Pavement black ice detection and warning system and method based on distributed acoustic sensing

Abstract

This invention belongs to the field of road traffic safety monitoring technology, and provides a road surface black ice detection and early warning system and method based on distributed acoustic sensing. It includes multiple sets of sensing optical fibers buried under the road surface and multiple sets of sensing optical fibers installed on the roadside guardrail. The system preprocesses the detection signals acquired by the distributed acoustic sensing subsystem, decomposes the signals, removes interfering intrinsic mode components, and processes the remaining components using an adaptive noise cancellation algorithm to obtain a clean signal containing road surface condition characteristics. Time-domain, frequency-domain, and time-frequency-domain features are extracted from the clean signal. A pre-trained recognition model is used to process the extracted features to identify the road surface condition, and an early warning is issued based on the recognition results. This invention can achieve real-time detection, accurate differentiation, location, and graded early warning of road surface black ice and black snow, ensuring road traffic safety.

Description

Technical Field

[0001] This invention belongs to the field of road traffic safety monitoring technology, specifically relating to a system and method for detecting and warning of black ice on the road surface based on distributed acoustic sensing. Background Technology

[0002] The statements in this section are merely background information related to the present invention and do not necessarily constitute prior art.

[0003] Black ice and snow on the road surface are among the main hidden dangers causing road traffic accidents in winter and in low temperature and high humidity environments. Because black ice and snow are similar in color to the road surface, thin in thickness and highly concealed, drivers have difficulty predicting them in advance, which can easily lead to safety accidents such as increased vehicle braking distance and skidding and loss of control.

[0004] Existing road surface ice and snow detection technologies have many limitations: manual inspections are inefficient, have limited coverage, and are risky to operate in adverse weather conditions; point sensors (such as temperature and humidity sensors) can only monitor local points, have blind spots, and cannot reflect the overall road surface condition; video image-based detection technologies are significantly affected by lighting conditions and weather conditions, and their accuracy drops sharply at night and in rainy or snowy weather; radar detection technology is expensive and lacks sufficient sensitivity to detect thin ice layers (thickness < 5 mm).

[0005] Distributed acoustic sensing (DAS) technology uses optical fiber as the sensing medium, enabling long-distance, high-density, distributed vibration and acoustic signal monitoring along fiber optic links. It boasts advantages such as wide monitoring range, high spatial resolution, strong resistance to electromagnetic interference, and good environmental adaptability, and has been successfully applied in multiple fields. However, applying DAS technology to the detection of black ice and snow on road surfaces still faces core technical bottlenecks: the correlation between vibration signals generated by vehicle movement and the state of road ice and snow is unclear, making it difficult to extract effective features from complex traffic noise; and there is a lack of dedicated recognition models for black ice and snow, making it impossible to accurately distinguish between these two hidden road surface conditions. Summary of the Invention

[0006] To address the aforementioned problems, this invention proposes a road surface black ice detection and early warning system and method based on distributed acoustic sensing. This invention overcomes the shortcomings of existing road surface black ice and snow detection technologies, such as numerous monitoring blind spots, weak ability to identify hidden hazards, and poor real-time performance. It enables real-time detection, accurate differentiation, location, and graded early warning of road surface black ice and snow, ensuring road traffic safety.

[0007] According to some embodiments, the present invention adopts the following technical solution:

[0008] A road surface black ice detection and early warning system based on distributed acoustic sensing includes a distributed acoustic sensing subsystem, a signal transmission subsystem, and a data processing and intelligent analysis center, wherein:

[0009] The distributed acoustic sensing subsystem includes a buried optical fiber unit, a guardrail optical fiber unit, a host, and a fixing component. The buried optical fiber unit includes multiple sets of sensing optical fibers buried under the road surface, the guardrail optical fiber unit includes multiple sets of sensing optical fibers installed on the roadside guardrail, the host is used to transmit pulsed laser signals to the sensing optical fibers and receive feedback detection signals, and the fixing component is used to fix the sensing optical fibers.

[0010] The signal transmission subsystem is used to transmit the detection signals acquired by the distributed acoustic sensing subsystem to the data processing and intelligent analysis center.

[0011] The data processing and intelligent analysis center is used to preprocess the received detection signals, decompose the signals, remove interfering intrinsic mode components, process the remaining components through an adaptive noise cancellation algorithm to obtain a clean signal containing road surface condition features, extract time-domain, frequency-domain, and time-frequency-domain features from the clean signal, process the extracted features using a pre-trained recognition model, identify the road surface condition, and issue warnings based on the recognition results.

[0012] As an alternative implementation, the host includes a pulsed laser source, an optical fiber coupler, a photodetector, and a signal conditioning unit. The pulsed laser source uses a set communication band and emits laser signals at a set pulse width and repetition frequency. The emitted laser signal passes through the sensing optical fiber of the optical fiber coupler. During transmission in the optical fiber, the laser signal is affected by vehicle movement and road vibrations, generating backscattered light. The backscattered light is transmitted through the coupler to the photodetector and converted into an electrical signal.

[0013] The signal conditioning unit is used to amplify, filter, and perform analog-to-digital conversion on electrical signals to generate digital signals.

[0014] As an alternative implementation, the host also includes a calibration module, which is used to send a calibration signal through a standard vibration source according to a set period, and to correct the detection data in combination with ambient temperature and humidity data.

[0015] As an alternative implementation, the buried optical fiber unit includes multiple sets of sensing optical fibers buried in parallel at a certain distance below the road lane, with a certain laying spacing between the sensing optical fibers, and the burial range of the sensing optical fibers covers the entire lane.

[0016] The guardrail fiber optic unit consists of multiple sets of sensing optical fibers spirally bonded from top to bottom along both sides of the road guardrail;

[0017] By burying fiber optic units and guardrail fiber optic units, a two-layer monitoring network is formed between the road surface and the roadside.

[0018] As an alternative implementation, the fixing component includes an anti-corrosion sleeve, a buffer rubber pad, and a buckle. The embedded sensing optical fiber is fitted with an anti-corrosion sleeve, and a buffer rubber pad is filled between the anti-corrosion sleeve and the sensing optical fiber.

[0019] The sensing fiber optic cables attached to the guardrail are secured with clips, with a certain distance between adjacent clips.

[0020] As an alternative implementation, the data processing and intelligent analysis center includes a data receiving module, a multi-source noise reduction module, an ice and snow feature extraction module, an intelligent recognition module, a precise positioning module, and a database module, wherein:

[0021] The data receiving module is used to receive detection data sent by the transmission subsystem;

[0022] The multi-source noise reduction module is used to decompose the mixed signal into several intrinsic mode components through empirical mode decomposition and remove interference components; taking the traffic vibration signal during the snow-free period as a reference, the module uses an adaptive noise cancellation algorithm to filter out redundant interference and extract the pure road surface condition feature signal.

[0023] The ice and snow feature extraction module is used to extract multi-dimensional feature parameters from pure signals, including time-domain features, frequency-domain features, and time-frequency-domain features.

[0024] The intelligent recognition module is used to process the extracted features using a pre-trained recognition model to identify the road surface condition. The recognition model is a support vector machine model with radial basis kernel function and feature weight factor introduced.

[0025] The precise positioning module is used to calculate the precise location of dark ice and / or dark snow based on the principle of optical time domain reflection, combined with the topology map of the sensor fiber optic cable laying and the GIS map.

[0026] The database module is used to store the original detection signals, the noise-reduced signals, features, road surface condition recognition results, location information, and road maintenance records.

[0027] A method for detecting and warning of black ice on road surfaces based on distributed acoustic sensing includes the following steps:

[0028] Acquire the detection signals from the distributed acoustic sensing subsystem;

[0029] The detected signal is preprocessed, decomposed, and interference intrinsic mode components are removed. The remaining components are then processed by an adaptive noise cancellation algorithm to obtain a clean signal containing road surface characteristics.

[0030] Time-domain, frequency-domain, and time-frequency-domain features are extracted from the clean signal. The extracted features are then processed using a pre-trained recognition model to identify the road surface condition and issue warnings based on the recognition results.

[0031] As an alternative implementation method, the process of decomposing the signal, removing interfering intrinsic mode components, and processing the remaining components using an adaptive noise cancellation algorithm includes:

[0032] First, the mixed signal is decomposed into several intrinsic mode components through empirical mode decomposition to remove interference components; then, using traffic vibration signals during periods without ice or snow as a reference, redundant interference is filtered out through an adaptive noise cancellation algorithm to extract pure road surface condition feature signals.

[0033] Among them, the empirical mode decomposition screening condition is that the correlation coefficient between two adjacent intrinsic mode components must satisfy... , The modal component number;

[0034] The objective function of the adaptive noise cancellation algorithm is: ;

[0035] in, Let be the error cost function. The original signal contains noise. For reference noise signal, This is the adaptive filter weight vector. This represents the expectation operation; the optimal weights are solved using the least squares method. ( () is the autocorrelation matrix of the reference signal. This is the cross-correlation vector between the reference signal and the original signal.

[0036] As an alternative implementation, the process of extracting time-domain, frequency-domain, and time-frequency-domain features from a clean signal includes: time-domain features including peak value, kurtosis, impulse factor, and waveform factor extracted from the clean signal; frequency-domain features including center frequency, spectral entropy, and characteristic frequency band energy percentage extracted from the clean signal; and time-frequency-domain features including wavelet packet energy entropy extracted from the clean signal.

[0037] As an alternative implementation method, the process of using a pre-trained recognition model to process the extracted features, identify the road surface condition, and issue a warning based on the recognition result includes: using a recognition model based on an improved support vector machine, the training samples of the recognition model include signal samples of different types of road surface conditions such as normal road surface, dark ice road surface and dark snow road surface, and improving feature discrimination by introducing a kernel function optimization strategy;

[0038] Improvements to support vector machines include using a radial basis function kernel and introducing feature weight factors. The kernel function expression is as follows: ;

[0039] in, For the input feature vector, For kernel function parameters, For feature dimension, For the first The weights of each feature are obtained through optimization using the particle swarm optimization algorithm. ;

[0040] The classification decision function of the recognition model is: ;

[0041] in, For the Lagrange multipliers of the support vector machine, The sample labels are -1 for normal road surfaces, 0 for dark snow surfaces, and 1 for dark ice surfaces. For bias terms, This represents the number of support vectors.

[0042] As an alternative implementation method, when the identification model identifies dark ice and dark snow, it calculates the precise location of the dark ice and dark snow based on the principle of optical time-domain reflectometry, combined with the laying topology map of the sensing optical fiber and the GIS map. The positioning process is as follows: ;

[0043] in, This represents the distance from the potential hazard point to the starting end of the optical fiber. The speed of light in a vacuum The round-trip time difference of the laser transmission. The refractive index of the sensing fiber;

[0044] And perform topology calibration on the positioning results: ;

[0045] in, For the final positioning distance, This is the initial positioning distance. This is a topology correction value, calibrated by fiber optic laying error. rice.

[0046] Compared with the prior art, the beneficial effects of the present invention are as follows:

[0047] This invention employs a dual-layer fiber optic cable laying method that combines underground installation and roadside guardrail binding. By leveraging the full-area monitoring advantages of DAS technology, it achieves real-time monitoring of the entire road area without blind spots. A single fiber optic cable can cover tens of kilometers of road, completely solving the limitations of traditional point-based monitoring and effectively capturing potential hazards of black ice and snow at any location on the road.

[0048] This invention innovatively designs a noise reduction algorithm that combines empirical mode decomposition and adaptive noise cancellation. By using a clear mathematical model to filter out traffic noise and environmental interference signals in road scenes, and combined with an optimized recognition model, it can accurately distinguish between dark ice, dark snow and normal road surfaces with an accuracy rate of ≥97%, solving the technical problem that dark ice and dark snow are highly concealed and difficult to identify.

[0049] Based on the principle of optical time-domain reflectometry, the positioning formula and topology calibration formula can accurately locate hidden dangers with a positioning accuracy of ±0.5m. Combined with maintenance work orders, it can significantly shorten the time for hidden danger investigation and handling, improve road maintenance efficiency, and reduce the incidence of traffic accidents.

[0050] This invention has strong anti-electromagnetic interference capability, good weather resistance, and can adapt to harsh environments such as high temperature, severe cold, rain and snow. It can achieve uninterrupted monitoring, high operational stability, and low maintenance cost. It is suitable for various road scenarios such as highways, bridges, and tunnels, and has broad application value.

[0051] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, preferred embodiments are described below in detail with reference to the accompanying drawings. Attached Figure Description

[0052] The accompanying drawings, which form part of this invention, are used to provide a further understanding of the invention. The illustrative embodiments of the invention and their descriptions are used to explain the invention and do not constitute an improper limitation of the invention.

[0053] Figure 1 This is a structural block diagram of a road surface black ice and snow detection and early warning system according to one embodiment;

[0054] Figure 2 This is a flowchart illustrating a method for detecting and warning of black ice and snow on the road surface according to one embodiment.

[0055] Figure 3 A schematic diagram illustrating the deployment of a DAS sensing deployment subsystem according to one embodiment;

[0056] Figure 4 This is a schematic diagram of the structure of an improved SVM model according to one embodiment. Detailed Implementation

[0057] The present invention will be further described below with reference to the accompanying drawings and embodiments.

[0058] It should be noted that the following detailed description is illustrative and intended to provide further explanation of the invention. Unless otherwise specified, 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 invention pertains.

[0059] It should be noted that the terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit the scope of exemplary embodiments according to the invention. As used herein, the singular form is intended to include the plural form as well, unless the context clearly indicates otherwise. Furthermore, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they indicate the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0060] Where there is no conflict, the embodiments and features described in this application may be combined with each other.

[0061] Example 1

[0062] A road surface black ice detection and early warning system based on distributed acoustic sensing, such as Figure 1 As shown, it includes a DAS sensor deployment subsystem, a signal transmission subsystem, a data processing and intelligent analysis center, and an early warning and linkage subsystem. These subsystems work collaboratively to achieve real-time, full-area monitoring and early warning of road conditions.

[0063] The DAS sensing deployment subsystem, as the core of signal acquisition, is used to realize the real-time acquisition of acoustic vibration signals across the entire road surface. It includes the DAS host, sensing optical fiber, optical fiber fixing components, and a calibration module.

[0064] The DAS main unit consists of a pulsed laser source, fiber optic coupler, photodetector, and signal conditioning unit. The pulsed laser source uses a 1550nm communication band with a pulse width of 10-30ns and a repetition frequency of 30-80kHz to ensure accurate capture of weak vibration signals. The pulsed laser source emits a narrow-pulse laser signal, which is injected into the sensing fiber via the fiber optic coupler. As the laser propagates through the fiber, it is affected by vehicle movement and road vibrations, resulting in backscattered light. This backscattered light is transmitted to the photodetector via the coupler and converted into an electrical signal. The signal conditioning unit amplifies, filters (passband frequency 10Hz-15kHz), and performs analog-to-digital conversion on the electrical signal to generate a digital signal.

[0065] Sensing fiber optics, such as Figure 3 As shown: This embodiment uses weather-resistant armored single-mode optical fiber, and the laying method adopts a dual mode of road surface burial + roadside guardrail binding; it is buried parallel to the road lane at a distance of 5-8cm, with a laying spacing of 0.3-0.8m, covering the entire lane; at the same time, it is spirally bound to the guardrails on both sides of the road from top to bottom to form a "road surface-roadside" dual-layer monitoring network.

[0066] Fiber optic fixing components include corrosion-resistant sleeves, buffer rubber pads, and stainless steel clips; the buried optical fiber is encased in a high-density polyethylene corrosion-resistant sleeve, and a buffer rubber pad is filled between the sleeve and the optical fiber to prevent damage to the optical fiber from road subsidence and vehicle rolling; the optical fiber is fixed to the guardrail using stainless steel clips with a clip spacing of 1-1.5m.

[0067] Calibration module: Composed of a standard vibration source, temperature sensor and humidity sensor, it periodically emits calibration signals and combines them with ambient temperature and humidity data to correct acquisition errors, ensuring that monitoring accuracy is not affected by environmental factors.

[0068] The signal transmission subsystem is used to achieve stable data transmission between the DAS sensing deployment subsystem and the data processing and intelligent analysis center. It adopts a dual-link architecture of fiber optic leased line + 5G redundant backup.

[0069] Among them, the fiber optic leased line adopts a single-mode fiber direct connection method with a transmission rate of no less than 1Gbps, which is used for the real-time transmission of massive amounts of raw monitoring data.

[0070] 5G Redundancy Backup Module: Deployed at communication base stations along roads, it automatically switches when the fiber optic leased line fails, ensuring uninterrupted data transmission with a transmission rate of ≥100Mbps;

[0071] Data encryption: This embodiment uses the AES-256 encryption protocol to encrypt the transmitted data and adds a CRC32 checksum to prevent data leakage or tampering.

[0072] The data processing and intelligent analysis center is the core processing unit of the system, used for noise reduction, feature extraction, status recognition and positioning of the collected signals. It includes a data receiving module, a multi-source noise reduction module, an ice and snow feature extraction module, an intelligent recognition module, a precise positioning module and a database module.

[0073] Specifically, the data receiving module receives encrypted data from the transmission subsystem and performs decryption, parsing, and format conversion.

[0074] Multi-source noise reduction module: It adopts a combination algorithm of "Empirical Mode Decomposition (EMD) + Adaptive Noise Cancellation"; firstly, the mixed signal is decomposed into several intrinsic mode components (IMF) through EMD to remove interference components such as vehicle horn and environmental wind noise; then, taking the traffic vibration signal during the non-icy and snowy period as a reference, the adaptive noise cancellation algorithm is used to further filter out redundant interference and extract the pure road surface condition feature signal.

[0075] Empirical Mode Decomposition (EMD) screening criteria: The correlation coefficient between two adjacent intrinsic mode components must meet the following conditions. ( (As modal component indices), ensuring the effectiveness of mode separation;

[0076] Objective function of adaptive noise cancellation algorithm: ;

[0077] in, Let be the error cost function. The original signal contains noise. For reference noise signal, This is the adaptive filter weight vector. This represents the expectation operation; the optimal weights are solved using the least squares method. ( () is the autocorrelation matrix of the reference signal. (This is the cross-correlation vector between the reference signal and the original signal).

[0078] Ice and snow feature extraction module: Extracts multi-dimensional feature parameters from pure signals, including time-domain features, frequency-domain features, and time-frequency-domain features. The specific calculation formula is as follows:

[0079] Temporal characteristics:

[0080] Peak value: , These are pure signal sample values. , This represents the number of sampling points;

[0081] kurtosis: ;

[0082] Pulse factor: ;

[0083] Waveform factor: ;

[0084] Frequency domain characteristics:

[0085] Center frequency: , The result is the Fourier transform of the signal. For the first Frequency values ​​of each frequency point Number of frequency points;

[0086] Spectral entropy: ,in Normalized spectral probability;

[0087] Energy percentage of characteristic frequency bands: ,in The frequency band is sensitive to ice and snow characteristics, with values ​​ranging from 50 to 500 Hz;

[0088] Time-frequency domain characteristics:

[0089] Wavelet packet energy entropy: ,in , For the first Each small wavelet carries energy. This represents the number of subbands corresponding to the wavelet packet decomposition layer number.

[0090] Intelligent recognition module: It adopts a recognition model based on improved support vector machine (SVM); the model training samples include three types of signal samples: normal road surface, dark ice road surface, and dark snow road surface (4500 sets in this embodiment). In some embodiments, they can be divided into training set (3150 sets) and test set (1350 sets) in a 7:3 ratio. The penalty parameter C=10 and kernel function parameter γ=0.1 are optimized by grid search, and the feature weight factor is obtained by particle swarm optimization algorithm; the number of model training iterations is set to 300 times; after training, the recognition accuracy of the test set reaches 97.8%.

[0091] like Figure 4 As shown, by introducing a kernel function optimization strategy to improve feature discrimination, accurate identification of road surface conditions is achieved with an accuracy rate of ≥97%.

[0092] Improved SVM kernel function: A radial basis function (RBF) kernel is selected, and feature weight factors are introduced. The kernel function expression is as follows: ;

[0093] in, For the input feature vector, For kernel function parameters, For feature dimension, For the first The weights of each feature (obtained through particle swarm optimization algorithm) );

[0094] Classification decision function: ;

[0095] in, For the Lagrange multipliers of the support vector machine, The sample labels are: -1 represents normal road surface, 0 represents dark snow road surface, and 1 represents dark ice road surface. For bias terms, This represents the number of support vectors.

[0096] Precise positioning module: Based on the principle of optical time domain reflectance (OTDR), combined with the fiber optic laying topology map and GIS map, it calculates the precise location of black ice and black snow, with a positioning accuracy of ±0.5m;

[0097] The core formula for positioning is: ;

[0098] in, This represents the distance from the potential hazard point to the starting end of the optical fiber. The speed of light in a vacuum The round-trip time difference of the laser transmission. The refractive index of the sensing fiber;

[0099] The topology calibration formula is: ;

[0100] in, For the final positioning distance, This is the initial positioning distance. This is a topology correction factor (calibrated by fiber optic laying error). ).

[0101] Database module: Stores raw signals, noise-reduced signals, feature parameters, road surface condition recognition results, location information, and road maintenance records (such as road material, maintenance records, historical ice and snow hazard points, etc.).

[0102] This embodiment also includes an early warning and linkage subsystem, which is used to realize graded early warning and linkage response for black ice and snow on the road surface, including a risk classification module, a multi-channel early warning module and a linkage response module.

[0103] Risk classification module: Based on the identification results, three levels of warning are divided: normal road surface (green), dark snow road surface (yellow warning), and dark ice road surface (red warning);

[0104] Multi-channel early warning module: Yellow warnings are pushed to maintenance personnel and traffic management departments via pop-up windows and SMS on the road maintenance platform; Red warnings simultaneously trigger roadside sound and light alarms and variable message signs (displaying "Black ice ahead, slow down"), and push warning information to passing vehicles via navigation APP.

[0105] Joint response module: Links with high-definition cameras along the road to automatically capture images of the potential hazard location, assisting staff in confirming the road surface condition; at the same time, it generates a maintenance and response work order, including the location of the hazard, the risk level, and response suggestions (such as spreading de-icing agents, temporary traffic control, etc.).

[0106] Example 2

[0107] A method for detecting and warning of black ice and snow on road surfaces based on distributed acoustic sensing, such as Figure 2 As shown, it includes the following steps:

[0108] Sensor deployment and system initialization: Based on the road design drawings, complete the laying and fixing of the double-layer sensor optical fiber buried under the road surface and bound to the roadside guardrail;

[0109] Install the DAS host, calibration module, and transmission equipment to build a data processing and intelligent analysis center; configure system parameters: pulsed laser source wavelength 1550nm, pulse width 20ns, repetition frequency 50kHz, fiber optic laying spacing 0.5m, and EMD decomposition termination condition is that the correlation coefficient of adjacent intrinsic mode components is ≤0.1; import road maintenance ledger data and complete system self-test and initialization.

[0110] Full-domain vibration signal acquisition and calibration: The DAS host is started, and the pulsed laser source emits a narrow pulse laser signal, which is injected into the sensing fiber through the fiber coupler. When the laser is transmitted in the fiber, it is affected by vehicle movement and road vibration, resulting in backscattered light. The scattered light is transmitted to the photodetector through the coupler and converted into an electrical signal. The signal conditioning unit amplifies, filters (passband frequency 10Hz-15kHz) and performs analog-to-digital conversion on the electrical signal to generate a digital signal. The calibration module sends a calibration signal through a standard vibration source every half month, and combines the data from the temperature sensor (measurement range -40℃-85℃, accuracy ±0.2℃) and humidity sensor (measurement range 0-100%RH, accuracy ±2%RH) to correct the acquired data.

[0111] Multi-source signal noise reduction processing: After the data receiving module receives and decrypts the digital signal, the multi-source noise reduction module processes it: the mixed signal is decomposed by the EMD algorithm to remove the interfering intrinsic mode components; then the remaining components are processed by the adaptive noise cancellation algorithm to obtain a clean signal containing road surface condition characteristics.

[0112] Ice and snow feature extraction and state recognition: The ice and snow feature extraction module extracts multi-dimensional feature parameters in the time domain, frequency domain, and time-frequency domain from the clean signal; the feature parameters are input into the trained improved SVM model, and the road surface state (normal, dark ice, dark snow) and confidence level are output; if the confidence level is ≥97%, it is determined to be a valid recognition result and the subsequent positioning process is triggered; otherwise, it is determined to be an interference signal and the signal acquisition step is returned.

[0113] Accurate Hazard Location and Risk Classification: The accurate location module is based on the OTDR principle and calculates the location coordinates of black ice and black snow through the location formula, and obtains the road station number and specific road section information by associating with the GIS map; the risk classification module classifies the warning level according to the identification results.

[0114] Tiered early warning and coordinated response: The multi-channel early warning module pushes hazard information according to the early warning level; the coordinated response module links with cameras to capture on-site images and generates operation and maintenance response work orders; maintenance personnel formulate and execute response plans based on early warning information and images; after the response is completed, the results are entered into the database, forming a closed-loop management of "monitoring-identification-early warning-response-recording".

[0115] Model Iteration and Optimization: Newly collected road surface condition signal samples and handling records are regularly entered into the database to iteratively train the improved SVM model, continuously improving the recognition accuracy and early warning precision of black ice and black snow.

[0116] In some embodiments, the system was tested for one winter (3 months), and a total of 18 hidden snow hazards and 7 hidden ice hazards were detected. All hazards were accurately located using the positioning formula and calibration formula, with a positioning error of ≤0.5m. The early warning response time was ≤5s, and the average handling time of maintenance personnel was shortened by 70% compared with traditional methods. During the trial operation, no traffic accidents caused by hidden ice or hidden snow occurred on this section of road, verifying the effectiveness and practicality of the system provided in this embodiment.

[0117] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of one or more computer-usable storage media (including, but not limited to, disk storage, etc.) containing computer-usable program code. CD - ROM It takes the form of a computer program product implemented on (such as optical memory, etc.).

[0118] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0119] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0120] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0121] The above description is merely a preferred embodiment of the present invention and is not intended to limit the invention. Various modifications and variations can be made to the present invention by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made by those skilled in the art without creative effort within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A distributed acoustic sensing based pavement black ice detection and warning system, characterized in that, It includes a distributed acoustic sensing subsystem, a signal transmission subsystem, and a data processing and intelligent analysis center, among which: The distributed acoustic sensing subsystem includes buried fiber optic units, guardrail fiber optic units, a host unit, and fixing components. The buried fiber optic units include multiple sets of sensing fibers buried under the road surface; the guardrail fiber optic units include multiple sets of sensing fibers installed on the roadside guardrails; the host unit is used to transmit pulsed laser signals to the sensing fibers and receive feedback detection signals; and the fixing components are used to fix the sensing fibers. The buried fiber optic units include multiple sets of sensing fibers buried parallel to each other at a certain distance below the roadway, with a certain spacing between them, and the buried range covers the entire roadway. The guardrail fiber optic units include multiple sets of sensing fibers spirally bound from top to bottom along both sides of the roadside guardrails. Through the buried fiber optic units and the guardrail fiber optic units, a two-layer monitoring network of road surface and roadside is formed. The signal transmission subsystem is used to transmit the detection signals acquired by the distributed acoustic sensing subsystem to the data processing and intelligent analysis center. The data processing and intelligent analysis center includes a multi-source noise reduction module, an ice and snow feature extraction module, and an intelligent recognition module. These modules are used to preprocess the received detection signals, decompose the signals, remove interfering intrinsic mode components, process the remaining components using an adaptive noise cancellation algorithm, obtain a clean signal containing road surface condition features, extract time-domain, frequency-domain, and time-frequency-domain features from the clean signal, process the extracted features using a pre-trained recognition model, identify the road surface condition, and issue warnings based on the recognition results. The multi-source noise reduction module is used to decompose the mixed signal into several intrinsic mode components through empirical mode decomposition and remove interference components; taking the traffic vibration signal during the snow-free period as a reference, the module uses an adaptive noise cancellation algorithm to filter out redundant interference and extract the pure road surface condition feature signal. The ice and snow feature extraction module is used to extract multi-dimensional feature parameters from pure signals, including time-domain features, frequency-domain features, and time-frequency-domain features. The intelligent recognition module is used to process the extracted features using a pre-trained recognition model to identify the road surface condition. The recognition model is a support vector machine model with radial basis kernel function and feature weight factor introduced.

2. The distributed acoustic sensing based pavement black ice detection and warning system of claim 1, wherein, The host includes a pulsed laser source, an optical fiber coupler, a photodetector, and a signal conditioning unit. The pulsed laser source selects a set communication band and emits laser signals at a set pulse width and repetition frequency. The emitted laser signal passes through the sensing optical fiber of the optical fiber coupler. When the laser signal is transmitted in the optical fiber, it is affected by vehicle movement and road vibration, resulting in backscattered light. The backscattered light is transmitted to the photodetector via the coupler and converted into an electrical signal. The signal conditioning unit is used to amplify, filter, and perform analog-to-digital conversion on electrical signals to generate digital signals; The host also includes a calibration module, which is used to send a calibration signal through a standard vibration source according to a set period, and correct the detection data in combination with ambient temperature and humidity data.

3. The distributed acoustic sensing based pavement black ice detection and warning system of claim 1, wherein, The fixing components include a corrosion-resistant sleeve, a buffer rubber pad, and a buckle. The embedded sensing optical fiber is fitted with a corrosion-resistant sleeve, and a buffer rubber pad is filled between the corrosion-resistant sleeve and the sensing optical fiber. The sensing fiber optic cables attached to the guardrail are secured with clips, with a certain distance between adjacent clips.

4. The distributed acoustic sensing based pavement black ice detection and warning system of claim 1, wherein, The data processing and intelligent analysis center also includes a data receiving module, a precise positioning module, and a database module, wherein: The data receiving module is used to receive detection data sent by the transmission subsystem; The precise positioning module is used to calculate the precise location of dark ice and / or dark snow based on the principle of optical time domain reflection, combined with the topology map of the sensor fiber optic cable laying and the GIS map. The database module is used to store the original detection signals, the noise-reduced signals, features, road surface condition recognition results, location information, and road maintenance records.

5. A method for detecting and warning of dark ice on a road surface based on distributed acoustic sensing, based on the system of any one of claims 1-4, characterized in that, Includes the following steps: Acquire the detection signals from the distributed acoustic sensing subsystem; The detected signal is preprocessed, decomposed, and interference intrinsic mode components are removed. The remaining components are then processed by an adaptive noise cancellation algorithm to obtain a clean signal containing road surface characteristics. Time-domain, frequency-domain, and time-frequency-domain features are extracted from the clean signal. The extracted features are then processed using a pre-trained recognition model to identify the road surface condition and issue warnings based on the recognition results.

6. The method for detecting and warning of black ice on road surfaces based on distributed acoustic sensing as described in claim 5, characterized in that, The process of decomposing the signal, removing interfering intrinsic mode components, and processing the remaining components using an adaptive noise cancellation algorithm includes: First, the mixed signal is decomposed into several intrinsic mode components through empirical mode decomposition to remove interference components; then, using traffic vibration signals during periods without ice or snow as a reference, redundant interference is filtered out through an adaptive noise cancellation algorithm to extract pure road surface condition feature signals. Among them, the empirical mode decomposition screening condition is that the correlation coefficient between two adjacent intrinsic mode components must satisfy... , The modal component number; The objective function of the adaptive noise cancellation algorithm is: ; in, Let be the error cost function. The original signal contains noise. For reference noise signal, This is the adaptive filter weight vector. This represents the expectation operation; the optimal weights are solved using the least squares method. ( () is the autocorrelation matrix of the reference signal. This is the cross-correlation vector between the reference signal and the original signal.

7. The method for detecting and warning of black ice on road surfaces based on distributed acoustic sensing as described in claim 5, characterized in that, The process of extracting time-domain, frequency-domain, and time-frequency-domain features from a clean signal includes: time-domain features include peak value, kurtosis, impulse factor, and waveform factor extracted from the clean signal; frequency-domain features include center frequency, spectral entropy, and characteristic frequency band energy percentage extracted from the clean signal; and time-frequency-domain features include wavelet packet energy entropy extracted from the clean signal.

8. The method for detecting and warning of black ice on the road surface based on distributed acoustic sensing as described in claim 5, characterized in that, The process of using a pre-trained recognition model to process extracted features, identify road surface conditions, and issue warnings based on the recognition results includes: using a recognition model based on an improved support vector machine; the training samples of the recognition model include signal samples of different types of road surface conditions such as normal road surface, dark ice road surface, and dark snow road surface; and improving feature discrimination by introducing a kernel function optimization strategy. Improvements to support vector machines include using a radial basis function kernel and introducing feature weight factors. The kernel function expression is as follows: ; in, For the input feature vector, For kernel function parameters, For feature dimension, For the first The weights of each feature are obtained through optimization using the particle swarm optimization algorithm. ; The classification decision function of the recognition model is: ; in, For the Lagrange multipliers of the support vector machine, The sample labels are -1 for normal road surfaces, 0 for dark snow surfaces, and 1 for dark ice surfaces. For bias terms, This represents the number of support vectors.

9. The method for detecting and warning of black ice on road surfaces based on distributed acoustic sensing as described in claim 5, characterized in that, When the identification model detects dark ice and dark snow, it calculates the precise location of the dark ice and dark snow based on the principle of optical time-domain reflectometry, combined with the laying topology map of the sensing optical fiber and the GIS map. The positioning process is as follows: ; in, This represents the distance from the potential hazard point to the starting end of the optical fiber. The speed of light in a vacuum The round-trip time difference of the laser transmission. The refractive index of the sensing fiber; And perform topology calibration on the positioning results: ; in, For the final positioning distance, This is the initial positioning distance. This is a topology correction value, calibrated by fiber optic laying error. rice.

Images