighting control module. This technical solution can realize a local self-triggering mechanism, improve response timeliness and system independence, and complete local sensing and self-triggering decision-making instantly upon collision, shortening response delay. It has multi-device collaborative processing capabilities. Once a device confirms a collision event, it can broadcast the judgment result in the form of a "decision data packet" through a chain network structure to form a linked response, realizing the technical path of "local self-triggering + distributed collaborative judgment".

Description

Technical Field

[0001] This application relates to the field of smart streetlights, and more particularly to a collaborative response method and system for collision events based on smart streetlights and railings. Background Technology

[0002] Currently, intelligent streetlights and guardrails are widely used in vehicle collision monitoring and accident response systems for road traffic safety. Existing technologies typically collect collision-related data by installing sensing devices such as deformation sensors, vibration sensors, sound sensors, or image acquisition modules on streetlights or guardrails, and then transmit the data to a central control system for analysis and processing via wired or wireless communication networks. In addition, some systems incorporate manual triggering devices at the vehicle or road infrastructure level. For example, physical trigger points are installed on vehicles, allowing drivers to actively activate warning functions of nearby streetlights or road sections in the event of a collision or emergency; alternatively, emergency buttons are directly installed on streetlight poles, allowing pedestrians or drivers to manually trigger warnings upon discovering an accident. These systems mostly rely on a central control platform for unified judgment and response, or on manual intervention for event reporting and triggering.

[0003] However, the aforementioned technologies still have significant limitations. First, the centralized judgment mode relying on a central control system suffers from response delays. When network communication is disrupted or the central node's processing load is too high, the timeliness of early warning triggering and response cannot be guaranteed. Second, manual triggering relies on human operation. If the driver is injured, unconscious, or the environment does not allow for human access to the triggering device, the accident may not be reported in a timely manner. Third, although some existing streetlights or guardrails have sensing capabilities, their detection results are often limited to a single point, lacking the ability to fuse multi-source information and make joint judgments, which can easily lead to misjudgments or missed judgments, especially in edge events such as low-speed collisions, small-amplitude deformations, or environmental noise interference, where accuracy is insufficient. More importantly, existing technologies generally adopt a single decision center, with most streetlight nodes only serving as passive information collection or signal transmission tools. This cannot achieve a distributed intelligent decision-making mode where "any node can independently judge and directly drive surrounding nodes to respond collaboratively," resulting in poor system resilience and insufficient localized rapid response capabilities to sudden collision events. Summary of the Invention

[0004] The purpose of this application is to address the problems in existing technologies for collision event detection and response, such as reliance on a centralized decision-making center, passive node functions, poor response timeliness, and lack of multi-source information fusion for judgment.

[0005] According to one aspect of this application, a collaborative response method for collision events based on smart streetlights and guardrails is provided, comprising the following steps:

[0006] S100. Collect local collision sensing signals through the dynamic sensing module installed on this street light;

[0007] S200: Receive event-aware data packets sent by other streetlights, wherein the event-aware data packets include adjacent collision sensing signals sensed by other streetlights and address information of the other streetlights;

[0008] S300, The event-aware data packet and the local collision-aware signal are fused together. Based on the processing result, it is determined whether to generate a decision data packet and whether to send it via broadcast. Specifically, this includes:

[0009] S310. Extract the adjacent collision sensing signal and its corresponding source street light address information contained in the event sensing data packet;

[0010] S320. The adjacent collision sensing signal and the local collision sensing signal are time-aligned and fused according to the preset timestamp, sampling interval and address priority to obtain the multi-source sensing fusion result of the current road segment.

[0011] S330. This street light determines whether the multi-source sensing fusion result meets the preset collision determination condition. If so, a decision data packet is generated. The decision data packet includes at least: the multi-source sensing fusion result, the address information and timestamp of each source street light in all event sensing data packets participating in the fusion.

[0012] S340. This street light broadcasts the decision data packet to the surrounding street lights.

[0013] Preferably, the local collision sensing signal and the adjacent collision sensing signal include:

[0014] The information includes railing deformation, vibration amplitude, and noise signal amplitude.

[0015] The railing deformation refers to the degree of physical deformation of the railing.

[0016] The vibration amplitude information refers to the vibration amplitude of the street lamp body;

[0017] The noise signal amplitude is the monitored instantaneous noise in decibels;

[0018] Accordingly, in S330, determining that the multi-source sensing fusion result satisfies the preset collision determination condition includes:

[0019] S331, The deformation of the railing exceeds the preset railing deformation threshold; and / or,

[0020] A preset number of adjacent streetlights all detected vibration amplitude values ​​exceeding a preset vibration amplitude threshold within a preset time window; and / or,

[0021] A preset number of adjacent streetlights were found to have noise signal amplitude values ​​exceeding a preset noise signal amplitude threshold within a preset time window.

[0022] Preferably, S330 further includes: based on the determination that the preset collision determination conditions are met, further locating the collision center point based on the following information sources: the coordinates of the monitoring point with the largest railing deformation, the location of the street lamp node with the largest vibration amplitude, and the sound source directional positioning result;

[0023] The above information is fused and calculated to output the spatial location information of the collision center point, and this location information is attached to the decision data packet and broadcast.

[0024] Preferably, after step S331, the method further includes: based on determining that a preset collision determination condition is met, judging the severity of the collision, and classifying the collision event into different severity levels according to a preset severity level determination rule, wherein the levels include:

[0025] Minor collision event: The maximum deformation of the railing is less than the first deformation threshold, the vibration acceleration value is less than the first acceleration threshold and the vibration duration is shorter than the first time threshold, and the audio energy value is lower than the first sound intensity threshold;

[0026] Medium-level collision event: The maximum deformation of the railing is between the first deformation threshold and the second deformation threshold, the vibration acceleration value is between the first acceleration threshold and the second acceleration threshold, or the vibration duration is between the first time threshold and the second time threshold, and the audio energy value is between the first sound intensity threshold and the second sound intensity threshold.

[0027] Severe collision event: The maximum deformation of the railing exceeds the second deformation threshold, or the vibration acceleration value is higher than the second acceleration threshold, or the vibration duration exceeds the second time threshold, or the audio energy value is higher than the second sound intensity threshold.

[0028] Preferably, based on determining the severity and center location of the collision event, streetlights within different ranges are controlled to enter corresponding lighting warning response modes, the response modes including:

[0029] When a minor collision is determined, the streetlights within a radius of no more than 50 meters around the center of the collision will switch to yellow slow-flash warning lights.

[0030] When a collision is determined to be a moderate collision event, streetlights within a radius of 50 to 100 meters around the collision center will switch to alternating red and yellow flashing lights.

[0031] When a serious collision is determined, streetlights within a radius of no less than 100 meters around the collision center point will switch to a red flashing or rotating light mode.

[0032] Preferably, it further includes:

[0033] S400. When a street light node receives the decision data packet and identifies itself as being within the warning coverage area, it activates a secondary accident prevention mechanism, which includes the following steps:

[0034] S410. Collect oncoming vehicle speed information for the current road segment through a speed detection module installed on the target street light node;

[0035] S420. Determine whether the speed of the approaching vehicle exceeds a preset high-speed approach threshold. If so, perform the following operations:

[0036] S421. Determine the target street light node on the path of the oncoming vehicle based on the direction of the oncoming vehicle and the location of the collision center point.

[0037] S422. Broadcast deceleration guidance control commands to multiple street light nodes along the path of the oncoming vehicle, gradually increasing the intensity and frequency of the warning lights to form a progressive "intelligent deceleration guidance" chain.

[0038] S430, the intelligent deceleration guidance process continues until any of the following conditions are met: the oncoming vehicle speed is lower than the set safety threshold and remains stable, or a collision event response end signal is received.

[0039] Preferably, step S330 further includes: the multi-source sensing fusion result does not meet the preset collision determination conditions, but there is a suspicion of collision, specifically including:

[0040] S332, The railing deforms, but the deformation does not exceed the preset railing deformation threshold; and / or,

[0041] If fewer than a preset number of adjacent streetlights are detected to have vibration amplitude values ​​exceeding a preset vibration amplitude threshold within a preset time window; and / or,

[0042] If fewer than a preset number of adjacent streetlights are detected to have noise signal amplitude values ​​exceeding a preset noise signal amplitude threshold within a preset time window;

[0043] S333. Generate a collision verification request packet, wherein the collision verification request packet includes: the multi-source perception fusion result, the address information and timestamp marker of each source street light in all event perception data packets participating in the fusion;

[0044] Accordingly, S340 further includes:

[0045] S341. Send the collision verification request packet to multiple target streetlights, wherein the multiple target streetlights are located at the entrance and exit of the road segment where each source streetlight is located, and are equipped with an image acquisition module;

[0046] S342. The target street light retrieves video images of the road segment where each source street light is located based on the source street light address information and timestamp in the collision verification request packet.

[0047] S343. Analyze whether there is any abnormal behavior in the vehicle in the video image. The abnormal behavior includes: within a preset time window, the vehicle enters the road segment where each source street light is located, but does not leave.

[0048] S344. If the abnormal behavior exists, the decision data packet is generated and sent via broadcast.

[0049] Preferably, S100 further includes:

[0050] S110. Based on a preset threshold, threshold detection and analysis are performed on the local raw sensing signal. If a preset activation condition is met, the communication module is switched from normal communication state to active communication state. The preset threshold is a numerical threshold set for the feature values ​​of the local raw sensing signal, and the feature values ​​include at least one of amplitude, rate of change, and duration. The activation condition is that within a preset time interval, any or a combination of features of the local raw sensing signal reaches the preset threshold. The local raw sensing signal is coarse-grained sampling data of railing deformation, vibration amplitude information, and noise signal amplitude obtained by low-power sampling under normal communication state.

[0051] When the communication module is in active communication state, S200 is executed.

[0052] The system also provides a collaborative response system for collision events based on smart streetlights and guardrails, applying the collaborative response method for collision events based on smart streetlights and guardrails as described above. The system includes:

[0053] The dynamic sensing module communicates with local streetlights to collect local collision sensing signals;

[0054] The communication module receives adjacency event sensing data packets sent by adjacent streetlights;

[0055] The collaborative decision-making module determines whether to send a collision event based on the local collision perception signal and the adjacent event perception data packet.

[0056] The lighting control module controls the illumination mode of the local streetlights according to the instructions sent by the collaborative decision-making module.

[0057] Preferably, the dynamic sensing module includes:

[0058] A deformation sensor, which is connected in communication with the local street light, is installed on the railing to collect the deformation data of the railing;

[0059] A vibration sensor is installed on the local street light to collect vibration amplitude data of the local street light pole.

[0060] A sound sensor collects decibel data of ambient sound around the local streetlights;

[0061] Some of the local streetlights are also equipped with an image acquisition module to collect image information about the area around the local streetlight.

[0062] This application offers the following advantages: Distributed intelligent decision-making: Any streetlight node can make independent decisions after receiving necessary multi-source sensing information, avoiding the communication bottlenecks and failure risks of a single central node. Multi-source sensing fusion: Combining local detection with neighboring node detection results improves the accuracy and robustness of collision event determination, reducing false alarms and missed alarms. Self-triggered instant response: Without relying on external manual triggering or remote scheduling, front-end nodes can complete judgment and broadcasting in milliseconds, significantly shortening response time and improving the timeliness of accident handling. Highly collaborative decision-making: Data packets can drive surrounding nodes to execute functions such as lighting, prompts, or speed guidance as needed, forming a regionalized and hierarchical intelligent early warning network, improving the level of road traffic safety. Attached Figure Description

[0063] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0064] Figure 1 This is a logic block diagram of a collision event collaborative response method based on smart streetlights and railings according to an embodiment of this application;

[0065] Figure 2 This is a logic block diagram of the secondary accident prevention mechanism described in one embodiment of this application;

[0066] Figure 3 This is a schematic diagram of a collision event collaborative response system based on smart streetlights and railings according to an embodiment of this application.

[0067] Reference numerals: 100, Smart street light; 10, Image acquisition module; 20, Vibration sensor; 30, Sound sensor; 40, Millimeter-wave radar module; 50, Sound and light warning module; 60, Solar power supply panel; 70, Deformation sensor. Detailed Implementation

[0068] To facilitate understanding of this application, a more complete description will be provided below with reference to the accompanying drawings. Preferred embodiments of this application are shown in the drawings. However, this application can be implemented in many different forms and is not limited to the embodiments described herein. Rather, these embodiments are provided to provide a more thorough and complete understanding of the disclosure of this application.

[0069] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the specification of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. The term "and / or" as used herein includes any and all combinations of one or more of the associated listed items.

[0070] Please refer to Figure 1 One embodiment of this application provides a collaborative response method for collision events based on smart streetlights and railings, including the following steps:

[0071] S100. The local collision sensing signal is acquired through the dynamic sensing module installed on this streetlight. In this step, it should be noted that the dynamic sensing module is a sensing unit composed of various sensors installed on the local streetlight and its supporting facilities. Its function is to acquire physical signs of potential collision events in real time. Specifically, this module includes a deformation sensor installed on the guardrail to detect the deformation of the guardrail under external force; a vibration sensor installed on the streetlight pole to acquire the vibration amplitude of the streetlight under collision impact; a sound sensor to acquire ambient sound decibels to obtain changes in noise signal amplitude during a collision; and some streetlights also have an image acquisition module to acquire image information around the streetlight to analyze the movement of vehicles, pedestrians, or objects. The real-time data acquired by these sensors is processed and encoded after a preset sampling interval to constitute the local collision sensing signal, which is used for subsequent event determination and collaborative processing.

[0072] S200: Receive event-aware data packets sent by other streetlights. These event-aware data packets include adjacent collision sensing signals sensed by other streetlights and the address information of the streetlight itself. In this step, it should be noted that the event-aware data packets are data units generated by other streetlight nodes based on adjacent collision sensing signals collected by their own dynamic sensing modules, and sent to this streetlight via the communication module. The adjacent collision sensing signals are consistent with the local collision sensing signals in type and data structure; the only difference is that the data originates from adjacent streetlight nodes. The event-aware data packets also contain the address information of the source streetlight, used to uniquely identify the data collection node, so as to accurately distinguish different data sources and maintain data traceability in subsequent fusion processing.

[0073] S300. The event-aware data packet and the local collision-aware signal are fused together. Based on the processing result, a decision data packet is generated and sent via broadcast. Specifically, this includes: In this step, it should be noted that the fusion processing aims to integrate multi-source sensing data from local and adjacent streetlights into a unified judgment criterion, thereby improving the accuracy and stability of collision event recognition. This step includes the following sub-steps:

[0074] S310. Extract the adjacent collision sensing signals and corresponding source street light address information contained in the event sensing data packet. In this step, it should be noted that the extraction process refers to parsing the received event sensing data packet, reading the adjacent collision sensing signals (i.e., collision-related data collected by the adjacent street light dynamic sensing module, such as railing deformation, street light amplitude values, abnormal image frames, sound intensity data, acceleration values, etc.) separately, and simultaneously extracting the source street light address information corresponding to the signal. This information is used for subsequent data alignment and fusion, ensuring that each sensing signal carries a traceable source node identifier, thereby providing a reliable data source for subsequent fusion judgment.

[0075] S320. Align and fuse adjacent collision sensing signals and local collision sensing signals according to preset timestamps, sampling intervals, and address priorities to obtain the multi-source sensing fusion result for the current road segment. In this step, it should be noted that the alignment process ensures consistency of all sensing data in the time dimension, avoiding misjudgments caused by sampling time differences. Simultaneously, address priorities can be set based on road layout (e.g., arterial road priority, intersection priority), assigning different weights to data from different sources during fusion to form a more representative basis for determining regional events.

[0076] S330. This streetlight determines whether the multi-source sensing fusion result meets the preset collision judgment conditions. If so, a decision data packet is generated. The decision data packet includes: the multi-source sensing fusion result, and the address information and timestamp of each source streetlight in all event sensing data packets participating in the fusion. In this step, it should be noted that the multi-source sensing fusion result refers to the result obtained after comprehensively processing detection data from different sensing devices, such as guardrail deformation data, streetlight vibration amplitude data, and noise signal amplitude data, through a data fusion algorithm. This result is used to more accurately determine whether a collision event has occurred. The preset collision judgment conditions refer to a set of judgment parameters and their threshold combinations pre-set before system deployment, based on the actual road environment, equipment sensitivity, and historical data statistics, used to determine whether a collision event exists. The thresholds can be parameterized according to the traffic conditions and safety requirements of different road sections. For example, situations such as vibration intensity exceeding the set vibration amplitude threshold and accompanied by an increase in the sound pressure level of the noise signal, or the appearance of blurred areas caused by high-speed movement in the image acquired by the image acquisition module, can all trigger collision judgments. To meet the criteria for collision determination, when the multi-source sensing fusion result meets the aforementioned preset criteria, the system automatically determines that a potential collision event exists and generates a decision data packet. The decision data packet is a set of data containing key information required for collision event determination, specifically including: the final determination result of multi-source sensing fusion, and the address information and timestamp of each source street light in all event sensing data packets participating in data fusion. The address information is used to identify the specific street light node from which the data originates, and the timestamp is used to record the specific time point of data collection for subsequent verification and traceability. By completely encapsulating the above fusion result and its determination criteria in the decision data packet, it can be ensured that the street light node receiving the data packet can clearly know the determination conclusion and its source, thereby ensuring information transparency and result reliability in the subsequent collaborative response process.

[0077] S340. This streetlight broadcasts a decision data packet to surrounding streetlights. In this step, it's important to note that broadcasting means simultaneously sending the decision data packet to multiple streetlight nodes within a preset communication range via the communication module. This eliminates the need for individual point-to-point connections; information is simply propagated within the coverage area, achieving rapid, low-latency data dissemination. This ensures that upon detecting a potential collision event, surrounding streetlights can obtain the judgment result immediately and initiate their respective response measures as needed, such as activating warning lights, broadcasting alarms, or uploading data to the control center, thereby improving the overall system's collaborative processing efficiency and emergency response speed.

[0078] Implementing the technical solution of this embodiment can achieve the following beneficial effects: It realizes a local self-triggered mechanism. Unlike traditional alarm systems that require manual triggering or remote control, this embodiment relies on a dynamic sensing module and has the ability to self-trigger collision perception, resulting in a faster response and effectively reducing the risk of missed alarms and delays. It improves the accuracy and robustness of event perception. Through a multi-source sensing data fusion mechanism, the sensing data of adjacent streetlights is fused with local signals, enhancing the system's ability to distinguish sudden collision events and avoiding single-point false alarms. It supports regional-level collaborative judgment and linkage response. This embodiment constructs a chain-like communication collaboration mechanism, enabling the judgment information of any node to quickly spread to surrounding nodes after triggering, forming a networked early warning system and improving the overall system response efficiency and coverage. It facilitates source tracing and intelligent control. The decision data packet contains detailed source streetlight addresses and timestamps, which can be used for subsequent record analysis and event reconstruction, while also providing a data foundation for the backend platform to dynamically adjust strategies.

[0079] Specifically, local collision sensing signals and adjacent collision sensing signals include: guardrail deformation, vibration amplitude information, and noise signal amplitude. Guardrail deformation refers to the degree of physical deformation of the guardrail; vibration amplitude information refers to the vibration amplitude of the streetlight body; and noise signal amplitude refers to the detected instantaneous noise in decibels. Local collision sensing signals and adjacent collision sensing signals refer to multiple types of physical signals acquired by acquisition devices (such as sensing modules installed on streetlights or guardrails). These signals can be used to determine whether a vehicle collision event has occurred. Guardrail deformation refers to the degree of physical deformation of the guardrail when subjected to an external impact; vibration amplitude information refers to the vibration intensity of the local or adjacent streetlight body, reflecting the kinetic energy transmission characteristics generated by the impact; and noise signal amplitude refers to the instantaneous high-decibel noise detected within the monitoring area, which can be used to reflect the acoustic characteristics of the collision. This sensing information possesses timeliness and locality, reflecting the multidimensional characteristics of collision events and serving as key input parameters for multi-source fusion judgment.

[0080] Accordingly, in S330, determining whether the multi-source sensing fusion result meets the preset collision determination conditions includes:

[0081] S331, The deformation of the railing exceeds a preset railing deformation threshold, and / or, a preset number of adjacent streetlights are detected to have vibration amplitude values ​​exceeding a preset vibration amplitude threshold within a preset time window, and / or, a preset number of adjacent streetlights are detected to have noise signal amplitude values ​​exceeding a preset noise signal amplitude threshold within a preset time window.

[0082] In this step, determining whether a collision event has occurred includes:

[0083] The railing deformation exceeds the preset railing deformation threshold: This means that when the railing deformation value detected by the deformation sensor in the dynamic sensing module is greater than the deformation threshold set by the system during the deployment phase, it is considered that the railing may have suffered a strong physical impact. This threshold can be parameterized based on the material, structural strength and historical impact data of the railing on site.

[0084] If a preset number of adjacent streetlights detect vibration amplitude values ​​exceeding a preset vibration amplitude threshold within a preset time window, it means that when at least a preset number of adjacent streetlight nodes (e.g., 2 streetlights) have vibration amplitude values ​​collected by their vibration sensors that are higher than the vibration amplitude threshold set by the system within a set time window (e.g., 500ms or 1s), it is determined that a major impact event may have occurred in that area. This condition can effectively eliminate false alarms caused by sporadic vibrations at a single point.

[0085] If a preset number of adjacent streetlights detect noise signal amplitude values ​​exceeding a preset noise signal amplitude threshold within a preset time window, it means that when at least a preset number of adjacent streetlight nodes detect instantaneous noise decibel values ​​higher than the noise amplitude threshold set by the system within a set time window, it is considered that an event accompanied by a strong impact sound may have occurred in the area. This condition can effectively capture the acoustic characteristics generated during a collision.

[0086] Combined Decision Scheme: In practical applications, the above three schemes can trigger the collision judgment process individually or in combination. For example, when the deformation of the railing exceeds the threshold and the vibration amplitude of the adjacent street light also exceeds the threshold, the accuracy of the collision judgment is higher; or, if the subsequent decision response is only triggered when at least two of the three signals exceed the threshold simultaneously, false alarms can be effectively reduced.

[0087] The technical solution implemented in this embodiment enables distributed intelligent decision-making. Any street light node can make an independent decision after receiving the necessary multi-source sensing information, avoiding the communication bottleneck and failure risk of a single central node. Multi-source sensing fusion combines local detection with the detection results of neighboring nodes, improving the accuracy and robustness of collision event determination and reducing false alarms and missed alarms. Self-triggered instant response does not rely on external manual triggering or remote scheduling. Front-end nodes can complete judgment and broadcasting in milliseconds, significantly shortening response time and improving the timeliness of accident handling. Highly collaborative decision-making capabilities allow data packets to drive surrounding nodes to execute functions such as lighting, prompts, or speed guidance as needed, forming a regionalized and hierarchical intelligent early warning network and improving the level of road traffic safety.

[0088] Furthermore, S330 also includes: based on the judgment that the preset collision determination conditions are met, further locating the collision center point based on the following information sources: the coordinates of the monitoring point with the largest railing deformation, the location of the street lamp node with the largest vibration amplitude, and the sound source directional positioning result; fusing and calculating the above information, outputting the spatial location information of the collision center point, and attaching this location information to the decision data packet for broadcasting.

[0089] In this step, it should be noted that the location of the collision center is based on the joint analysis of multiple sensory information, specifically including the following three core information sources:

[0090] The coordinates of the monitoring point with the largest deformation of the railing are obtained. Railings usually deform significantly when subjected to direct impact. Therefore, by collecting and comparing the railing deformation at each monitoring point in real time, the point with the largest deformation can be identified as the potential collision area.

[0091] In a collision event, the street light structure at close range will experience varying degrees of mechanical vibration due to the shock wave. By comparing the vibration sensor values ​​of multiple street light nodes, the node with the largest vibration amplitude is selected as one of the criteria for judgment.

[0092] The sound source directional localization results are obtained by collecting noise signals at the time of collision using microphone arrays deployed at different street light nodes, and performing directional analysis based on time difference and signal strength difference to determine the estimated location of the intersection of sound source directions.

[0093] The three types of information mentioned above will be uniformly input into the fusion algorithm, which will calculate the spatial location information of the collision center point using methods such as weighted center estimation and multi-sensor data fusion. Finally, the location result will be attached to the generated decision data packet and sent by nodes with broadcast permissions in the chain network to surrounding street light nodes and the upper control platform, realizing coordinated response and accurate alarm for collision events within the area.

[0094] The technical solution implemented in this embodiment enables rapid and accurate perception and localization of collision events, improving the system's response timeliness and spatial resolution. By fusing three types of heterogeneous information—barrier deformation, streetlight vibration amplitude, and sound source direction localization—the system effectively overcomes the limitations of single-sensor errors or occlusion interference, enhancing the reliability of collision detection. Simultaneously, by appending the collision center point coordinates to the broadcast data in real time, it not only provides targeted response data for nearby streetlights but also offers spatial coordinate support for subsequent traffic control, accident handling, and early warning strategy decisions, thereby constructing a highly collaborative and accurate intelligent urban roadside event detection and response mechanism.

[0095] Furthermore, S331 and beyond also includes: based on the determination that the preset collision determination conditions are met, judging the severity of the collision, and classifying the collision event into different severity levels according to preset severity level determination rules, the levels include:

[0096] Minor collision event: The maximum deformation of the railing is less than the first deformation threshold, the vibration acceleration value is less than the first acceleration threshold and the vibration duration is shorter than the first time threshold, and the audio energy value is lower than the first sound intensity threshold;

[0097] Medium-level collision event: The maximum deformation of the railing is between the first deformation threshold and the second deformation threshold, the vibration acceleration value is between the first acceleration threshold and the second acceleration threshold, or the vibration duration is between the first time threshold and the second time threshold, and the audio energy value is between the first sound intensity threshold and the second sound intensity threshold.

[0098] Severe collision event: The maximum deformation of the railing exceeds the second deformation threshold, or the vibration acceleration value is higher than the second acceleration threshold, or the vibration duration exceeds the second time threshold, or the audio energy value is higher than the second sound intensity threshold.

[0099] In this step, it's important to note that determining the severity of a collision is a crucial step in the fusion and analysis of multi-source sensing data. This provides an accurate basis for subsequent early warning level control and response mechanisms. The system first determines that the preset collision criteria are met, and then, based on sensor signals from the guardrail, streetlights, and ambient sound sources, classifies the collision level according to the following rules:

[0100] Minor collision events: When the maximum deformation of the railing is less than the first deformation threshold, the vibration acceleration value is less than the first acceleration threshold, the vibration duration is shorter than the first time threshold, and the audio energy value is lower than the first sound intensity threshold, the system identifies it as a minor collision. These events usually do not pose a substantial threat to the facility itself or the traffic environment and are only used for recording and mild reminders.

[0101] Moderate Collision Event: When the maximum deformation of the railing is between the first and second deformation thresholds, or when the vibration acceleration value, duration, and audio energy value are any combination between the first and second corresponding thresholds, the system will identify the event as moderate. Such events may cause some structural impact or safety hazards. The system will trigger a moderate-intensity area-level alert and, depending on the situation, coordinate with local or neighboring nodes to take further action.

[0102] Severe Collision Event: When any of the following conditions are met: the maximum deformation of the railing exceeds the second deformation threshold, the vibration acceleration value is higher than the second acceleration threshold, the vibration duration exceeds the second time threshold, or the audio energy value exceeds the second sound intensity threshold, the system will determine the event as a severe collision and quickly escalate the response mechanism, including but not limited to activating strong warning lights, regional broadcasting, and linking with the city-level emergency mechanism.

[0103] To improve the robustness and accuracy of the judgment, this step supports parallel analysis of multi-source features and a threshold fault tolerance mechanism. That is, when a certain sensing path fails or the error is too large, the system can complete the event level judgment through the other paths to ensure the stable operation of the system.

[0104] The technical solution implemented in this embodiment enables automated, multi-source fusion-based hierarchical judgment of the severity of collision events, giving the system the ability to quickly identify collision events of different levels and thus adopt corresponding early warning response strategies based on the level differences. This hierarchical mechanism not only improves the pertinence and effectiveness of early warning information but also helps avoid resource waste and information overload, enhancing the overall intelligence level and practical value of the system. Furthermore, through flexible and adjustable threshold settings and multi-channel fault-tolerant strategies, the system's adaptability and robustness in complex urban road environments are significantly enhanced, demonstrating good engineering feasibility and promising prospects for widespread application.

[0105] Furthermore, based on determining the severity and center location of the collision event, streetlights within different ranges are controlled to enter corresponding lighting warning response modes. These response modes include:

[0106] When a minor collision is determined, the streetlights within a radius of no more than 50 meters around the center of the collision will switch to yellow slow-flash warning lights.

[0107] When a collision is determined to be a moderate collision event, streetlights within a radius of 50 to 100 meters around the collision center will switch to alternating red and yellow flashing lights.

[0108] When a serious collision is determined, streetlights within a radius of no less than 100 meters around the collision center point will switch to a red flashing or rotating light mode.

[0109] In this embodiment, it should be noted that after a collision event occurs, the system not only completes the spatial positioning of the collision center point and the intelligent judgment of the severity of the event, but also, based on the collision center point as a reference, sets different lighting warning response modes for smart streetlights within different radii according to the severity level determined therein. Specifically, this includes:

[0110] Minor collision events: such as slight bending of railings, localized low-decibel impacts, weak vibrations, etc. These events are localized and low-risk. Therefore, within a radius of no more than 50 meters from the center of the collision, the intelligent streetlights within the control range will switch to yellow slow-flash lights to alert oncoming vehicles, but without excessively interfering with traffic order.

[0111] Medium-level collision events: such as significant deformation of the guardrail, significant vibration amplitude, and high sound source intensity. In this case, the intelligent streetlights within a radius of 50 to 100 meters will be controlled to switch to an alternating red and yellow flashing mode to guide oncoming vehicles to slow down and pay attention with a more obvious dynamic light effect.

[0112] In cases of serious collisions, such as broken railings, accompanied by loud impacts and violent vibrations, a warning radius of at least 100 meters is set out with the collision center as the center to prevent secondary accidents to the greatest extent possible. The streetlights in this area are switched to red flashing or rotating light modes to create a strong visual warning and prompt drivers at a distance to take deceleration or avoidance measures in advance.

[0113] To ensure accurate implementation of different response levels, the system needs to combine the center point coordinates and event level field attached to the decision data packet during execution, compare the spatial distance between each node and the center point, and initiate the corresponding response strategy accordingly.

[0114] The technical solution implemented in this embodiment can:

[0115] Achieve graded response: Dynamically control the response range and lighting pattern according to the severity of the collision event, enhancing the system's adaptability and response accuracy to events of different risk levels;

[0116] Improve warning efficiency: By dynamically dividing the warning range in conjunction with the spatial location of the collision center point, the warning vehicles are more targeted, reducing the interference of blind, all-area warnings on traffic.

[0117] Reduce the risk of secondary accidents: The expanded coverage of high-level warnings helps to warn approaching vehicles at high speeds earlier, effectively improving traffic safety control capabilities in the post-collision area;

[0118] Enhancing System Intelligence: The system integrates perception and judgment, spatial positioning, and visual response control, significantly improving the event perception and proactive early warning capabilities of the smart street light network. It embodies a collaborative mechanism of "local perception + proactive judgment + hierarchical linkage early warning."

[0119] Furthermore, this method also includes:

[0120] S400. When a streetlight node receives a decision data packet and identifies itself as being within the warning coverage area, the secondary accident prevention mechanism is activated. The warning coverage area refers to the spatial range calculated based on the center point of the collision event, road structure, and the pre-set safety protection radius, used to determine which streetlight nodes need to participate in secondary accident prevention. This range is typically calculated using a combination of geographic location databases, road topology information, and the relative coordinates of the streetlight nodes to ensure coverage of areas potentially at risk of secondary impact. When a streetlight node receives a decision data packet containing the above information and confirms through its internal location determination module that it is within this coverage area, the system immediately triggers the "secondary accident prevention mechanism" to reduce the risk of chain traffic accidents caused by the collision event. This includes the following steps:

[0121] S410. The speed detection module installed on the target streetlight node collects the speed information of oncoming vehicles on the current road segment. It should be noted that the speed detection module can be implemented in various ways, such as radar-based speed measurement modules, laser rangefinder speed measurement modules, video image recognition speed measurement modules, or inductive loop speed measurement modules. The specific selection can be determined based on the characteristics of the road segment and installation conditions. This module collects the speed information of vehicles approaching the target road segment in real time and removes invalid noise data through filtering algorithms to ensure speed measurement accuracy. The collected data is recorded in timestamp form and provided to the subsequent speed determination logic for analysis.

[0122] S420. Determine if the oncoming vehicle's speed exceeds the preset high-speed approach threshold. In this step, it's important to note that the high-speed approach threshold is a speed value preset before system deployment based on road speed limits, accident response time requirements, and traffic flow characteristics. It distinguishes between normal approach and potentially dangerous approach. When the speed detection module reports an oncoming vehicle speed exceeding this threshold, it is considered that the vehicle poses a risk of high-speed approach to the collision zone, and active intervention measures must be initiated immediately.

[0123] If so, then perform the following operations:

[0124] S421. Based on the oncoming vehicle direction and the collision center point location, determine the target streetlight node on the oncoming vehicle's path. In this step, it should be noted that the oncoming vehicle direction can be determined through the direction detection function of the speed detection module, video recognition algorithms, or multi-node collaborative speed measurement data. The collision center point location is provided by the decision data packet. After combining this information, the system can calculate the positions of all streetlight nodes on the oncoming vehicle's path using a road topology mapping table and mark these nodes as "intervention nodes" to execute orderly deceleration guidance control.

[0125] S422. Broadcast deceleration guidance control commands to multiple streetlight nodes along the path of oncoming vehicles, progressively increasing the intensity and frequency of the warning lights to form a progressive "intelligent deceleration guidance" chain. In this step, it should be noted that the deceleration guidance control commands include parameters for light brightness control, flashing frequency control, and execution duration. Progressive control means that streetlight nodes closer to the collision center point have higher warning light intensity and faster flashing frequency, thus creating a clear visual warning gradient as the driver approaches the accident area. This progressively increasing warning effect creates a gradual sense of urgency in the driver's visual perception, prompting the driver to gradually decelerate, improve reaction time, and reduce the probability of secondary accidents.

[0126] S430. The intelligent deceleration guidance process continues until either of the following conditions is met: the oncoming vehicle speed is below the set safety threshold and remains stable, or a collision event response termination signal is received. It should be noted that the set safety threshold is a lower speed limit set by the system based on road safety requirements. When the vehicle speed drops below this value and remains stable within the duration monitoring window, the vehicle is considered to have safely decelerated, and high-intensity guidance is no longer necessary. The collision event response termination signal is issued by the incident handling center or core streetlight node after confirming that there is no risk of secondary accidents at the scene. Upon receiving this signal, the system immediately terminates the deceleration guidance process and restores the streetlights to normal operating mode to save energy and reduce interference with normal traffic.

[0127] The technical solution implemented in this embodiment can quickly determine whether a streetlight node is within the warning range after a collision event is detected, and dynamically execute an "intelligent deceleration guidance" strategy based on the speed of oncoming vehicles, thereby significantly improving the ability to prevent secondary accidents. Through visualized guidance gradients and precise path judgment, it effectively achieves early warning and proactive intervention for high-speed vehicles, making it particularly suitable for nighttime, low-visibility, or high-traffic scenarios, greatly enhancing the intelligent response capability and accident handling efficiency of the traffic system.

[0128] Furthermore, S330 also includes: the multi-source sensing fusion result does not meet the preset collision determination conditions, but there is a suspicion of collision. It should be noted that "suspicion of collision" means that although the multi-source sensing fusion result does not reach the threshold combination that fully triggers the collision determination conditions, some sensing data has shown abnormal trends or local features related to collision. In this case, to avoid missing potential collision events due to overly strict threshold settings, the system will enter a verification mode, using further image analysis methods to confirm the nature of the event, thereby improving the overall accuracy and robustness of the determination. Specifically, this includes:

[0129] S332. The railing deforms, but the deformation amount does not exceed a preset railing deformation threshold; and / or, fewer than a preset number of adjacent streetlights detect vibration amplitude values ​​exceeding a preset vibration amplitude threshold within a preset time window; and / or, fewer than a preset number of adjacent streetlights detect noise signal amplitude values ​​exceeding a preset noise signal amplitude threshold within a preset time window. This step further judges cases where the multi-source perception fusion result does not meet the collision judgment conditions but is suspected of collision, and is used to trigger the subsequent image verification mechanism. This step sets multiple warning critical conditions to avoid misjudging or missing specific collision events. Specifically: when the railing deforms but the deformation amount does not exceed the preset threshold, it may be a slight contact or a false deformation caused by non-collision, and therefore cannot be directly identified as a collision event. When only fewer than a threshold number of adjacent streetlights detect vibration signal amplitudes exceeding the threshold within the preset time window, it may be a local non-collision vibration. Similarly, if the detected high noise signal only appears in a few streetlights, it may be caused by other interference sources. This step demonstrates the system's "tolerant triggering" mechanism for suspicious events, ensuring that subsequent verification can still be initiated even under weak signal conditions.

[0130] S333. Generate a collision verification request packet, which includes: the multi-source perception fusion result, and the address information and timestamp markers of each source street light in all event perception data packets participating in the fusion. In this step, it should be noted that the purpose of the collision verification request packet is to completely encapsulate the perception data of suspected collision events and distribute it to the street light nodes capable of performing video verification. The address information is used to uniquely identify the location of the street light node from which the data originates; the timestamp marker is used to locate the corresponding time segment during video retrieval to ensure the accuracy of image analysis.

[0131] Accordingly, S340 also includes:

[0132] S341. Send collision verification request packets to multiple target streetlights, where each target streetlight is located at the entrance and exit of the road segment where the source streetlights are located, and is equipped with an image acquisition module. In this step, it should be noted that the selection principle for target streetlights is to ensure that complete images of vehicles entering and exiting can be obtained to determine whether vehicle trajectories are abnormal. The image acquisition module can be a high-definition camera, a low-light night vision camera, or a panoramic fisheye camera; the specific model is matched according to the installation location and lighting conditions. Send collision verification request packets to the multiple target streetlights (located at both ends of the road segment where each source streetlight is located) through a chain network. Since these streetlights are equipped with image acquisition modules and have video analysis capabilities, selecting them as the execution nodes for video verification helps to comprehensively grasp the dynamic information of suspicious road segments and form a cross-device collaborative judgment mechanism.

[0133] S342. Target streetlights: Based on the source streetlight address information and timestamp markers in the collision verification request packet, retrieve video images of the road segments where each source streetlight is located. It should be noted that video retrieval can be completed through local caching or a background video storage server. The target streetlight quickly locates the corresponding road segment and time window video data segment by parsing the information in the verification request packet, avoiding delays caused by a full video search. Based on the address and timestamp information in the verification request packet, the target streetlight retrieves video images of the road segment where the source streetlight is located within the corresponding time period, ensuring that the acquired image data has clear temporal and spatial correlations, thereby supporting subsequent accurate behavior recognition and analysis.

[0134] S343. Analyze whether there is any abnormal behavior in the video images. Abnormal behavior includes: within a preset time window, a vehicle enters the road segment where each source street light is located but does not leave. In this step, it should be noted that the abnormal behavior determination can be based on video target detection and trajectory tracking algorithms. For example, by detecting the vehicle's license plate, outline, or movement trajectory, it can be determined whether the vehicle stays for a long time after entering the road segment, suddenly disappears, or is not detected at the exit, thereby inferring that a collision or blockage event may have occurred, which can be used as one of the core criteria for collision verification.

[0135] S344. If abnormal behavior is detected, a decision data packet is generated and broadcast. In this step, it should be noted that after detecting abnormal vehicle behavior, the target streetlight automatically generates a decision data packet. This packet is consistent with the scenario described in S330, containing the final collision determination result, the address information of the data source, and a timestamp. It is then broadcast in the chain network to ensure that relevant nodes (such as upstream smart light poles, backend systems, or area control nodes) can receive and respond immediately. The broadcast method can employ wired communication (such as fiber optic ring networks) or wireless communication (such as LoRa, NB-IoT, 5G, etc.) to ensure coverage of all streetlight nodes within the warning range and achieve rapid response.

[0136] The technical solution implemented in this embodiment can perform secondary verification using video image acquisition and behavior recognition capabilities when the perception fusion result has not yet clearly indicated a collision event. This effectively improves the system's accuracy in identifying minor collisions and signal weakening events. Through standardized construction of verification requests, multi-device collaborative image retrieval, dynamic behavior analysis, and broadcast warning strategies, this solution achieves a closed-loop judgment path from "possible event" to "confirmed event," enhancing system robustness, reducing false alarms and false negatives, and providing a reliable triggering basis for subsequent multi-level response mechanisms.

[0137] In an optional embodiment, S100 further includes:

[0138] S110. Based on a preset threshold, perform threshold detection and analysis on the local raw sensing signal. If the preset activation condition is met, switch the communication module from normal communication state to active communication state. The preset threshold is a numerical threshold set for the feature values ​​of the local raw sensing signal. The feature values ​​include at least one of amplitude, rate of change, and duration. The activation condition is that any or a combination of features of the local raw sensing signal reaches the preset threshold within a preset time interval. The local raw sensing signal is coarse-grained sampling data of railing deformation, vibration amplitude information, and noise signal amplitude obtained by low-power sampling under normal communication state.

[0139] When the communication module is in active communication state, S200 is executed.

[0140] In this embodiment, it should be noted that, to balance low-power standby in daily use and rapid response in the event of a collision, a state wake-up step S110 is set before step S100. When the street light node is in normal communication mode, the communication module operates in low-power listening mode, only periodically monitoring the local raw sensing signal, without establishing communication sessions with neighboring nodes or performing broadcasts. The system pre-sets a threshold for the local raw sensing signal and calculates its characteristic values ​​(including but not limited to amplitude, rate of change, and duration at least one) within a preset time interval; each characteristic value is matched one-to-one with the corresponding preset threshold to trigger a judgment. This embodiment does not limit the specific threshold value and can be calibrated according to the application scenario during deployment or operation.

[0141] When any or a combination of features of the local raw sensing signal reaches or exceeds the corresponding preset threshold within the preset time interval, it is determined that the activation condition is met, and a wake-up operation is performed to switch the communication module from the normal communication state to the active communication state for sending and receiving adjacent data packets and / or broadcasting. After entering the active communication state, S100 is executed immediately to complete the event-level local data acquisition, and subsequent steps such as S200 to S340 are executed to realize adjacent data reception, fusion determination, and (when necessary) broadcasting of decision data packets. To avoid conceptual confusion, the periodic monitoring under normal conditions is only used for activation determination and does not constitute data acquisition in S100.

[0142] When necessary communication is completed in active communication mode and the activation conditions are not met again within a fallback period, the communication module can automatically return to normal communication mode, thereby maintaining low power consumption and controllable communication load during non-event periods.

[0143] The technical solution implemented in this embodiment can perform only low-power listening and brief monitoring in the normal stage, and only enter the active communication state and execute S100 when the activation conditions are met, avoiding energy consumption and resource waste caused by long-term high duty cycle communication. Through the front-end link of "local monitoring → threshold wake-up → event-level acquisition and (conditional) broadcasting", local rapid diffusion and coordination can be achieved without relying on the central end for unified scheduling, shortening the front-end response latency and improving the efficiency of localized handling of sudden collisions. Exchanging data packets only in necessary time windows significantly reduces invalid communication volume, which is conducive to maintaining link stability and capacity planning in large-scale road network deployments. Even if individual links are not working, local broadcasting in the active state can drive neighboring nodes to complete fusion, positioning, classification, early warning and verification according to the established process, enhancing the overall resilience and robustness of the system. By judging thresholds for characteristics such as amplitude / rate of change / duration, and combining engineering de-jittering and automatic fallback design, false triggers caused by occasional noise or brief disturbances can be effectively suppressed, improving judgment accuracy and operational reliability.

[0144] This application also provides a collision event collaborative response system based on smart streetlights and railings. Applying the above-mentioned collision event collaborative response method based on smart streetlights and railings, the system includes: a dynamic sensing module, which is communicatively connected to the local streetlights and collects local collision sensing signals; a communication module, which receives adjacent event sensing data packets sent by neighboring streetlights; a collaborative decision-making module, which determines whether to send a collision event based on the local collision sensing signals and adjacent event sensing data packets; and a lighting control module, which controls the lighting mode of the local streetlights according to the instructions sent by the collaborative decision-making module.

[0145] In this embodiment, it should be noted that the collision event collaborative response system based on the smart street light 100 and the guardrail uses the smart street light 100 on one side of the road as the carrier and is deployed in conjunction with the sensing points on the side of the guardrail. The system consists of a dynamic sensing module, a communication module, a collaborative decision-making module, and a lighting control module, which work together to complete the localized perception, distributed judgment, and multi-node linkage response of collision events. The dynamic sensing module communicates with the local street light to acquire local collision perception signals; the communication module is responsible for exchanging event perception data packets with adjacent street lights; the collaborative decision-making module performs fusion judgment based on local and adjacent information and generates decision data packets; the lighting control module switches the local street light's illumination mode according to the decision instructions to achieve on-site visual early warning.

[0146] In the dynamic perception stage, collision-related physical quantities are collected and preprocessed, providing timestamped data input for subsequent fusion and judgment. Typical implementations include: a deformation sensor 70 mounted on the railing to characterize the deformation caused by external impact; a vibration sensor 20 mounted on the streetlight pole to reflect structural vibrations generated by impact transmission; a sound sensor 30 to capture abnormal acoustic features; and image acquisition on streetlights to form video / image evidence in sections requiring image verification or trajectory analysis. The module can perform basic filtering, noise reduction, initial threshold detection, and data encoding, outputting a unified format of local collision perception signals.

[0147] In the communication phase, the module establishes data exchange with adjacent streetlights via wired or wireless links, receives adjacent event sensing data packets according to a predetermined frame structure, and supports necessary forwarding. To ensure timing consistency, the data packets include identification fields such as sampling timestamps and source streetlight addresses; to improve robustness, multi-hop forwarding and broadcast propagation are supported. This phase ensures that adjacent nodes grasp the sensing dynamics of surrounding road segments in real time, providing context for distributed decision-making.

[0148] In the collaborative decision-making phase, the module performs time alignment, spatial correlation, and multi-source fusion on local collision perception signals and adjacent event perception data packets to form a fusion result for the current road segment. It then makes a judgment based on preset collision determination conditions: when the fusion result meets the triggering conditions, a decision data packet is generated and the source addresses and timestamps of the participating fusion components are written into it; when the fusion result is insufficient to directly trigger a collision but there is suspicion, a verification request can be constructed according to a predetermined process. Nodes with image capabilities retrieve video from the corresponding road segment and time window for abnormal behavior analysis. If the verification is successful, the process proceeds to decision broadcasting. For confirmed collision events, event classification and (optionally) center location estimation can be further performed according to predetermined rules, and the classification and location fields are encapsulated together in the decision data packet.

[0149] In the lighting control phase, the module receives instructions from the collaborative decision-making module to switch the local streetlight illumination mode, providing visual alerts to oncoming vehicles and pedestrians. To match event levels and spatial ranges, the control strategy can employ different flashing frequencies, brightness, or colors within different radii according to preset rules. When necessary, it can be linked with audio-visual devices to enhance visibility. In scenarios involving vehicle speed intervention, the alert can be progressively strengthened along the vehicle's path according to a predetermined guidance strategy, guiding high-speed approaching vehicles to decelerate until safety conditions are met or a response end signal is received, at which point the warning state is exited.

[0150] The technical solution implemented in this embodiment can complete the closed-loop processing from "multi-source acquisition - adjacency sharing - fusion judgment - hierarchical early warning" at the roadside front end without relying on a single central node, significantly shortening the link latency from perception to response. By aligning and fusing heterogeneous information such as guardrail deformation, guardrail vibration and environmental acoustics in time and space, it improves the accuracy of judgment in minor collisions, noise disturbances and complex background scenarios, and reduces false alarms and missed alarms. Relying on decision data packets containing source addresses and timestamps, it achieves traceability of results and consistent response of multiple nodes, enhancing the system's resilience in scenarios of link fluctuations and local failures. By linking light patterns according to event level and spatial location, it forms a range-based and hierarchical visual warning, which helps oncoming vehicles to slow down and avoid in time, reducing the risk of secondary accidents. The system architecture is modular and the interface is standardized, which facilitates flexible configuration of sensing and communication capabilities according to road conditions, meeting the engineering adaptation and large-scale promotion under different deployment densities and energy consumption targets.

[0151] Furthermore, the dynamic sensing module includes: a deformation sensor, which is communicatively connected to the local street light and installed on the railing to collect deformation data of the railing; a vibration sensor, which is installed on the local street light to collect vibration amplitude data of the local street light pole; a sound sensor to collect ambient sound decibel data around the local street light; and some local street lights are also equipped with an image acquisition module to collect image information around the local street light.

[0152] In this embodiment, it should be noted that the dynamic sensing module is communicatively connected to the local streetlights and is composed of a deformation sensor 70, a vibration sensor 20, a sound sensor 30, and an image acquisition module 10. It is used to collect, preprocess, and encode multi-source information related to collisions on-site, and output a time-stamped local collision sensing signal to the collaborative decision-making module. The deformation sensor 70 is installed at a force-sensitive location on the guardrail post or upper edge to collect the deformation (or equivalent signal related to the deformation) of the guardrail under external force in real time. The sensor can be zero-point calibrated and temperature drift compensated in conjunction with the guardrail structure, and a reasonable sampling period and upper limit of the range can be set to cover the variation range from minor scraping to medium-to-high intensity impacts. After low-pass / band-pass filtering, outlier removal, and amplitude normalization, the raw data forms a guardrail deformation feature sequence, and a local clock timestamp is added for subsequent synchronization and tracing.

[0153] Vibration sensor 20 is installed at a structurally stable part of the streetlight pole to collect the vibration amplitude (which can be characterized as the amplitude / energy index of displacement, velocity, or acceleration) transmitted to the pole by collision impact. To reduce the influence of environmental vibration and equipment resonance, vibration reduction and limiting measures can be taken in the installation location and fixing method. On the algorithm side, bandpass / envelope detection, short-time energy statistics, and rate of change analysis are combined to obtain the pole vibration characteristic sequence. When vibration peak values ​​are collected from multiple streetlight nodes, a reference can be provided for subsequent spatial correlation and center point estimation.

[0154] The sound sensor 30 is used to monitor changes in ambient sound levels in decibels, focusing on the acoustic characteristics of short-duration, high-energy impacts. To reduce interference from wind noise and continuous background noise, a windproof shield and appropriate sensitivity can be configured on the hardware side. On the algorithm side, A-weighting (or equivalent weighting), short-duration energy, and duration statistics are performed to obtain acoustic amplitude / duration characteristics. When multiple nodes simultaneously detect acoustic abrupt changes within a preset time window, the reliability of collision event assessment can be enhanced.

[0155] Image acquisition module 10 is installed on some local streetlights to acquire video / image information of road segments, primarily for image verification and trajectory analysis of suspicious events. To ensure the validity of the evidence, image frames are timestamped and associated with streetlight address information to quickly locate the corresponding road segment and time window video clip when a verification request arrives. In this embodiment, the image acquisition results are not considered essential sensing conditions, but are only used as auxiliary verification / location information when needed.

[0156] The technical solution implemented in this embodiment enables multi-dimensional, time-synchronized, and spatially relevant acquisition of collision-related physical quantities at the front-end node: through complementary sensing of guardrail deformation, guardrail vibration, and environmental acoustics, the detection rate and robustness of collisions with different intensities and contact methods are significantly improved, reducing false alarms / missed alarms; through local preprocessing and time alignment, local data and adjacent data have consistent timing and source identification during fusion judgment, facilitating the rapid generation of traceable decision data packets; when using an activation mechanism, the above features can be directly used for threshold triggering, enabling the system to maintain low-power monitoring under normal conditions and immediately enter active communication upon triggering, balancing energy consumption and timeliness; when secondary verification or central positioning is required, deformation / vibration features can be cross-verified with image information, further improving spatial positioning accuracy and event confirmation credibility; overall, the local collision sensing signal output by the dynamic sensing module provides high-quality input for subsequent fusion judgment, hierarchical early warning, and guidance control, thereby forming a closed loop of front-end rapid detection—adjacent sharing—cooperative response, improving the system's real-time performance, reliability, and scalability in actual road environments.

[0157] In an optional embodiment, the system can be equipped with the following functional units according to the road scene and power supply conditions: a millimeter-wave radar module 40, used to acquire kinematic information such as distance, speed, and approach direction of moving targets on the near road section, and output feature quantities bound to timestamps and node addresses for use by the collaborative decision-making module in event fusion judgment and vehicle path recognition. It is preferably installed in the upper middle part or near the lamp arm of the intelligent street light 100, and has basic filtering and anti-interference processing capabilities. When secondary accident prevention is involved, it can be used as one of the implementation methods of the speed detection unit, providing measurement basis for judging "whether the high-speed approach threshold is exceeded", and supporting a progressive deceleration guidance strategy along the oncoming vehicle path.

[0158] The millimeter-wave radar module 40 is used to collect near-field / environmental reference quantities (such as ambient noise level, light intensity, near-range disturbances / vibrations, etc.), and outputs auxiliary features for threshold adaptation, background compensation, and false trigger suppression, improving the stability of judgment in complex background noise or lighting change scenarios. Its sampling results are synchronized with the local clock and encoded along with the node address, facilitating time alignment and spatial correlation with local / adjacent data during the fusion stage. This probe does not replace existing deformation, vibration, or sound acquisition; it only serves as an enhanced auxiliary input in front-end judgment or triggering logic.

[0159] The solar power panel 60 is used to provide supplemental power or off-grid backup power for roadside equipment. It works in conjunction with local power management to maintain continuity between critical links and front-end sensing when mains power is limited or abnormally interrupted, thus ensuring uninterrupted S110→S100 wake-up and broadcasting under the "normal low power consumption, trigger-only communication" strategy. It is preferably installed on the upper part of the lamp arm or pole, and the output is connected to the system DC bus or energy storage unit after power management. Specific electrical parameters are not limited.

[0160] The technical solution implemented in this embodiment can enhance end-side perception and power supply resilience without changing the necessary components of the system: kinematic measurements provided by the millimeter-wave radar module 40 improve the accuracy of judging oncoming vehicle speed and approach direction, enhancing the triggering and execution effect of the secondary accident prevention chain; background references provided by the millimeter-wave radar module 40 enable adaptive correction of thresholds and triggering conditions, reducing false triggers and missed detections; and the power supply provided by the solar power panel 60 ensures continuous operation of critical perception and communication under abnormal power supply or remote road conditions, thereby further improving the system's real-time performance, reliability, and scalability in real-world road environments. All of the above units are optional configurations and do not limit the scope of protection of the claims.

[0161] The embodiments described above are merely illustrative of several implementations of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the scope of protection of this application. Therefore, the scope of protection of this patent application should be determined by the appended claims.

Claims

1. A collaborative response method for collision events based on smart streetlights and railings, characterized in that, Includes the following steps: S100. Collect local collision sensing signals through the dynamic sensing module installed on this street light; S200. Receive event sensing data packets sent by other streetlights, wherein the event sensing data packets include adjacent collision sensing signals sensed by other streetlights and address information of the other streetlights, and the local collision sensing signals and adjacent collision sensing signals include: railing deformation, vibration amplitude information, and noise signal amplitude; S300, The event-aware data packet and the local collision-aware signal are fused together. Based on the processing result, it is determined whether to generate a decision data packet and whether to send it via broadcast. Specifically, this includes: S310. Extract the adjacent collision sensing signal and its corresponding source street light address information contained in the event sensing data packet; S320. The adjacent collision sensing signal and the local collision sensing signal are time-aligned and fused according to the preset timestamp, sampling interval and address priority to obtain the multi-source sensing fusion result of the current road segment. S330. This streetlight determines whether the multi-source sensing fusion result meets the preset collision determination conditions, including: S331, The deformation of the railing exceeds the preset railing deformation threshold; and / or, A preset number of adjacent streetlights all detected vibration amplitude values ​​exceeding a preset vibration amplitude threshold within a preset time window; and / or, A preset number of adjacent streetlights were found to have noise signal amplitude values ​​exceeding a preset noise signal amplitude threshold within a preset time window. S340. If so, locate the collision center point based on the coordinates of the monitoring point with the largest railing deformation, the location of the street lamp node with the largest vibration amplitude, and the sound source directional positioning results. S341. Determine the severity of the collision and classify the collision event into different severity levels according to the preset severity level determination rules; S342. Control the streetlights in different ranges to enter the corresponding lighting warning response mode; S343. Generate a decision data packet, wherein the decision data packet includes: the multi-source sensing fusion result, the address information and timestamp of each source street light in all event sensing data packets participating in the fusion; S340. This street light broadcasts the decision data packet to surrounding street lights. S400. When a street light node receives the decision data packet and identifies itself as being within the warning coverage area, a secondary accident prevention mechanism is activated, specifically including: S410. Collect oncoming vehicle speed information for the current road segment through a speed detection module installed on the target street light node; S420. Determine whether the speed of the approaching vehicle exceeds a preset high-speed approach threshold. If so, perform the following operations: S421. Determine the target street light node on the path of the oncoming vehicle based on the direction of the oncoming vehicle and the location of the collision center point. S422. Broadcast deceleration guidance control commands to multiple street light nodes along the path of the oncoming vehicle, progressively increasing the intensity and frequency of the warning lights to form a progressive "intelligent deceleration guidance" chain; S430, the intelligent deceleration guidance process continues until any of the following conditions are met: the oncoming vehicle speed is lower than the set safety threshold and remains stable, or a collision event response end signal is received.

2. The collaborative response method for collision events based on intelligent streetlights and railings according to claim 1, characterized in that, The railing deformation refers to the degree of physical deformation of the railing. The vibration amplitude information refers to the vibration amplitude of the street lamp body; The noise signal amplitude is the instantaneous noise in decibels detected.

3. The collision event collaborative response method based on intelligent streetlights and railings according to claim 2, characterized in that, The S340 also includes: The coordinates of the monitoring point with the largest railing deformation, the location of the street lamp node with the largest vibration amplitude, and the sound source directionality positioning results are fused together to calculate and output the spatial location information of the collision center point. This location information is then attached to the decision data packet and broadcast.

4. The collision event collaborative response method based on intelligent streetlights and railings according to claim 3, characterized in that, The severity levels include: Minor collision event: The maximum deformation of the railing is less than the first deformation threshold, the vibration acceleration value is less than the first acceleration threshold and the vibration duration is shorter than the first time threshold, and the audio energy value is lower than the first sound intensity threshold; Medium-level collision event: The maximum deformation of the railing is between the first deformation threshold and the second deformation threshold, the vibration acceleration value is between the first acceleration threshold and the second acceleration threshold, or the vibration duration is between the first time threshold and the second time threshold, and the audio energy value is between the first sound intensity threshold and the second sound intensity threshold. Severe collision event: The maximum deformation of the railing exceeds the second deformation threshold, or the vibration acceleration value is higher than the second acceleration threshold, or the vibration duration exceeds the second time threshold, or the audio energy value is higher than the second sound intensity threshold.

5. The collision event collaborative response method based on intelligent streetlights and railings according to claim 4, characterized in that, The response modes include: When a minor collision is determined, the streetlights within a radius of no more than 50 meters around the center of the collision will switch to yellow slow-flash warning lights. When a collision is determined to be a moderate collision event, streetlights within a radius of 50 to 100 meters around the collision center will switch to alternating red and yellow flashing lights. When a serious collision is determined, streetlights within a radius of no less than 100 meters around the collision center point will switch to a red flashing or rotating light mode.

6. The collaborative response method for collision events based on intelligent streetlights and railings according to claim 1, characterized in that, S330 further includes: the multi-source sensing fusion result does not meet the preset collision determination conditions, but there is a suspicion of collision, specifically including: S332, The railing deforms, but the deformation does not exceed the preset railing deformation threshold; and / or, If fewer than a preset number of adjacent streetlights are detected to have vibration amplitude values ​​exceeding a preset vibration amplitude threshold within a preset time window; and / or, If fewer than a preset number of adjacent streetlights are detected to have noise signal amplitude values ​​exceeding a preset noise signal amplitude threshold within a preset time window; S333. Generate a collision verification request packet, wherein the collision verification request packet includes: the multi-source perception fusion result, the address information and timestamp marker of each source street light in all event perception data packets participating in the fusion; Accordingly, S340 further includes: S341. Send the collision verification request packet to multiple target streetlights, wherein the multiple target streetlights are located at the entrance and exit of the road segment where each source streetlight is located, and are equipped with an image acquisition module; S342. The target street light retrieves video images of the road segment where each source street light is located based on the source street light address information and timestamp in the collision verification request packet. S343. Analyze whether there is any abnormal behavior in the vehicle in the video image. The abnormal behavior includes: within a preset time window, the vehicle enters the road segment where each source street light is located, but does not leave. S344. If the abnormal behavior exists, the decision data packet is generated and sent via broadcast.

7. The collaborative response method for collision events based on intelligent streetlights and railings according to claim 1, characterized in that, The S100 further includes: S110. Based on a preset threshold, threshold detection and analysis are performed on the local raw sensing signal. If a preset activation condition is met, the communication module is switched from normal communication state to active communication state. The preset threshold is a numerical threshold set for the feature values ​​of the local raw sensing signal, and the feature values ​​include at least one of amplitude, rate of change, and duration. The activation condition is that within a preset time interval, any or a combination of features of the local raw sensing signal reaches the preset threshold. The local raw sensing signal is coarse-grained sampling data of railing deformation, vibration amplitude information, and noise signal amplitude obtained by low-power sampling under normal communication state. When the communication module is in active communication state, S200 is executed.

8. A collision event collaborative response system based on intelligent streetlights and railings, employing the collision event collaborative response method based on intelligent streetlights and railings as described in any one of claims 1-7, characterized in that, The system includes: The dynamic sensing module communicates with local streetlights to collect local collision sensing signals; The communication module receives adjacency event sensing data packets sent by adjacent streetlights; The collaborative decision-making module determines whether to send a collision event based on the local collision perception signal and the adjacent event perception data packet. The lighting control module controls the illumination mode of the local streetlights according to the instructions sent by the collaborative decision-making module.

9. The collision event collaborative response system based on intelligent streetlights and railings according to claim 8, characterized in that, The dynamic sensing module includes: A deformation sensor, which is connected in communication with the local street light, is installed on the railing to collect the deformation data of the railing; A vibration sensor is installed on the local street light to collect vibration amplitude data of the local street light pole. A sound sensor collects decibel data of ambient sound around the local streetlights; Some of the local streetlights are also equipped with an image acquisition module to collect image information about the area around the local streetlight.

Images