Intelligent speed measurement, snapshot and warning system for expressway maintenance deployment and control area

Abstract

The invention relates to the technical field of expressway traffic monitoring, in particular to an intelligent speed measurement snapshot warning system for an expressway maintenance deployment and control area, which comprises a millimeter-wave radar module, a snapshot camera module, a multi-degree-of-freedom holder bracket, an edge calculation module, an LED information board module, a dynamic grading warning unit and a rapid deployment structure, reighting coordinate mapping is established through joint calibration, and an edge calculation module fuses multi-source data and calculates risk coefficients in combination with vehicle speed, acceleration, distance and the like; the dynamic grading warning unit drives the LED information board to perform grading warning in different colors and modes; the multi-degree-of-freedom holder automatically adapts to road conditions, and efficient installation is achieved by quickly deploying the structure. According to the system, the problems of low precision, slow deployment and single warning of traditional equipment are solved, the monitoring accuracy, the risk early warning capability and the deployment efficiency of the maintenance deployment and control area are improved, and the operation safety and the traffic order are guaranteed.

Description

Technical Field

[0001] The present invention relates to the technical field of highway traffic monitoring, and in particular to an intelligent speed measurement, capture and warning system for a highway maintenance control area. Background Art

[0002] Highway maintenance zones, special areas for ensuring the safety of road maintenance operations, are crucial for traffic order management and risk prevention. Traditional maintenance control systems rely primarily on fixed speed limit signs, temporary warning cones, and manual patrols. These methods have gradually exposed numerous limitations in practical application and are unable to meet safety management requirements in complex traffic environments.

[0003] From the perspective of speed measurement and image capture technology, traditional equipment has significant functional shortcomings. Early radar speed guns were mostly single-band, susceptible to interference from adjacent vehicles in multi-lane scenarios. Insufficient angular resolution resulted in misalignment between the speed measurement target and the actual vehicle, with errors often exceeding 5 km / h. Fixed camera capture relies on preset trigger areas, making it prone to missed or incorrect captures when vehicles deviate from their lanes or on slopes. This significantly reduces license plate recognition accuracy, making it difficult to provide a reliable basis for law enforcement. Furthermore, radar and cameras are deployed as independent devices, lacking coordinate correlation and data fusion. Speed ​​information and image information for the same vehicle cannot be accurately matched, leading to the disconnect between "speed measurement without image, and image without speed."

[0004] In terms of warnings and guidance, fixed speed limit signs cannot dynamically adjust their information based on real-time road conditions. For example, during heavy rain, the normal speed limit will still be displayed, making it difficult to reflect the actual safe speed. Single warning modes (such as static text) provide insufficient visual stimulation to drivers. At high speeds, drivers can take up to 3-5 seconds to recognize and react to risks, which can easily lead to rear-end collisions. Furthermore, the installation and deployment of traditional equipment is cumbersome, requiring professionals to carry tools to secure the poles, connect cables, and perform parameter adjustments. The installation of a single set of equipment often takes over two hours, which not only takes up significant travel time but also increases the risk of conflicts between maintenance operations and normal traffic.

[0005] With the continuous growth of highway traffic and the diversification of vehicle types, the traffic environment in maintenance control areas has become increasingly complex. Factors such as the long braking distances of large trucks, the speeding behavior of small passenger cars, and the sudden drop in visibility in inclement weather have placed higher demands on the dynamic monitoring, real-time warning, and rapid response capabilities of control areas. Although some existing intelligent speed measurement devices attempt to combine radar and cameras, the lack of a precise joint calibration mechanism results in insufficient data fusion accuracy. Furthermore, most devices rely on mains power and wired networks, limiting deployment flexibility and making it difficult to adapt to the rapid deployment requirements of temporary maintenance and emergency repairs. Therefore, to address these issues, an intelligent speed measurement, capture, and warning system for highway maintenance control areas was proposed. SUMMARY

[0006] The present application aims to provide a highway maintenance control area intelligent speed measurement snapshot warning system to solve the problems raised in the background art.

[0007] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0008] A highway maintenance control area intelligent speed measurement snapshot warning system, comprising:

[0009] A millimeter wave radar module for transmitting frequency-modulated continuous wave and receiving echo, and outputting distance, speed and angle data of multi-lane vehicles in real time;

[0010] A snapshot camera module with high-resolution dynamic capture capability, responding to trigger instructions to perform millisecond-level speed measurement snapshot;

[0011] An edge computing module connecting the millimeter wave radar module and the snapshot camera module, performing radar and vision data fusion, license plate feature extraction and overspeed risk calculation;

[0012] A multi-degree-of-freedom gimbal support bearing the millimeter wave radar module and the snapshot camera module, supporting three-dimensional space angle automatic adjustment;

[0013] An LED information board module receiving instructions from the edge computing module through wireless communication, dynamically displaying overspeed vehicle information and speed limit warning;

[0014] A dynamic hierarchical warning unit driving the LED information board to switch display mode and color according to the overspeed determination result.

[0015] As a preferred scheme, the millimeter wave radar module and the snapshot camera module establish a coordinate mapping relationship through joint calibration:

[0016] An array of corner reflectors and a checkerboard calibration board are arranged in the overlapping field of view, and an EPnP algorithm is used to solve the rotation and translation matrix from the radar coordinate system to the camera image coordinate, with a calibration error less than or equal to 0.1 pixels.

[0017] As a preferred scheme, the automatic adjustment method of the multi-degree-of-freedom gimbal support includes:

[0018] Adjustment parameters are calculated based on lane line geometric features:

[0019] The pitch angle is obtained by subtracting half the image height from the longitudinal coordinate of the lane center point, dividing by the camera focal length, and then taking the inverse tangent function;

[0020] The horizontal angle is obtained by subtracting the horizontal coordinate of the intersection of the lower edge of the lane and the bottom edge of the image from one quarter of the image width, dividing by the projection length of the lower edge of the lane, and then taking the inverse sine function.

[0021] As a preferred solution, the speed measurement snapshot process is:

[0022] The millimeter wave radar detects the target vehicle 200 meters in front of the control area and outputs the real-time position coordinates;

[0023] When the target enters the 80-100 meter visual range, the edge computing triggers the vehicle detection model to locate the target;

[0024] When the target enters the preset snapshot area, the camera is instructed to perform snapshot;

[0025] Car plate recognition is performed and a dynamic risk value is calculated using a dynamic risk prediction model.

[0026] As a preferred solution, the edge computing module executes a radar-vision trajectory correction algorithm:

[0027] An improved extended Kalman filter model is used, wherein the Kalman gain is equal to the prediction error covariance multiplied by the observation matrix transpose, and then divided by the result of the observation matrix multiplied by the prediction error covariance multiplied by the observation matrix transpose plus the visual confidence factor multiplied by the observation noise covariance;

[0028] The visual confidence factor is calculated by dividing the negative ambient light noise variance of the natural constant e by the square of 30.

[0029] As a preferred solution, the car plate recognition uses a light adaptive convolution model:

[0030] In the feature extraction layer, the output feature map is equal to the sum of the convolution operation performed after superimposing the light mask on the input feature map, and then multiplied by the adaptive weight;

[0031] The light mask is the absolute value of the pixel gray value minus 128;

[0032] The adaptive weight is equal to 1 divided by 1 plus the local area gray variance value.

[0033] As a preferred solution, the dynamic risk prediction model is:

[0034] The risk coefficient is equal to the difference between the real-time vehicle speed and the dynamic speed limit threshold divided by the dynamic speed limit threshold, plus 0.3 times the absolute value of the vehicle acceleration divided by the gravitational acceleration, plus 0.5 times the negative distance from the control area of the natural constant e divided by the square of 50.

[0035] As a preferred solution, the LED information board control strategy includes:

[0036] When there is no vehicle, display the basic speed limit information in green with low brightness;

[0037] When the risk coefficient is greater than or equal to 0.8, display the speeding license plate and warning words in red with high brightness.

[0038] When the risk coefficient is greater than 0.3 and less than 0.8, a yellow flashing speed limit warning is displayed;

[0039] The data is transmitted wirelessly through LoRa, and the transmission delay is less than or equal to 1 second.

[0040] As a preferred solution, the dynamic speed limit threshold is adaptively adjusted according to the environment:

[0041] The dynamic speed limit threshold is equal to the reference speed limit value multiplied by 1 minus 0.2 multiplied by the rainfall intensity normalized value, minus 0.3 multiplied by 1 minus the real-time visibility divided by 200.

[0042] As a preferred solution, a rapid deployment structure is also included:

[0043] The device box and the gimbal support are connected at a second level through an electromagnetic lock.

[0044] The solar panels and lithium batteries form an off-grid power supply system.

[0045] The Beidou positioning module uploads the device spatial coordinates to the cloud monitoring platform in real time.

[0046] As can be seen from the above technical solutions provided by the present application, the intelligent speed measurement and snapshot warning system for the highway maintenance control area provided by the present application has the following beneficial effects:

[0047] The accuracy and reliability of speed measurement and snapshot are improved:

[0048] The long-distance monitoring of the millimeter wave radar module and the high-definition visual acquisition of the snapshot camera module are combined to realize data fusion, solving the shortcomings of traditional single sensors in complex environments, allowing speed measurement and license plate recognition to maintain stable performance under various lighting and weather conditions; The dynamic speed limit mechanism can flexibly adjust the judgment standard according to the actual environmental conditions, making the speed measurement result more consistent with the real-time road conditions.

[0049] Dynamic risk classification warning is realized, effectively preventing safety accidents:

[0050] Risk assessment is no longer based on vehicle speed alone, but considers multiple factors such as vehicle driving state and distance from the control area, providing a more comprehensive assessment of potential risks; The LED information board visually transmits risk levels to drivers through different colors and display modes, helping them quickly identify risks and slow down in time, thereby reducing the probability of accidents in the maintenance control area;

[0051] Adapting to diverse maintenance scenarios improves deployment efficiency:

[0052] The multi-degree-of-freedom pan / tilt bracket automatically adjusts the device's angle based on the actual road conditions, allowing one set of equipment to be used in a variety of road conditions, including straights, curves, and slopes, eliminating the need for repeated deployment. The rapid deployment structure features convenient connections and a detachable design, combined with an off-grid power supply system, significantly shortening installation time and minimizing traffic disruption, making it particularly suitable for temporary deployments.

[0053] Edge computing improves system response speed and reduces external dependence:

[0054] The edge computing module processes data on-site, reducing latency in data transmission to the cloud and ensuring the system can quickly respond to various situations. Local storage and short-range communication technologies also reduce reliance on the network, allowing the system to operate stably even in remote areas while also reducing the consumption of cloud computing power and data traffic.

[0055] The entire chain ensures stable system operation and improves management efficiency:

[0056] The system has hardware status monitoring and fault self-diagnosis capabilities, which can promptly detect and handle abnormal situations. The redundant design further ensures the continuous operation of the system. The local storage and traceability of data facilitates subsequent law enforcement and system optimization. Overall, the system reduces deployment and maintenance costs and improves the pertinence and efficiency of law enforcement through intelligent management. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] Figure 1 This is a schematic diagram of the overall structure of an intelligent speed measurement, capture and warning system for a highway maintenance control area according to the present invention. DETAILED DESCRIPTION

[0058] In order to make the purpose, technical solutions and advantages of the present invention more clearly understood, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0059] In order to better understand the above technical solution, the above technical solution will be described in detail below with reference to the accompanying drawings and specific implementation methods.

[0060] like Figure 1 As shown, an embodiment of the present invention provides an intelligent speed measurement, capture and warning system for a highway maintenance control area, comprising:

[0061] Millimeter-wave radar module, used to transmit frequency-modulated continuous waves and receive echoes, outputting real-time distance, speed, and angle data for multi-lane vehicles;

[0062] The snapshot camera module has high-resolution motion capture capabilities and responds to trigger commands to perform millisecond-level speed measurement and snapshots;

[0063] The edge computing module connects the millimeter-wave radar module and the snapshot camera module to perform radar data fusion, license plate feature extraction, and speeding risk calculation;

[0064] The multi-degree-of-freedom gimbal supports the millimeter-wave radar module and the snapshot camera module, and supports automatic adjustment of the three-dimensional space angle.

[0065] The LED information board module receives instructions from the edge computing module through wireless communication and dynamically displays speeding vehicle information and speed limit warnings;

[0066] The dynamic graded warning unit drives the LED information board to switch the display mode and color according to the speeding judgment result.

[0067] In this embodiment, the millimeter wave radar module and the snapshot camera module establish a coordinate mapping relationship through joint calibration:

[0068] A corner reflector array and a checkerboard calibration plate are placed in the overlapping field of view. The EPnP algorithm is used to solve the rotation and translation matrix from the radar coordinate system to the camera image coordinate system. The calibration error is less than or equal to 0.1 pixel.

[0069] Furthermore, the millimeter-wave radar module and the snapshot camera module, as the core sensing components of the intelligent speed measurement and warning system in highway maintenance control areas, do not work independently. Instead, they work closely together to build a high-precision vehicle monitoring system. The coordinate mapping relationship established through joint calibration is the key link to achieve this collaboration. The following details each module and the joint calibration mechanism.

[0070] Millimeter-wave radar module: the "sensing antennae" for long-distance dynamic monitoring:

[0071] The millimeter-wave radar module, based on frequency-modulated continuous wave (FMCW) technology, is capable of long-range detection of vehicles in multiple lanes within a 200-meter range. Its hardware configuration and functional design fully meet the requirements for precise monitoring in high-speed scenarios. It utilizes a 4-transmitter, 8-receiver antenna array operating in the 77-80 GHz frequency band, with a transmit signal bandwidth of 500 MHz. This configuration achieves a range resolution of 0.3 meters, enabling clear distinction between vehicles traveling closely in adjacent lanes. The signal processing unit, equipped with a high-performance FPGA chip, supports a target update rate of 100 times per second, enabling real-time tracking of high-speed vehicles.

[0072] In the workflow, the transmitting unit generates a linear frequency modulation signal, the frequency of which changes with time in a sawtooth wave with a period of 10ms. The receiving unit captures the echo signal reflected by the vehicle and mixes it with the local oscillation signal to obtain an intermediate frequency signal. After amplification by a low-noise amplifier (LNA) and bandpass filtering, the signal-to-noise ratio is increased to more than 30dB. The intermediate frequency signal is then subjected to a two-dimensional spectrum analysis using a fast Fourier transform (FFT): in the distance dimension, according to the formula (in, is the target distance, is the speed of light, is the difference in transmit and receive frequencies, is the frequency modulation period, is the signal bandwidth) to extract the target distance information; in the velocity dimension, using the Doppler effect, through the formula (in, is the vehicle radial velocity, is the millimeter wave wavelength, The vehicle's radial velocity is calculated using the Doppler shift algorithm. Angle measurement is based on the principle of a multi-channel phased array, using the MUSIC super-resolution algorithm to analyze the phase difference of the echo signal, achieving a horizontal and pitch angle measurement accuracy of ±0.5 degrees, accurately locating the vehicle's lane.

[0073] To cope with complex environments, the module has multiple anti-interference mechanisms: the Constant False Alarm Rate (CFAR) algorithm estimates the background noise in real time through a sliding window, dynamically adjusts the detection threshold, and controls the false alarm rate caused by fixed clutter such as guardrails and trees to 10 -6 Below; in rainy and foggy weather, the transmission power is automatically increased by 10% and the receiving gain is increased to compensate for signal attenuation; when the intensity of the co-frequency interference signal exceeds -80dBm, the transmission frequency is adjusted in steps of 1MHz until the interference frequency band is avoided to ensure the stability of the monitoring data;

[0074] Snapshot camera module: the "acquisition window" for high-definition visual information:

[0075] The snapshot camera module focuses on capturing high-definition images of vehicles within a visual range of 5-30 meters, providing high-quality visual data for license plate recognition and vehicle feature extraction. Equipped with a 9-megapixel CMOS sensor and an 8-32mm motorized zoom lens, it can achieve imaging of ≥100 pixels in the license plate area within the monitoring range, ensuring clear character details. The shutter speed is adjustable from 1 / 500 to 1 / 2000 seconds, and with a frame rate of 30 frames per second, it effectively eliminates motion blur from high-speed vehicles. The fill light system uses an adaptive infrared LED array, with a nighttime fill light range of 30 meters. The fill light intensity automatically adjusts with ambient light to avoid overexposure or underexposure.

[0076] To address complex lighting conditions, the module is equipped with wide dynamic range (WDR) technology, achieving a dynamic range of 120dB, enabling clear visualization of both the foreground and background of vehicles in backlit scenes. A local adaptive exposure algorithm analyzes the brightness distribution of the license plate area in the image in real time, adjusting exposure parameters accordingly to avoid overexposure of the characters due to direct sunlight. When the edge computing module issues a capture command, the camera completes image acquisition within 50 milliseconds via a hardware trigger interface and transmits the original image to the edge computing module via Gigabit Ethernet, providing a foundation for subsequent processing.

[0077] Joint Calibration of Millimeter-Wave Radar and Snapshot Cameras: A "Spatial Bridge" for Data Fusion

[0078] The core of joint calibration is to establish a precise mapping relationship between the radar coordinate system (3D spatial coordinates) and the camera image coordinate system (2D pixel coordinates), ensuring that the position information of the same vehicle in the two sensor data strictly corresponds. The specific process is divided into two parts: hardware deployment and algorithm solution.

[0079] During the hardware deployment phase, the corner reflector array and checkerboard calibration plate need to be deployed synchronously within the overlapping monitoring field of view of the radar and camera (5-80 meters). The corner reflector is a strong reflective target of the radar, and its three-dimensional coordinates in the radar coordinate system ( ) can be accurately measured by radar; the two-dimensional pixel coordinates of the corner points of the checkerboard calibration plate (square length 50mm) in the camera image They can be extracted by visual inspection algorithms, and their physical dimensions provide a scale reference for coordinate conversion. The two types of calibration objects must be evenly distributed within the field of view, covering near, medium, and far distances and multi-lane areas to ensure that the calibration results are valid across the entire field.

[0080] The algorithm uses the EPnP (Efficient Perspective-n-Point) algorithm: First, the radar 3D coordinates of the corner reflector and the image pixel coordinates of the checkerboard corner points are extracted to form multiple sets of corresponding point pairs; then the point pairs are input into the algorithm, and by constructing four virtual control points, the 3D space points are represented as linear combinations of the control points, and the rotation matrix from the radar coordinate system to the camera image coordinate system is efficiently solved ( ) and the translation vector ( ), and finally get the mapping relationship formula (The control points represent the three-dimensional space points as linear combinations of the control points, and efficiently solve the rotation from the radar coordinate system to the camera image coordinate system. 、 is the image pixel coordinate (pixel), is the camera internal parameter matrix (including focal length, principal point coordinates and other parameters), is the rotation matrix, is the translation vector, 、 、 is the three-dimensional coordinate in the radar coordinate system (meters);

[0081] To ensure accuracy, multiple rounds of iterative optimization are required: the reprojection error (the difference between the converted pixel coordinates and the actual detected coordinates) of all calibration point pairs is calculated. If the error exceeds 0.1 pixel, outlier points are removed and the solution is recalculated until the average error is ≤ 0.1 pixel. Through this calibration process, the edge computing module can convert the 3D vehicle coordinates output by the radar into image pixel coordinates in real time, guiding the camera to accurately select the target area, providing a unified coordinate reference for radar-based data fusion, license plate recognition, and other functions, eliminating data misalignment caused by sensor installation deviations.

[0082] In summary, the long-range dynamic monitoring of millimeter-wave radar and the high-definition visual acquisition of snapshot cameras complement each other based on the coordinate mapping constructed through joint calibration: the radar provides the camera with target location and trigger timing, and the camera supplements the radar data with visual details and license plate information. The two work together to form the system's high-precision perception layer, laying a solid foundation for subsequent speeding determination, risk assessment, and warning issuance.

[0083] In this embodiment, the automatic adjustment method of the multi-degree-of-freedom pan / tilt bracket includes:

[0084] Calculate and adjust parameters based on lane line geometry:

[0085] The pitch angle is obtained by subtracting half of the image height from the vertical coordinate of the lane center point, divided by the camera focal length, and then taking the inverse tangent function.

[0086] The horizontal angle is obtained by subtracting the horizontal coordinate of the intersection of the lower edge of the lane and the bottom edge of the image from one-quarter of the image width, divided by the projected length of the lower edge of the lane, and then taking the inverse sine function;

[0087] Furthermore, the multi-degree-of-freedom pan-tilt bracket serves as the "dynamic adjustment hub" supporting the millimeter-wave radar module and snapshot camera module in the intelligent speed measurement and warning system for highway maintenance and control areas. Through precise three-dimensional angle control, it ensures that both sensors always cover the monitoring area in the optimal posture, adapting to complex road conditions and control requirements, and providing a stable spatial reference for the spatiotemporal alignment and fusion analysis of radar data. The following describes this module in detail, from the overall perspective:

[0088] 1. Overview of overall functions:

[0089] The multi-degree-of-freedom gimbal supports the millimeter-wave radar module and the snapshot camera module. It dynamically adapts the sensor's field of view through automatic control of its two degrees of freedom: horizontal rotation and pitch. Its core functions include: real-time calculation and adjustment of parameters based on lane geometry, driving the bracket's precise rotation in three dimensions to ensure that multi-lane vehicles within a 200-meter monitoring range remain within the sensor's field of view. Closed-loop control ensures angle adjustment accuracy, maintaining a 90% or greater overlap between the radar and camera fields of view. The system also features environmental adaptability, maintaining structural stability in harsh conditions such as strong winds and temperature swings, providing continuous and reliable system support. It serves as a key regulatory component connecting perception-layer equipment with the controlled environment.

[0090] 2. Submodule composition and functions:

[0091] (1) Mechanical bearing and drive unit:

[0092] Main structure: A "T"-shaped frame made of aviation aluminum alloy is used. The bottom is fixed to the control area poles with four sets of M16 expansion bolts. The top platform has standardized mounting holes (compatible with M10 threads) to fix the radar (front end) and camera (rear end) respectively. The distance between the two is 50cm to avoid obstruction. The frame has an IP66 protection level and can operate in environments of -30℃ to +70℃. The wind resistance level is ≤10.

[0093] Drive system:

[0094] Horizontal rotation mechanism: driven by a 60 series stepper motor with a harmonic reducer (reduction ratio 1:120), output torque ≥ 6N•m, rotation range ±120°, speed 3° / s, and control accuracy 0.05°;

[0095] Pitch adjustment mechanism: uses 57 series stepper motor and self-locking worm gear reducer, output torque ≥10N•m, pitch range -20° to +35°, speed 2° / s, control accuracy 0.03°;

[0096] Transmission protection: Built-in torque sensor triggers overload protection when the load exceeds 150% of the rated value, automatically cutting off the motor power supply; high-temperature grease is applied to the gear meshing to reduce mechanical wear in low-temperature environments;

[0097] (2) Posture perception and control unit:

[0098] Angle feedback module: Both the horizontal and pitch axes are equipped with 20-bit absolute encoders with a sampling frequency of 100Hz, which collect current angle information in real time with a feedback delay of ≤5ms, providing accurate data for closed-loop control;

[0099] Control core: equipped with ARM Cortex-M4 core microcontroller, receive adjustment instructions from edge computing module through CAN bus (baud rate 500kbps), execute PID control algorithm, make the deviation between actual angle and target angle ≤0.1°;

[0100] Environmental adaptation component: integrate temperature and humidity sensor and vibration accelerometer, start cooling fan when detecting environmental temperature > 75℃, trigger filtering algorithm when recognizing > 2° / s² instantaneous vibration, eliminate interference signals;

[0101] (Three) automatic adjustment algorithm unit:

[0102] Parameter calculation module: based on lane line image features collected by camera, real-time calculation of adjustment parameters:

[0103] Pitch angle formula: , wherein, is the target pitch angle, is the longitudinal coordinate of the center point of the middle lane, is the image height, is the camera focal length;

[0104] Horizontal angle formula: , wherein, is the target horizontal angle, is the image width, is the horizontal coordinate of the intersection point of the lower edge of the lane and the bottom edge of the image, is the projection length of the lower edge of the lane;

[0105] Dynamic correction mechanism: receive lane line feature update every 300ms, when detecting angle deviation > 0.5° for 3 times in a row, start incremental adjustment mode, adjustment step 0.1° each time, avoid overshoot; for curved road scenes, additional curvature correction coefficient is introduced, make the horizontal angle adjustment amount inversely proportional to the lane curvature;

[0106] Three, key technology principles:

[0107] (One) double-degree-of-freedom closed-loop control principle:

[0108] Based on the "instruction-execution-feedback" closed-loop control model, the microcontroller compares the target angle instruction from the edge computing module with the actual angle feedback from the encoder, calculates the adjustment amount through the PID algorithm, and drives the stepper motor to rotate; the proportional term (P) quickly reduces the current deviation, the integral term (I) eliminates cumulative error, and the differential term (D) suppresses adjustment oscillation, so that the angle control precision is stable within 0.1°, ensuring the stability of the sensor field of view;

[0109] (Two) lane line feature driven adjustment principle:

[0110] A geometric model is constructed using lane marking image features captured by the camera, converting image pixel coordinates into physical angle parameters. Pitch angle adjustment is performed by calculating the longitudinal position deviation of the middle lane marking in the image, ensuring that the monitoring area is centered. Horizontal angle adjustment is based on the projection characteristics of the lower lane edge, ensuring that multiple lanes are evenly distributed in the image. Essentially, this involves dynamically aligning the sensor's field of view with the actual lanes through the mapping relationship between visual features and physical angles.

[0111] (3) Principle of environmental adaptability and stability:

[0112] Environmental adaptability is achieved through a combination of mechanical structural design and electronic protection: a high-strength frame made of aviation aluminum alloy withstands strong wind loads; IP66 protection grade isolates against rain and snow; temperature adaptive control prevents motor overheating; and a vibration filtering algorithm eliminates transient interference, ensuring structural deformation ≤0.5mm and angular drift ≤0.1° / h in environments ranging from -30°C to +70°C and wind speeds below level 10.

[0113] 4. Module workflow:

[0114] (1) Initialization phase:

[0115] After the system is started, the gimbal performs a "zero return" operation: the horizontal rotation is to 0° (straight ahead), the pitch is adjusted to the initial angle of +5°, the deviation between the mechanical zero point and the electrical zero point is calibrated using the encoder, and the compensation amount is stored in the local register;

[0116] Load initial parameters (such as camera focal length and image size), establish a CAN communication connection with the edge computing module, and send a "ready" signal. Initialization takes ≤30 seconds.

[0117] (2) Parameter reception and calculation stage:

[0118] The edge computing module sends lane line feature parameters every 300ms: the vertical coordinate of the center point of the middle lane , image height ,width , the horizontal coordinate of the intersection point of the lower edge of the lane and projection length ;

[0119] The gimbal control unit substitutes the parameters into the pitch angle and horizontal angle formula to calculate the target angle and , generate adjustment instructions;

[0120] (III) Angle adjustment and closed-loop control stage:

[0121] After the drive unit receives the command, the stepper motor rotates according to the calculated angle, and the encoder feeds back the current angle in real time;

[0122] The microcontroller continuously compares the target angle with the actual angle through the PID algorithm. When the deviation is ≤0.1°, it stops adjusting and sends back an "adjustment completed" signal. A single adjustment takes ≤2 seconds.

[0123] (IV) Exception handling stage:

[0124] If overload (torque > 9N•m) or overtemperature (> 75°C) is detected, the adjustment will be stopped immediately, a fault alarm will be sent, and manual intervention will be awaited;

[0125] When vibration disturbance is detected (acceleration > 2° / s²), the Kalman filter algorithm is activated to smooth the angle feedback, responding only to trend deviations that last for more than 1 second.

[0126] (V) Ending stage:

[0127] When the system receives a stop command, the gimbal bracket is reset to the initial angle (horizontal 0°, pitch +5°), the motor power is cut off, the operation log (such as the number of adjustments and maximum deviation) is saved, the communication connection is closed, and the workflow is completed;

[0128] 5. Application value of the module:

[0129] (1) Improving system environment adaptability:

[0130] Through dual-degree-of-freedom angle adjustment, one set of equipment can adapt to various road conditions such as straight roads, curves, and slopes, covering 3-5 lanes. This reduces the amount of equipment deployed in the maintenance control area and lowers initial investment costs. It is particularly suitable for the flexible deployment needs of temporary maintenance operations.

[0131] (2) Ensuring the accuracy of radar and visual data fusion:

[0132] Stable angle control and guaranteed field of view overlap keep the coordinate mapping error between the radar and camera ≤ 0.1 pixel, providing a reliable spatial reference for the radar trajectory correction algorithm (improved EKF), improving the accuracy of speeding determination and license plate recognition.

[0133] (3) Reduce operation and maintenance complexity:

[0134] The automatic adjustment function reduces the need for manual debugging, shortening the deployment time of a single device from the traditional 20 minutes to 5 minutes. The fault self-diagnosis and protection mechanism reduces the risk of equipment damage, extends the service life, and reduces the average annual maintenance cost by more than 30%.

[0135] (IV) Enhance system reliability:

[0136] The structural stability and angle control accuracy in harsh environments increase the system's mean time between failures (MTBF) to more than 1,000 hours, providing uninterrupted support for traffic monitoring in maintenance control areas.

[0137] In this embodiment, the edge computing module executes the radar trajectory correction algorithm:

[0138] An improved extended Kalman filter model is used, in which the Kalman gain is equal to the prediction error covariance multiplied by the observation matrix transpose, divided by the observation matrix multiplied by the prediction error covariance multiplied by the observation matrix transpose plus the visual confidence factor multiplied by the observation noise covariance;

[0139] The visual confidence factor is calculated by dividing the negative ambient light noise variance of the natural constant e by the square of 30;

[0140] Furthermore, the edge computing module is the "core brain" of the intelligent speed measurement and warning system in highway maintenance control areas. It undertakes key tasks such as radar data fusion, target feature extraction, speeding risk calculation, and equipment collaborative control. By processing perception layer data in real time at the control site, it avoids cloud transmission delays and bandwidth consumption, providing computing power support for the system's rapid response and accurate decision-making. Specifically, it is reflected in the following:

[0141] 1. Overview of overall functions:

[0142] With "low latency and high reliability" as its core goals, the edge computing module integrates multi-source sensor data for real-time analysis and decision-making. Its main functions include: receiving distance, speed, and angle data from millimeter-wave radar and image data from snapshot cameras, and accurately correcting the target trajectory through a radar-visual fusion algorithm; performing license plate recognition and feature extraction on vehicle images, and calculating risk factors based on dynamic speed limit thresholds; sending angle adjustment commands to the multi-degree-of-freedom gimbal based on the analysis results, outputting graded warning information to the LED information board, and simultaneously interacting with the cloud platform via wireless communication. It is the key hub connecting the perception layer and the execution layer, ensuring the real-time and accuracy of the system in complex traffic scenarios.

[0143] 2. Submodule composition and functions:

[0144] (1) Radar and vision data fusion unit:

[0145] Data preprocessing and spatiotemporal alignment: Receive target data (range, speed, angle) from the radar at 100 frames per second and image data from the camera at 30 frames per second. Align the asynchronous data using hardware synchronization pulses (microsecond level) and a linear interpolation algorithm to ensure that the spatiotemporal coordinate error of the same target is ≤10ms. Filter and denoise the radar data (using a median filter to remove transition points), and perform distortion correction on the image data (based on the camera's intrinsic parameter matrix), laying the foundation for fusion analysis.

[0146] Coordinate transformation and target association: The radar three-dimensional coordinates are transformed using the rotation and translation matrix obtained by joint calibration. Convert to image pixel coordinates , the radar target and the image detection box are associated by the IOU (Intersection over Union) algorithm (IOU≥0.5 is determined as the same target), and the identity confusion problem in the multi-target intersection scene is solved;

[0147] Trajectory correction algorithm: an improved extended Kalman filter (EKF) model is used to optimize the target trajectory, and the formula is:

[0148] Observation update: , wherein, is the Kalman gain, is the prediction error covariance, is the observation matrix, is the visual confidence factor, is the ambient light noise variance (lx), is the light noise reference value (30lx), is the observation noise covariance matrix; by dynamically adjusting the weight (when the light is good increase, enhance the influence of visual data; when the light is poor decrease, rely on radar data), the trajectory smoothness is improved by 40%, and the position error is less than or equal to 0.5 meters;

[0149] (2) Target feature extraction and recognition unit:

[0150] Vehicle detection and positioning: when the radar target enters the 80-100 meter visual range, the YOLOv8 target detection model is triggered, the vehicle area is located in the image (detection accuracy≥98%), and the vehicle bounding box coordinates and confidence are output, providing target area guidance for subsequent snapshot;

[0151] License plate recognition and character extraction: a light adaptive convolution model is used for the snapshot image, and the feature extraction layer output formula is:

[0152] , wherein, is the output feature map, is the number of convolution kernels, is the weight coefficient of the th convolution kernel, is the convolution operation, is the input feature map, is the feature map addition operation, is the light mask , is the pixel gray value, is the local area gray variance; by dynamically adjusting the convolution kernel weight, the license plate recognition accuracy in backlight and low light environment still remains above 99.2%, and the character extraction time is less than or equal to 20ms;

[0153] Vehicle type and motion feature analysis: Combining radar speed data with the vehicle's aspect ratio in the image, the vehicle type (sedan, truck, bus) is identified and the vehicle's real-time acceleration is extracted (via velocity differential calculation), providing multi-dimensional parameters for risk assessment.

[0154] (3) Speeding risk calculation and decision-making unit:

[0155] Dynamic speed limit threshold calculation: The speed limit value is adjusted in real time according to environmental parameters. The formula is:

[0156] ,in, is the dynamic speed limit threshold, is the base speed limit (default 80km / h), is the rainfall intensity influence coefficient, is the visibility influence coefficient, , , For real-time visibility, is the normalized value of rainfall intensity, ranging from 0 to 1. is the upper limit of visibility (200 meters); for example, heavy rain ( ) and visibility is 50 meters, the speed limit is automatically reduced to 52km / h;

[0157] Risk factor assessment: The risk value is calculated based on the vehicle speed, acceleration, and distance from the control area. The formula is:

[0158] ,in, is the dynamic risk coefficient, is the real-time vehicle speed, is the dynamic speed limit threshold, is the acceleration weight coefficient, is the distance weight coefficient, is away from the baseline value, and , , , is the vehicle acceleration, is the acceleration due to gravity, is the distance from the control area; When it is judged as high risk, The risk is medium, which is used as the basis for graded warnings;

[0159] Equipment collaborative control command generation: Based on the risk assessment results, angle adjustment commands are sent to the multi-degree-of-freedom pan / tilt bracket, and graded warning information is output to the LED information board to ensure coordinated response of the equipment;

[0160] (4) Communication and data management unit:

[0161] Local data storage: uses a 128GB industrial-grade SD card to store original images (only speeding vehicles), target trajectory data, equipment operation logs, etc., with a cycle coverage period of ≥30 days and support for breakpoint continuation;

[0162] Wireless communication interface: Communicates with the LED information board via the LoRa module (transmission distance ≥ 3km), with a delay of ≤ 100ms; interacts with the cloud platform via the 4GCat.1 module, uploads key data (batch upload every 5 minutes), and supports remote configuration updates and firmware upgrades;

[0163] Fault self-diagnosis: Real-time monitoring of CPU load (≤80% is normal), memory usage (≤70% is normal), and communication link status. When an abnormality occurs, a local buzzer alarm is triggered and a fault code is sent to the cloud.

[0164] 3. Key technical principles:

[0165] (1) Principles of radar-visual fusion and trajectory correction:

[0166] Based on the improved extended Kalman filter algorithm, the visual confidence factor is dynamically adjusted , achieving complementary advantages between radar and camera data; radar provides high-precision distance and velocity information, but has limited angular resolution; cameras provide high-resolution visual features, but are susceptible to lighting effects; during the fusion process, the filtering algorithm continuously corrects the target state through a prediction-update cycle, so that the trajectory maintains the stability of the radar while incorporating the detailed positioning of the camera, solving the problem of single sensor monitoring failure in scenes with occlusion and changing lighting;

[0167] (2) Principle of illumination adaptive license plate recognition:

[0168] To solve the problem of license plate recognition under complex lighting conditions, we introduce a light mask and local grayscale variance , dynamically adjust the convolution kernel weights In bright light areas, Enlargement Reduce, reduce the feature weight of the overexposed area; in the shadow area, Reduce Increase and enhance the feature response in low-light areas, thereby suppressing light interference and improving the robustness of character recognition;

[0169] (3) Principles of dynamic risk assessment:

[0170] Breaking through the traditional speeding judgment mode based solely on vehicle speed, the relative speeding ratio (reflecting the severity of speeding), acceleration shock (reflecting driving stability), and distance attenuation factor (reflecting urgency) are coupled as risk factors; through the weight coefficient 、 Reasonable settings make risk assessment more tailored to the safety needs of maintenance control areas. For example, a vehicle approaching a control area at low speed but accelerating rapidly may pose the same risk as a vehicle traveling at high speed but steadily. This avoids the limitations of a single indicator.

[0171] 4. Module workflow:

[0172] (1) Initialization phase:

[0173] After the module is powered on, it completes a hardware self-test (CPU, memory, storage, and communication interface), loads the pre-trained model (vehicle detection and license plate recognition) and initial parameters (baseline speed limit, calibration matrix, etc.), and takes ≤60 seconds.

[0174] Establish communication connections with millimeter-wave radar, snapshot camera, gimbal bracket, and LED information board, and synchronize system clocks (error ≤ 1ms);

[0175] (2) Data reception and preprocessing stage:

[0176] Receive radar data (every 10ms) and camera images (every 33ms) in real time, perform spatiotemporal alignment and noise filtering;

[0177] Perform distortion correction and region of interest (ROI) cropping on the image, retaining only the area containing the lane to reduce the amount of computation;

[0178] (3) Target detection and fusion stage:

[0179] When the radar detects a target entering the 200-meter range, the vehicle detection model is activated; when the target enters the 80-100-meter range, the camera is triggered to focus on monitoring the target area;

[0180] The radar and visual data are fused through coordinate transformation and IOU association, and the target trajectory is corrected using the improved EKF algorithm to output a smooth vehicle position and velocity sequence.

[0181] (IV) Feature extraction and risk calculation stage:

[0182] When the target enters the capture area (preset to be 50-80 meters), the camera is instructed to capture the image and perform license plate recognition and vehicle model analysis;

[0183] Calculate the risk factor by combining real-time vehicle speed, acceleration, distance from the control area, and dynamic speed limit threshold , determine the risk level;

[0184] (V) Decision-making and implementation stage:

[0185] Generate control instructions based on risk level: High risk ( ) when the LED information board is instructed to highlight the license plate and warning words in red; medium risk ( ) when the speed limit warning is displayed, the LED information board flashes yellow;

[0186] Extract lane line features from the image, calculate the gimbal adjustment angle, and send it to the gimbal bracket to ensure that the target remains in the center of the field of view;

[0187] Store key data and upload it to the cloud periodically, updating local logs;

[0188] (6) Ending stage:

[0189] After receiving the system stop command, the module stops receiving data, saves the current configuration and unuploaded data, and executes the safety shutdown process, which takes ≤ 10 seconds;

[0190] 5. Application value of the module:

[0191] (1) Improve system response speed:

[0192] The edge computing module controls data processing latency to less than 200ms, reducing latency by over 90% compared to cloud-based processing. This ensures that speeding vehicles are captured in real time and trigger alerts, giving drivers ample time to slow down and react.

[0193] (2) Enhanced adaptability to complex scenarios:

[0194] Through radar-visual fusion and adaptive illumination algorithms, the system maintains a target recognition accuracy of over 98% in scenarios such as rain, fog, backlight, and multi-lane traffic, solving the performance degradation problem of traditional speed measurement systems in harsh environments.

[0195] (3) Reduce deployment and operation and maintenance costs:

[0196] The integrated edge computing architecture reduces reliance on cloud computing power and lowers communication traffic consumption. Local storage and fault self-diagnosis functions reduce the frequency of manual on-site maintenance, reducing system operation and maintenance costs by more than 40%.

[0197] (4) Ensuring the effectiveness of law enforcement and security:

[0198] The dynamic risk assessment model makes speeding determinations more scientific and, combined with the accuracy of license plate recognition, provides a reliable basis for traffic enforcement. The graded warning function effectively reminds drivers to slow down. Field tests have shown that the average speed of vehicles in maintenance control areas has decreased by 15%, and the accident rate has dropped by 30%.

[0199] The edge computing module builds the system's "real-time decision-making center" by integrating multi-source data, optimizing algorithm models, and strengthening equipment collaboration. Its performance directly determines the accuracy and reliability of the entire speed measurement and warning system, and is the core technical support for realizing intelligent management of highway maintenance and control areas.

[0200] In this embodiment, the speed measurement and capture process is as follows:

[0201] The millimeter-wave radar detects the target vehicle 200 meters in front of the control area and outputs the real-time position coordinates;

[0202] When a target enters the visual range of 80 to 100 meters, edge computing triggers the vehicle detection model to locate the target;

[0203] When the target enters the preset capture area, the camera is instructed to capture the target;

[0204] Perform license plate recognition and calculate dynamic risk value using dynamic risk prediction model.

[0205] In this embodiment, the control strategy of the LED information board includes:

[0206] When there are no vehicles, the basic speed limit information is displayed in green with low brightness;

[0207] When the risk factor is greater than or equal to 0.8, the speeding license plate and warning message will be highlighted in red;

[0208] When the risk factor is greater than 0.3 and less than 0.8, the speed limit warning will flash in yellow;

[0209] Data is transmitted wirelessly via LoRa, with a transmission delay of less than or equal to 1 second;

[0210] Furthermore, the LED information board module serves as the "dynamic information window" for the intelligent speed measurement and warning system in highway maintenance control areas. It visually transmits speeding vehicle information and speed limit warnings to drivers, providing real-time reminders and safety guidance for violations. It is a key executive component for direct interaction between the system and road users, specifically:

[0211] 1. Overview of overall functions:

[0212] The LED information board module is primarily responsible for receiving wireless commands from the edge computing module and switching display modes based on the vehicle's dynamic risk level, accurately presenting the speeding vehicle's license plate, real-time speed, and corresponding warning information. Its core functions include: supporting multi-color display (red, yellow, and green) and dynamic flashing effects to meet graded warning requirements; responding to commands from the edge computing module in real time via wireless communication, ensuring information update latency ≤ 100ms; and possessing the ability to adaptively adjust to ambient light, maintaining display clarity in scenarios such as strong sunlight, backlight, and nighttime conditions, providing drivers with intuitive and timely traffic guidance and helping to reduce the risk of traffic accidents in maintenance control areas.

[0213] 2. Submodule composition and functions:

[0214] (1) Display unit:

[0215] LED array and panel structure: Utilizes a 32×128 dot matrix of high-brightness LED light-emitting tubes with a pixel pitch of 3mm and a display area ≥0.5㎡, supporting single-color (red, yellow, green) and dual-color combination displays. The panel is covered with anti-glare tempered glass with a nano-coating treatment, resulting in a light transmittance ≥90%, as well as scratch and UV resistance. It can emit light stably in environments ranging from -20℃ to +60℃, with a single-point brightness ≥8000cd / ㎡, ensuring clear visibility from 100 meters away.

[0216] Display mode control: supports three basic display modes:

[0217] Static display: used for regular speed limit information (such as "Speed ​​limit 80km / h in maintenance area"), with a font size of 50×50 pixels and a solid green color.

[0218] Flashing display: used for medium risk warning, yellow font flashes at a frequency of 2Hz (on 0.25s, off 0.25s), and the content includes "Note: Slow down and limit the speed to XXkm / h";

[0219] Highlight scrolling display: used for high-risk warnings, with red font scrolling horizontally at a speed of 10 characters per second, with the content "Yu A12345 speeding 10% slow down immediately", and the border flashing to enhance the warning effect;

[0220] (2) Wireless communication and control unit:

[0221] Communication interface: Equipped with an industrial-grade LoRa wireless module, operating frequency 470-510MHz, using spread spectrum communication technology, transmission distance ≥3km, strong anti-interference ability (can resist the interference strength of the same frequency signal ≤-90dBm); the module supports data encryption transmission (using AES-128 encryption algorithm) to ensure the security of command and information transmission and avoid malicious tampering;

[0222] Control core: Built-in STM32F103 microcontroller, after receiving the command frame (including display content, color, and mode parameters) from the edge computing module, parses the command and drives the display unit to perform the corresponding operation. At the same time, it feedbacks the current working status (such as normal display, communication interruption, power supply abnormality, etc.) to the edge computing module through heartbeat packets (once every 5 seconds);

[0223] (3) Environmental adaptation and power supply unit:

[0224] Light Sensing Adjustment Module: Integrated ambient light sensor (sampling frequency 10Hz) detects ambient light intensity in real time (range 0-100,000 lux). When light intensity is greater than 50,000 lux (strong light environment), the brightness is automatically increased to 100%. When light intensity is less than 500 lux (nighttime), the brightness is reduced to 30% to avoid glare. When backlit scenes are detected (light contrast > 100:1), local area brightness compensation is activated to improve edge clarity of displayed content.

[0225] Power Management: A wide voltage input (AC220V±20%) is converted to a DC5V / 10A output via a switching power supply to power the LED array and control unit. Built-in overcurrent, overvoltage, and short-circuit protection circuits automatically cut off the main power supply and activate the backup lithium battery (12V / 7Ah) when the input voltage exceeds the range (<176V or >264V). This provides a three-hour emergency display (only basic speed limit information is displayed), ensuring essential functionality in the event of a power outage.

[0226] 3. Key technical principles:

[0227] (1) Principle of graded warning display:

[0228] Based on the risk factor (Risk) output by the edge computing module, a mapping relationship between risk level, display mode, and color is established: when Risk < 0.3, a green static display is triggered to convey basic speed limit information; when 0.3 < Risk < 0.8, a yellow flashing mode is switched to issue a warning; when Risk ≥ 0.8, a red highlight scrolling mode is activated to highlight speeding vehicles. This grading mechanism uses differences in visual stimulation intensity (color conspicuity and dynamic effects) to enable drivers to quickly identify risk levels and take appropriate speed reduction measures.

[0229] (2) Wireless communication and command response principles:

[0230] Low-latency data interaction with the edge computing module is achieved through LoRa wireless communication technology. The communication protocol uses a custom frame format (including command type, display content length, and data check bit). The data volume per frame is ≤256 bytes, and the transmission rate is set to 9600bps, ensuring that the transmission time of a single command is ≤50ms. The receiving end verifies data integrity through CRC (cyclic redundancy check). If the verification fails, it automatically requests retransmission, with the number of retransmissions ≤3 to ensure the accuracy of command execution.

[0231] (3) Principle of adaptive adjustment of ambient light:

[0232] Using an ambient light sensor to collect real-time lighting data, the PID algorithm dynamically adjusts the LED drive current: for every 10,000 lux increase in light intensity, the drive current increases by 5% (maximum, no more than 120% of the rated current); for every 10,000 lux decrease in light intensity, the drive current decreases by 8% (minimum, no less than 20% of the rated current). This adjustment mechanism ensures display clarity under varying lighting conditions while avoiding energy waste and accelerated LED aging, extending the device's service life.

[0233] 4. Module workflow:

[0234] (1) Initialization phase:

[0235] After the module is powered on, it completes a hardware self-test (LED dot matrix, communication module, sensor), lights up all LEDs in turn for a full-brightness test (lasting 2 seconds), and then enters the standby state. By default, the green static message "Maintenance area speed limit 80km / h" is displayed;

[0236] The wireless communication module starts and searches for the edge computing module's signal. After establishing a connection, it sends a "ready" status frame and waits for instructions to be received.

[0237] (2) Instruction reception and analysis stage:

[0238] Monitor the LoRa communication channel in real time (signal detection every 10ms). When receiving the command frame from the edge computing module, perform CRC check and format analysis to extract the display content (such as "京B78901 20% overspeed"), color parameters (red / yellow / green), and display mode (static / flashing / scrolling).

[0239] If the command parsing fails (e.g. format error, verification failure), a "retransmission request" is sent to the edge computing module; if the parsing is successful, the display execution phase is entered;

[0240] (III) Display execution and adjustment stage:

[0241] According to the analysis results, the LED array is driven to display the corresponding content: static mode directly refreshes the dot matrix data; flashing mode controls the on and off switching through a timer; scrolling mode updates the display content column by column at a preset speed;

[0242] The ambient light sensor collects light data in real time and dynamically adjusts the LED brightness to ensure that the display effect is adapted to the current environment (e.g., the brightness is adjusted to maximum in strong sunlight at noon and to 30% at night);

[0243] (IV) Status feedback and exception handling stage:

[0244] After each display update, an "execution success" feedback frame is sent to the edge computing module, including the current display status and brightness value;

[0245] If a single-point LED fault is detected (three consecutive driver failures), the fault location is marked and recorded in the internal log. If communication is interrupted for more than 30 seconds, the system automatically switches to the default display mode (green static speed limit) and activates a buzzer (internal only, not external) to indicate the fault.

[0246] (V) Ending stage:

[0247] When the "stop display" command is received from the edge computing module or the power is cut off, the current display content is cleared first, the power supply of the LED array is turned off, the last display parameters and fault log are saved, and the shutdown process is completed.

[0248] In this embodiment, the system also includes a rapid deployment structure:

[0249] The equipment box and the PTZ bracket are connected in seconds through an electromagnetic lock;

[0250] Solar panels and lithium battery packs form an off-grid power supply system;

[0251] The Beidou positioning module uploads the device's spatial coordinates to the cloud monitoring platform in real time;

[0252] Furthermore, the rapid deployment structure is the key support for the "install-and-use" intelligent speed measurement and warning system in highway maintenance control areas. Through modular design and convenient connection methods, it shortens the on-site deployment time of system equipment to less than 5 minutes, meeting the urgent need for deployment efficiency in temporary maintenance operations while ensuring the installation stability of the equipment in field environments.

[0253] 1. Overview of overall functions:

[0254] The quick deployment structure is mainly responsible for the integrated installation and field fixation of each module of the system, and the core functions include: using standardized interfaces to realize the quick assembly of millimeter wave radars, snapshot cameras, edge computing modules, and gimbal supports, and the mechanical connection can be completed without professional tools; the quick fixation of equipment is realized through lightweight vertical rods and foundation components, which are suitable for different pavements such as asphalt and cement; the off-grid power supply system and Beidou positioning module are integrated to solve the deployment problem in the environment without city power and network in the wild; it has anti-wind and anti-seismic design to ensure the stable work of the equipment during deployment, and is the core structural component balancing the deployment efficiency and reliability of the system;

[0255] II. Submodule composition and function:

[0256] (I) Modular connection component:

[0257] Electromagnetic lock fixing mechanism: the connection between the equipment box and the gimbal support adopts electromagnetic lock design, the electromagnet (working voltage 12V, suction force ≥500N) is built-in the lock, and the locking and unlocking are controlled through the press-type trigger switch; when installing, insert the positioning pin of the gimbal support into the guide hole at the bottom of the box, trigger the electromagnetic lock to automatically attract, and complete the mechanical fixation; when disassembling, press the unlocking button, the electromagnet is powered off and released, the whole process does not need screwdrivers and other tools, and the time consumption is ≤30 seconds;

[0258] Integrated cable interface: the box side is equipped with a waterproof aviation plug (IP67 protection), which integrates power (DC12V), data (gigabit Ethernet), and control (CAN bus) interfaces, the plug adopts blind plug design (with guide slot), and automatically rotates and locks after insertion, the single interface connection time consumption is ≤10 seconds, avoiding the cable entanglement problem of traditional wiring;

[0259] (II) Lightweight vertical rod and foundation unit:

[0260] Segmented vertical rod: composed of 3 aluminum alloy round pipes (diameter 114mm, wall thickness 3mm), each length 1.5 meters, connected through quick buckle, total height ≥4 meters after unfolding; the flange (diameter 300mm) is welded at the bottom of the vertical rod, and 4 waist type holes are evenly distributed to adapt to different specifications of expansion bolts or counterweight block fixation methods;

[0261] Multifunctional foundation component: provides two foundation schemes:

[0262] Temporary foundation: uses 4 pieces of 20kg cast iron counterweight (total weight 80kg), fixed on the flange through bolts, suitable for scenes where the pavement cannot be damaged, and the deployment time is ≤5 minutes;

[0263] Permanent foundation: uses 4 groups of M20 expansion bolts (length 300mm), drilling depth ≥200mm, suitable for long-term deployment areas, anti-pulling force ≥5kN, can resist 10-level wind load;

[0264] (3) Off-grid power supply system:

[0265] Solar panels and brackets: Equipped with 300W monocrystalline silicon solar panels (conversion efficiency ≥ 23%), they are mounted in the middle of the pole using an adjustable bracket (pitch angle 0°-60°). The bracket adopts a quick-release design and is fixed with butterfly bolts, allowing for quick adjustment of the angle according to the sun's position. The panel surface is covered with hail-proof tempered glass (3.2mm thick) that can withstand the impact of hailstones with a diameter of 25mm.

[0266] Lithium battery pack and management module: Utilizes a 12V / 100Ah lithium iron phosphate battery pack (cycle life ≥ 2000 times), with a built-in BMS (battery management system) and overcharge, over-discharge, and overcurrent protection functions. Under conditions of an average of 4 hours of effective sunlight per day, the system can support continuous operation in rainy and cloudy weather for ≥ 3 days. When replenishing power through the AC charging port (AC220V), the full charge time is ≤ 8 hours.

[0267] (4) Beidou positioning and status monitoring unit:

[0268] Beidou Positioning Module: Integrates a Beidou dual-mode positioning chip (positioning accuracy ≤ 10 meters, timing accuracy ≤ 50ns), connects to an external high-gain antenna (gain ≥ 28dB) via an SMA interface, collects device latitude and longitude coordinates in real time, and reports location information to the cloud every 30 seconds for electronic fence management in the controlled area.

[0269] Attitude sensor: A triaxial accelerometer (range ±16g, sampling rate 100Hz) is installed on the top of the pole to monitor the equipment's tilt angle in real time. When the tilt angle is greater than 5°, a local sound and light alarm (buzzer + red LED) is triggered, and an early warning message is sent to the cloud, prompting maintenance personnel to check the equipment's fixed status.

[0270] 3. Key technical principles:

[0271] (1) Quick connection and locking principle:

[0272] The electromagnetic lock combines mechanical positioning with electromagnetic adsorption for rapid fastening. The alignment pin and guide hole ensure coaxiality (deviation ≤ 0.5mm). The axial suction generated by the electromagnet ensures a tight fit between the contact surfaces, and the friction torque can withstand the torsional force generated by force 10 winds. Compared to traditional bolt connections, this eliminates the need to tighten bolts individually, improving connection efficiency by over 80%. The on / off status of the electromagnetic lock can be remotely monitored via an edge computing module, preventing loose connections caused by human error.

[0273] (2) Principles of off-grid power supply energy management:

[0274] Based on the Maximum Power Point Tracking (MPPT) algorithm, the solar controller adjusts the charging voltage and current in real time to ensure that the solar panels operate at maximum power output (conversion efficiency ≥ 95%). The power supply strategy adopts a "solar priority + battery backup" mode: during the day when sufficient solar power is generated, the system directly supplies power to the load and charges the lithium battery. At night or on rainy days, the system automatically switches to lithium battery power. The BMS system monitors the battery SOC (state of charge) to control the depth of discharge (minimum SOC ≥ 20%), thereby extending the battery cycle life.

[0275] (3) Principle of wind resistance of lightweight structure:

[0276] The pole adopts a variable cross-section design, with a lower diameter of 114mm (for enhanced bending stiffness) and an upper diameter of 76mm (for reduced weight). The overall weight is ≤30kg, and the wind load coefficient is ≤0.6 (based on fluid dynamics simulation optimization). The bottom counterweight is designed with a downward-shifted center of gravity (the center of the counterweight is ≤15cm from the ground), which reduces the overall center of gravity of the equipment to below 2 meters. Combined with the rigid connection between the flange and the foundation, it forms a stable "light on top, heavy on the bottom" structure that can withstand the impact of a force 10 wind (wind speed 24.5-28.4m / s) without overturning.

[0277] 4. Module workflow:

[0278] (1) Preliminary preparation stage:

[0279] The millimeter-wave radar, snapshot camera, and edge computing module are pre-installed in the equipment box, and cable connections and functional testing are completed to form a "plug-and-play" integrated unit.

[0280] During transportation, the poles are folded into sections, and the solar panels, counterweights, and tool kits are packed separately. The entire set of equipment can be transported by an ordinary pickup truck and can be carried by one person.

[0281] (2) On-site assembly stage:

[0282] Pole erection: Deploy three sections of poles at the control point, connect and fix them in sequence with buckles, and adjust the verticality after erecting them (calibrate with a level bubble), which takes ≤5 minutes;

[0283] Equipment fixing: Install the integrated equipment box on the PTZ bracket at the top of the pole through the electromagnetic lock, press the lock button to confirm the lock, and connect the aviation plug to complete the electrical connection. It takes ≤ 1 minute.

[0284] Foundation fixing: Select counterweights or expansion bolts according to the road surface type to fix the bottom of the pole to ensure that the pole does not shake. It takes ≤3 minutes;

[0285] (III) Power supply and positioning startup phase:

[0286] Unfold the solar panel and adjust the angle to face due south (deviation ≤ 5°), connect it to the charging port of the lithium battery pack, and start the power supply system. The lithium battery will start charging and the device will automatically power on.

[0287] The Beidou positioning module searches for satellite signals (cold start time ≤ 30 seconds). After completing positioning, it sends the device coordinates to the cloud platform. The system displays the "deployment completed" status.

[0288] (IV) Deployment and operation phase:

[0289] The attitude sensor monitors the device status in real time. If there is a slight shake (tilt angle 3°-5°), the gimbal bracket will automatically adjust to compensate. If the tilt angle is greater than 5°, an alarm will be triggered immediately.

[0290] The off-grid power supply system automatically switches its operating mode according to the intensity of sunlight. The edge computing module records power supply parameters (such as battery voltage and charging current) and reports them to the cloud every hour.

[0291] (V) Recovery stage:

[0292] After receiving the recovery instruction, press the unlock button of the equipment box to release the electromagnetic lock, remove the integrated equipment box, and disconnect the aviation plug;

[0293] The poles were dismantled into sections, the counterweights and solar panels were recovered, and the site was cleaned up. The entire recovery process took ≤10 minutes, and no traces remained on the road surface.

[0294] While embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions, and variations may be made to these embodiments without departing from the principles and spirit of the invention, and that the scope of the invention is defined by the appended claims and their equivalents.

Claims

1. An intelligent speed measurement, capture and warning system for highway maintenance control areas, characterized by: include: Millimeter-wave radar module, used to transmit frequency-modulated continuous waves and receive echoes, outputting real-time distance, speed, and angle data for multi-lane vehicles; The snapshot camera module has high-resolution motion capture capabilities and responds to trigger commands to perform millisecond-level speed measurement and snapshots; The edge computing module connects the millimeter-wave radar module and the snapshot camera module to perform radar data fusion, license plate feature extraction, and speeding risk calculation; A multi-degree-of-freedom gimbal bracket carries the millimeter-wave radar module and the snapshot camera module, and supports automatic adjustment of three-dimensional spatial angles; The LED information board module receives instructions from the edge computing module through wireless communication and dynamically displays speeding vehicle information and speed limit warnings; The dynamic graded warning unit drives the LED information board to switch the display mode and color according to the speeding judgment result.

2. The intelligent speed measurement, capture and warning system for highway maintenance control areas according to claim 1 is characterized by: The millimeter wave radar module and the snapshot camera module establish a coordinate mapping relationship through joint calibration: A corner reflector array and a checkerboard calibration plate are arranged in the overlapping field of view, and the EPnP algorithm is used to solve the rotation and translation matrix from the radar coordinate system to the camera image coordinate system. The calibration error is less than or equal to 0.1 pixel.

3. The intelligent speed measurement, capture and warning system for highway maintenance control areas according to claim 1 is characterized by: The automatic adjustment method of the multi-degree-of-freedom pan / tilt bracket includes: Calculate and adjust parameters based on lane line geometry: The pitch angle is obtained by subtracting half of the image height from the vertical coordinate of the lane center point, divided by the camera focal length, and then calculating the inverse tangent function. The horizontal angle is obtained by subtracting the horizontal coordinate of the intersection of the lower edge of the lane and the bottom edge of the image from one-quarter of the image width, dividing by the projected length of the lower edge of the lane, and then taking the arcsine function.

4. The intelligent speed measurement, capture and warning system for highway maintenance control areas according to claim 1 is characterized by: The speed measurement and capture process is as follows: The millimeter-wave radar detects the target vehicle 200 meters in front of the control area and outputs the real-time position coordinates; When a target enters the visual range of 80 to 100 meters, edge computing triggers the vehicle detection model to locate the target; When the target enters the preset capture area, the camera is instructed to capture the target; Perform license plate recognition and calculate dynamic risk value using dynamic risk prediction model.

5. The intelligent speed measurement, capture and warning system for highway maintenance control areas according to claim 1 is characterized by: The edge computing module executes the radar trajectory correction algorithm: An improved extended Kalman filter model is used, in which the Kalman gain is equal to the prediction error covariance multiplied by the observation matrix transpose, divided by the observation matrix multiplied by the prediction error covariance multiplied by the observation matrix transpose plus the visual confidence factor multiplied by the observation noise covariance; The visual confidence factor is calculated by dividing the negative ambient light noise variance of the natural constant e by the square of 30.

6. The intelligent speed measurement, capture and warning system for highway maintenance control areas according to claim 4 is characterized by: The license plate recognition adopts the illumination adaptive convolution model: In the feature extraction layer, the output feature map is equal to the sum of the input feature map superimposed with the illumination mask, the convolution operation, and the multiplication by the adaptive weight; The light mask is the absolute value of the pixel gray value minus 128; The adaptive weight is equal to 1 divided by 1 plus the value of the grayscale variance of the local area.

7. The intelligent speed measurement, capture and warning system for highway maintenance control areas according to claim 4 is characterized by: The dynamic risk prediction model is: The risk factor is equal to the difference between the real-time vehicle speed and the dynamic speed limit threshold divided by the dynamic speed limit threshold, plus 0.3 multiplied by the absolute value of the vehicle acceleration divided by the acceleration of gravity, plus 0.5 multiplied by the negative distance from the control area of ​​the natural constant e divided by 50.

8. The intelligent speed measurement, capture and warning system for highway maintenance control areas according to claim 1 is characterized by: The LED information board control strategy includes: When there are no vehicles, the basic speed limit information is displayed in green with low brightness; When the risk factor is greater than or equal to 0.8, the speeding license plate and warning message will be highlighted in red; When the risk factor is greater than 0.3 and less than 0.8, the speed limit warning will flash in yellow; Data is transmitted wirelessly via LoRa, with a transmission delay of less than or equal to 1 second.

9. The intelligent speed measurement, capture and warning system for highway maintenance control areas according to claim 1 is characterized by: The dynamic speed limit threshold is adaptively adjusted according to the environment: The dynamic speed limit threshold is equal to the base speed limit value multiplied by 1 minus 0.2 multiplied by the normalized rainfall intensity value, minus 0.3 multiplied by 1 minus the real-time visibility divided by 200.

10. The intelligent speed measurement, capture and warning system for highway maintenance control areas according to claim 1 is characterized by: Also includes a quick deployment structure: The equipment box and the PTZ bracket are connected in seconds through an electromagnetic lock; Solar panels and lithium battery packs form an off-grid power supply system; The Beidou positioning module uploads the device's spatial coordinates to the cloud-based monitoring platform in real time.

Images