Method, device and system for monitoring road traffic and road conditions by using seismic data

Abstract

A method, apparatus, and system for monitoring traffic and road conditions using seismic data are provided. A method of monitoring traffic flow using seismic data may include: acquiring seismic data from a seismic recording device; obtaining a seismic data fluctuation graph based on the seismic data, wherein the seismic data fluctuation graph comprises a plurality of seismic data fluctuation curves of each seismic recording device; and monitoring the traffic flow based on the seismic data fluctuation map.

Description

Technical Field

[0001] The present invention generally relates to the field of data processing technology, and specifically includes a method, system, electronic device for traffic monitoring and road condition management, as well as a non-transitory computer-readable storage medium containing storage instructions and a computer program product. The invention can use the collected data to monitor and predict traffic dynamic changes and traffic accidents in real time, and conduct comprehensive analysis in combination with external information (such as seismic wave data) to improve the accuracy of prediction and response capabilities. Background Art

[0002] Road traffic accidents are one of the leading causes of death worldwide. The World Health Organization (WHO) estimates that approximately 1.3 million people die each year in traffic accidents. According to the Centers for Disease Control and Prevention (CDC), in 2019, the traffic accident death rate in rural areas was 3 to 10 times that in urban areas, and rural areas accounted for approximately 45% of the national total. Since accidents on these roads often take a long time to be reported, it is very beneficial to use traffic detection systems and methods to monitor these roads. Monitoring traffic flow on multiple roads or entire cities is essential for traffic management departments to identify any traffic problems in real time. With the advent of autonomous vehicles, the need for real-time traffic flow information has become more urgent. Traffic flow information provides opportunities for lane traffic density management systems and methods that can automatically optimize the spacing between autonomous vehicles and between other autonomous vehicles and / or non-autonomous vehicles on the road. More specifically, the operation of the autonomous vehicle can be automatically controlled to maintain an optimal spacing relative to the vehicles in front and behind.

[0003] Currently, there are many methods for vehicle detection, including video image processing, radar, and ultrasonic sensors. Deploying a large number of camera systems on the roadside can provide detailed traffic flow and individual vehicle information. However, using a large number of cameras to monitor traffic and road conditions faces many problems, such as:

[0004] 1. High cost: Rural roads cover long distances, which makes these systems difficult to implement in practice. Each traffic camera usually costs tens of thousands of dollars, and covering a wide distance requires a large number of such cameras or sensors, which is costly.

[0005] 2. Real-time data processing requirements: Quickly understanding the current traffic situation requires access to a large amount of video data and extracting useful information from it. This not only requires significant real-time computing power, but also requires expensive equipment, increasing the overall cost of the system.

[0006] 3. Privacy regulations: Many regions and countries have introduced privacy regulations that restrict or prohibit the use of video recording in public places. This poses legal and ethical challenges to the widespread use of cameras.

[0007] In addition, the mobile phone positioning services currently used in many navigation systems may not provide accurate traffic information. Therefore, it is necessary to develop a new traffic and road monitoring method to solve the above technical problems. Summary of the invention

[0008] The first aspect of the present invention application proposes a method for monitoring traffic flow using seismic data, which method may include the following steps:

[0009] ·Acquire seismic data from seismic recording equipment.

[0010] Generate a seismic data fluctuation map based on the acquired seismic data, the map containing a plurality of seismic data fluctuation curves of each seismic recording device.

[0011] Monitoring traffic flow based on earthquake data wave maps.

[0012] In this method, seismic recording devices may be arranged at fixed or variable intervals on one side, on both sides, in the middle, or any combination thereof of the road.

[0013] Additionally, the process of generating a wave map of seismic data may include:

[0014] ·Signal enhancement for seismic data.

[0015] After signal enhancement, perform two-way traffic wave field separation or attenuation.

[0016] The seismic data are balanced using the calibration curve and a seismic data fluctuation map is generated based on the balanced seismic data.

[0017] The method may further include:

[0018] Obtain vehicle velocity spectrum based on seismic data wave diagram to intuitively display the vehicle's motion status.

[0019] Determine vehicle speed and / or vehicle movement path by analyzing seismic data similarities between different seismic recording devices.

[0020] ·Determine whether the vehicle is speeding based on the vehicle speed.

[0021] Determine the type and weight of the vehicle by analyzing the peaks of the vehicle in the fluctuation curves of multiple seismic data.

[0022] Determine whether the vehicle is parked based on the vehicle's movement path.

[0023] Based on earthquake data, determine whether there are pedestrians walking on the road.

[0024] Detect if objects have fallen from the vehicle through seismic data.

[0025] A second aspect of the present invention application provides a system for monitoring traffic flow using seismic data, the system may include:

[0026] · Seismic recording equipment configured to acquire seismic data;

[0027] a data processing module configured to generate a seismic data fluctuation map based on the seismic data, wherein the seismic data fluctuation map includes a plurality of seismic data fluctuation curves for each seismic recording device;

[0028] An analysis module configured to analyze traffic flow based on seismic data wave patterns.

[0029] According to the second aspect, the analysis module thereof can also perform the method of the first aspect.

[0030] A third aspect of the present invention application provides a system for monitoring traffic flow by further utilizing processed seismic data, the system may include:

[0031] · Seismic recording equipment configured to acquire seismic data;

[0032] ·processor;

[0033] A memory coupled to the processor having stored therein instructions that, when executed by the processor, enable the processor to:

[0034] o obtaining a seismic data fluctuation map based on the seismic data, wherein the seismic data fluctuation map includes a plurality of seismic data fluctuation curves for each seismic recording device;

[0035] οMonitoring traffic flow based on seismic data wave maps.

[0036] A fourth aspect of the present invention provides a method for training a vehicle recognition model, which may include:

[0037] Get the first image from the video that contains the vehicle;

[0038] Transform the first image into a second image that represents the vehicle and the lane it is traveling in, where the second image is used as the training label;

[0039] Obtaining the first seismic data corresponding to the first image from the seismic data;

[0040] Train a vehicle recognition model based on the first seismic data and the second image.

[0041] A fifth aspect of the present invention application provides a vehicle identification method, which may include:

[0042] Input the seismic data to be identified into a vehicle identification model to determine the vehicle and the lane it is in, wherein the vehicle identification model is trained using the method of the fourth aspect.

[0043] A sixth aspect of the present invention application provides a method for training a vehicle recognition model, which may include:

[0044] Select earthquake data in multiple time windows as labels;

[0045] Identify the maximum peak or Gaussian distribution peak for each tag;

[0046] Determine the vehicle position corresponding to the maximum peak or the peak of the Gaussian distribution;

[0047] and training a vehicle recognition model based on the location and the corresponding maximum peak or Gaussian distribution peak.

[0048] A seventh aspect of the present application provides a vehicle identification method, which may include:

[0049] Input the seismic data to be identified into the vehicle identification model to determine the location of the vehicle;

[0050] and determining the speed of the vehicle and / or the moving path of the vehicle based on the position, wherein the vehicle recognition model is trained using the method of the sixth aspect.

[0051] An eighth aspect of the present application provides a method for monitoring geological conditions under a road using seismic data, the method may include:

[0052] Obtain target seismic data and reference seismic data from seismic recording equipment;

[0053] Generate a target Green's function based on the target seismic data; generate a reference Green's function based on the reference seismic data;

[0054] Generate near-surface relative velocity changes based on target Green’s function and reference Green’s function to monitor the geological conditions beneath the road.

[0055] According to the eighth aspect, the method may further include:

[0056] ·Perform signal preprocessing, noise removal, resampling, data displacement correction and filtering on target seismic data and reference seismic data.

[0057] According to an eighth aspect, wherein the filtering includes multiple frequency bands (f 1i ,f 2i ) bandpass filter, where f 0 <f 1i <f 1;f 0 <f 2i <f 1 ;f 0 and f 1 They are respectively the lower and upper limits of the frequency range of the seismic recording equipment.

[0058] According to the eighth aspect, the method may further include:

[0059] Extract one or more body waves, elastic P waves, S waves, SH waves, surface waves, coda waves, Rayleigh waves and Love waves from each target seismic data and reference seismic data;

[0060] and generating reference Green's function and target Green's function based on the extracted waves.

[0061] According to an eighth aspect, generating a near-surface relative velocity change may include:

[0062] Generate near-surface relative velocity variations based on target and reference Green's functions by using an ambient noise imaging method.

[0063] A ninth aspect of the present application provides a method for monitoring road surface conditions using seismic data, which may include:

[0064] Obtain target seismic data and reference seismic data from seismic recording equipment;

[0065] Calculate the reference total energy of a vehicle passing through a certain frequency range within a reference time based on the reference seismic data;

[0066] Calculate the target total energy of the vehicle passing through the same frequency range within the target time based on the target seismic data; and

[0067] Monitoring road surface conditions based on reference total energy and target total energy.

[0068] According to the ninth aspect, the calculation of the reference total energy and the calculation of the target total energy are performed in the frequency domain or the time domain.

[0069] According to a ninth aspect, calculation of the reference total energy and calculation of the target total energy are performed in a plurality of frequency bands.

[0070] According to a ninth aspect, monitoring the road surface condition based on the reference total energy and the target total energy may include:

[0071] Calculating a reference ratio between reference total energies in a plurality of frequency bands, and calculating a target ratio between target total energies in a plurality of frequency bands;

[0072] · Compare the reference ratio and target ratio and perform weighted summation or subtraction.

[0073] A tenth aspect of the present application provides a method for predicting a traffic accident, which may include:

[0074] Obtain seismic data from seismic recording equipment;

[0075] Obtain information about the vehicle based on seismic data;

[0076] Predicting traffic accidents based on vehicle information, current and historical road condition information, driver information and weather. The eleventh aspect of the present application provides a method for training a traffic accident prediction model, which may include:

[0077] Obtain earthquake data, vehicle information, road information, driver information and weather information at the time of the accident;

[0078] ·Training a traffic accident prediction model based on the earthquake data and the information.

[0079] A twelfth aspect of the present application provides a method for predicting a traffic accident, which may include:

[0080] Inputting the earthquake data to be identified into a traffic accident prediction model to predict traffic accidents, thereby obtaining a traffic accident score, wherein the traffic accident prediction model is trained using the method according to the eleventh aspect.

[0081] A thirteenth aspect of the present application provides a method for training a vehicle weight model, which may include:

[0082] Extracting peak values ​​of vehicles from seismic data; and

[0083] A vehicle weight prediction model is trained based on the peak value and the vehicle weight.

[0084] A fourteenth aspect of the present application provides a method for obtaining vehicle weight, which may include:

[0085] Inputting the earthquake data to be identified into a vehicle weight model to obtain the weight of the vehicle, wherein the vehicle weight model is trained using the method according to the thirteenth aspect.

[0086] A fifteenth aspect of the present application provides an electronic device, which may include:

[0087] at least one processor; and

[0088] a memory communicatively coupled to at least one processor,

[0089] Wherein, the memory stores instructions, and when the instructions are executed by at least one processor, the at least one processor executes the above method.

[0090] The sixteenth aspect of the present application provides a non-temporary computer-readable storage medium, which stores instructions that, when executed on a computer, cause the computer to perform the above method.

[0091] A seventeenth aspect of the present application provides a computer program product, comprising a computer program having instructions. When these instructions are executed by a computer, the computer is caused to perform the above method. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] · Figure 1 is a schematic diagram of a system architecture for executing a traffic and road monitoring method according to an exemplary embodiment of the present disclosure;

[0093] · Figure 2 is a schematic flow chart illustrating a method for monitoring traffic flow according to an exemplary embodiment of the present disclosure;

[0094] · Figure 3 Schematic diagram showing several example designs for placing seismic sensor recording locations along a roadway to monitor traffic;

[0095] · Figure 4 A schematic flow chart for processing seismic data is shown;

[0096] · Figures 5A-5C Graph showing amplitude correction for face consistency;

[0097] · Figure 6 The graph after bandpass signal processing is shown;

[0098] · Figure 7 shows the vehicle speed spectrum generated from the bandpass signal processing data;

[0099] · Figure 8 A graph showing the generation of vehicle speed spectra from signal-enhanced traffic data;

[0100] · Fig. 9 The vehicle paths and the vehicle speeds extracted from the speed spectra are shown;

[0101] · Fig.10 A graph for determining whether a vehicle is speeding is shown;

[0102] · Fig.11 An interpretation of the relationship between vehicle type, speed, and weight and earthquake amplitude is shown;

[0103] · Fig.12 A diagram illustrating vehicle stopping is shown;

[0104] · Fig.13 It illustrates the path of a single vehicle moving along the road (shown as a curve);

[0105] · Fig.14 The typical seismic response of a vehicle is shown;

[0106] · Fig.15 The seismic response of a person walking on the highway is given (as shown in the area);

[0107] · Fig.16 The extrapolation of the velocity spectrum in the spatial domain is shown;

[0108] · Fig.17 is a block diagram of a system 1700 for monitoring traffic flow using seismic data according to an embodiment;

[0109] · Fig.18 shows the process of converting from a first video image to a second two-color image, wherein (a) a video recording of a road is shown and vehicles are detected using image / video recognition techniques, and (b) training labels are generated using vehicles detected from the video recording;

[0110] · Fig.19 The use of machine learning for vehicle detection and interpretation is shown;

[0111] · Fig. 20 It shows that the direct Fig. 20 (a)) or ML-curve( Fig. 20 (b)) The vehicle path obtained;

[0112] · Fig.21 is a schematic flow chart illustrating a method for monitoring geological conditions under a road using seismic data according to an example embodiment of the present disclosure;

[0113] · Fig. 22 Reference Green's functions between pairs of sites (seismic sensors) are shown;

[0114] · Fig.23 A Green's function comparison between the reference and target times is shown;

[0115] · Fig.24 An image showing the relative changes in earthquake velocities under the road;

[0116] · Fig.25 is a schematic flow chart illustrating a method for monitoring road surface conditions using seismic data according to an exemplary embodiment of the present disclosure;

[0117] · Fig.26 The changes in the road surface characteristic response over time are shown;

[0118] · Fig. 27 is a schematic flow chart illustrating a method for predicting a traffic accident according to an exemplary embodiment of the present disclosure; and

[0119] · Fig.28 An electronic device according to an example embodiment of the present disclosure is shown. DETAILED DESCRIPTION

[0120] Below, various example embodiments of the present disclosure will be described with reference to the accompanying drawings. In the following description, specific details such as detailed configuration and components are provided only to help the overall understanding of these embodiments of the present disclosure. Therefore, it should be understood by those skilled in the art that various changes and modifications can be made to the embodiments described herein without departing from the scope and spirit of the present disclosure. In addition, for the sake of clarity and brevity, the description of well-known functions and structures is omitted.

[0121] These example embodiments and the terms used therein are not intended to limit the technology disclosed herein to a specific form, and should be understood to include various modifications, equivalents and / or alternatives of the corresponding embodiments. When describing the drawings, similar reference numerals may be used to refer to similar constituent elements. Unless clearly different in context, singular expressions may include plural expressions.

[0122] Furthermore, the various embodiments described herein are not necessarily mutually exclusive, as some embodiments can be combined with one or more other embodiments to form new embodiments.

[0123] Unless the context clearly indicates, the singular form used herein may also include the plural form. The expressions "first", "second", "the first" or "the second" used in various embodiments of the present disclosure can modify various components without considering order and / or importance, but do not limit the corresponding components. When an element (e.g., a first element) is referred to as "connected (functionally or in communication)" or "directly coupled" to another element (the second element), the element can be directly connected to another element, or connected to another element by another element (e.g., a third element).

[0124] The expression "configured to" used in various embodiments of the present disclosure may be used interchangeably with expressions such as "suitable for", "capable of", "designed to", "modified to", "made to", or "capable of", in terms of hardware or software, depending on the situation. In some cases, the expression "device configured to" may refer to, for example, a situation where the device is "capable of" together with other devices or components. For example, the phrase "a processor modified (or configured) to perform A, B, and C" may refer to (but not limited to) a dedicated processor (such as an embedded processor) for performing the corresponding operations, or a general-purpose processor (such as a central processing unit (CPU) or an application processor (AP)) that performs the corresponding operations by executing one or more software programs stored in a memory device.

[0125] Here, unless otherwise specified, the term "or" used herein refers to a non-exclusive "or". The examples used herein are only intended to facilitate the understanding of the practical mode of the embodiments of the present invention and to further enable those skilled in the art to practice the embodiments of the present invention. Therefore, these examples should not be interpreted as limiting the scope of the embodiments of the present invention.

[0126] According to the tradition of this field, the implementation method can be described and illustrated with blocks that perform the described functions. These blocks may be referred to as units, engines, managers, modules or similar names in this article, and they are physically implemented by analog and / or digital circuits, such as logic gates, integrated circuits, microprocessors, microcontrollers, memory circuits, passive electronic components, active electronic components, optical components, hard-wired circuits, etc., and can be selectively driven by firmware and / or software. For example, these circuits can be embodied in one or more semiconductor chips, or on substrate supports such as printed circuit boards. The circuits constituting a block can be implemented by dedicated hardware, or by a processor (for example, one or more programmed microprocessors and their associated circuits), or by a combination of dedicated hardware executing certain functions of the block and a processor executing other functions of the block. Each block of the implementation method can be physically separated into two or more discrete blocks that interact with each other without departing from the scope of this disclosure. Similarly, the blocks of the implementation method can be physically combined into more complex blocks without departing from the scope of this disclosure.

[0127] When expressions such as "at least one of A, B, and C, etc." are used, they should generally be interpreted according to the meaning commonly understood by those skilled in the art (for example, "a system comprising at least one of A, B, and C" should include but is not limited to a system comprising A, B, C, A and B, A and C, B and C, and / or A, B, C, etc.).

[0128] In earthquake monitoring, seismic sensors are installed to record elastic waves propagating from earthquake sources. They are usually designed to monitor large areas on the earth, and the sensor locations may be irregular due to the limitations of the installation site. Each sensor records a time series of ground vibrations at a uniform sampling interval, and the signal can be recorded in the form of displacement, velocity (velocity sensor) or acceleration (accelerometer) of the ground vibration. Through numerical calculations, these three forms of data are roughly interchangeable. Seismic sensors are designed to record ground vibrations within a specific frequency range, subject to the limitations of hardware and electronics. These records can be used to infer the location of the earthquake source, the magnitude, the focal mechanism, and the medium velocity between the source and the receiver. The sensor can include a single vertical component that records vertical ground vibrations, or three components that record ground vibrations along three orthogonal directions. A single three-component sensor is able to determine the direction of the incident earthquake P or S wave under simplifying assumptions.

[0129] Seismic research is not limited to earthquake problems, but is also widely used to help find underground oil, gas and other resources. In seismic exploration for minerals, coal, or oil and gas, seismic surveys use one or more controlled sources and one or more receivers to collect information about the phase and amplitude of seismic waves. Sources and receivers used in seismic surveys can be placed regularly or irregularly on the earth's surface along a line (for two-dimensional imaging) or within an area (for three-dimensional imaging) according to a designed geometry. In these surveys, sources and receivers are set at known locations. The velocity medium or rock interface is the target of imaging.

[0130] Seismic data may include a large number of seismic traces. Each sensor records a trace for one vertical component or three traces for three components. Seismic traces include data on ground motion over a period of time, which are recorded at receivers, which may be seismometers, hydrophones, micro-electromechanical systems (MEMS) sensors, distributed acoustic sensing (DAS), or other seismic monitoring devices. Receivers may be deployed at approximately constant intervals along the surface. While ground motion may be collected passively, seismic surveys may use a source to create seismic waves from a known origin. The source is the location where seismic waves originate. A shot is the release of energy at the source to create seismic waves, such as a single dynamite explosion or vibrator. The location where the shot occurs is called the source. A single shot may be made at the source to create seismic waves. Multiple shots may be made at different sources and times.

[0131] Seismometers are installed on the ground to record vibrations caused by earthquakes or controlled sources. Analysis of this data can reveal information about the source of the earthquake or the earth medium through which the seismic waves propagate. Seismic sensors can also be installed on the roadside to monitor ground motion caused by passing vehicles acting as passive sources. Analyzing this data with designed algorithms and methods can help reveal the speed, weight, spacing, location, and driving pattern of all vehicles along the entire road monitored by the seismic sensors. This data can also be used to monitor daily changes in road or bridge conditions for risk assessment. These results can help minimize the occurrence of accidents or disasters and help promote traffic control and safe driving at a lower cost.

[0132] Design of a traffic flow reconstruction process using seismic data may include one or more of the following steps:

[0133] 1) Receive traffic vibration data recorded on site and transmit it to the computer in real time;

[0134] 2) Data signal enhancement through digital processing;

[0135] 3) Two-way traffic wave field separation or amplitude attenuation processing;

[0136] 4) Ground surface consistency amplitude correction;

[0137] 5) Extract amplitude and frequency attributes to determine vehicle weight and type;

[0138] 6) scanning the vehicle speed to generate a speed spectrum;

[0139] 7) Use machine learning methods to detect vehicles and output vehicle labels with pulses or Gaussian signals on time data at the time of vehicle arrival;

[0140] 8) Correlating the pulses or Gaussian signals obtained in step (7) to generate a spatiotemporal trajectory curve reflecting the vehicle traveling along the road;

[0141] 9) Inferring driving behavior from the vehicle's spatiotemporal trajectory curve and lane position; and

[0142] 10) Reconstruct real-time traffic flow including vehicle type, weight, speed and position (location and lane).

[0143] One embodiment of the present application provides a method for monitoring traffic flow using seismic data, thereby reducing system layout costs, reducing the amount of data to be processed, protecting personal privacy, and having high timeliness. The method may include: acquiring seismic data from a seismic recording device; acquiring a seismic data fluctuation graph based on the seismic data, wherein the seismic data fluctuation graph includes multiple seismic data fluctuation curves for each seismic recording device; and monitoring traffic flow based on the seismic data fluctuation graph.

[0144] It is relatively simple to extract information such as the speed, weight, number of axles, position, lane position, and driving mode of all vehicles from ground vibration data, and the amount of data is several orders of magnitude smaller than video data. The results of vibration data analysis can provide a global view of all vehicles on the road or in the city, and help the camera system focus on certain specific vehicles or specific roads with minimal cost and computing effort. The camera system is like a person's "eyes", while the auditory ability is like a person's "ears". The combination of "eyes" and "ears" makes the monitoring capabilities of many applications more powerful.

[0145] Referring now to the drawings, in particular Figures 1 to 28 , wherein like reference characters indicate corresponding features consistently throughout the various drawings, a preferred embodiment is shown.

[0146] Figure 1 is a schematic diagram of a system architecture for executing a traffic and road monitoring method according to an exemplary embodiment of the present disclosure. It should be noted that Figure 1 It is only an example of a system architecture to which the implementation methods disclosed in this application can be applied, and is used to help those skilled in the art understand the technical content disclosed herein, but it does not mean that the disclosed implementation methods cannot be used in other devices, systems, environments or scenarios.

[0147] like Figure 1 As shown, the system architecture 100 according to this embodiment may include seismic recording devices, such as seismic sensors 101, 102, 103, a network 104 and a server 105. The network 104 is a medium for providing a communication link between the seismic sensors 101, 102, 103 and the server 105. The network 104 may include various types of connections, such as wired, wireless communication links or optical fiber cables.

[0148] In the embodiment, seismic data can be transmitted via 5G / 4G cellular network, Wi-Fi or fiber optic Internet cable. Server 105 can be a server that provides various services. Server 105 can be a cloud server, also known as a cloud computing server or cloud host. Server 105 can also be a server for a distributed system or a server combined with blockchain technology.

[0149] It should be noted that the traffic and road monitoring method in this embodiment is usually executed by the server 105. Accordingly, the unit or module for executing the method of this embodiment can be provided in the server 105. The traffic and road monitoring method according to this embodiment can also be executed by a device, server or server cluster that is different from the server 105 but can communicate with it. Accordingly, the unit or module for executing the traffic and road monitoring method according to this embodiment can also be set in a device, server or server cluster that is different from the server 105 but can communicate with it. DETAILED DESCRIPTION

[0151] Traffic monitoring

[0152] Figure 2 FIG. 2 is a schematic flow chart 200 illustrating a method for monitoring traffic flow according to an exemplary embodiment of the present disclosure. Figure 2 As shown, the flowchart 200 may include the following operations:

[0153] In operation S210, seismic data may be acquired from a seismic recording device;

[0154] In operation S220, a seismic data fluctuation map may be obtained based on the seismic data, wherein the seismic data fluctuation map includes a plurality of seismic data fluctuation curves of each seismic recording device;

[0155] In operation S230, traffic flow may be monitored based on the seismic data wave pattern.

[0156] In operation S210 of the embodiment, the earthquake recording device may be a seismometer, a seismic sensor, or a ground vibration recording device. In the embodiment, the earthquake recording device may be arranged at fixed or variable intervals on one side of the road, on both sides of the road, or in the middle of the road, or any combination thereof. This will be referred to by reference to Figure 3For detailed information about operation S220, please refer to Figures 4 to 6 The seismic vibrations generated by vehicles moving along the roadside are recorded by seismic sensors, and the recorded seismic data can be used to reconstruct traffic flow and estimate moving vehicle information in real time. Through operation S230, monitoring traffic flow may include digital traffic flow reconstruction and detecting abnormal activities.

[0157] Figure 3 Schematic diagram showing several example designs of seismic sensor recording locations along a road for monitoring traffic. Figure 3 As shown, example arrangements of seismic sensors may include: a single-sided arrangement (e.g., Figure 3 The right side of the road shown in (a) is Figure 3 left side of the road as shown in (b)); middle lane layout (e.g. Figure 3 central island or divider as shown in (c)); bilateral arrangement, such as Figure 3 staggered arrangement on both sides of the road as shown in (d), and any combination of these.

[0158] Frequency Range of Vehicle Vibration Data - The frequency range of vibration data recorded by seismic sensors varies from 0.01 Hz to several kilohertz. Across the entire frequency spectrum of the recorded data, the frequency response of larger and heavier vehicles will differ from that of smaller and lighter vehicles. Therefore, it becomes critical to capture the full frequency range of ground vibrations caused by vehicles. In addition, using the full frequency range of recorded vehicle vibrations is essential to determine the condition of the road surface and the structure beneath the road.

[0159] Figure 4 A schematic flow chart 400 for processing seismic data is shown. Figure 4 As shown, the flowchart 400 may include the following steps:

[0160] Step S410: Signal enhancement can be performed on the seismic data;

[0161] Step S420: After the signal is enhanced, the two-way traffic wave field may be separated or attenuated;

[0162] Step S430, the seismic data may be balanced using the calibration curve;

[0163] Step S440: generating a seismic data fluctuation map based on the balanced seismic data.

[0164] In operation S410, signal enhancement may be performed on the seismic data.

[0165] Vehicle signal enhancement – ​​Seismic sensors record vehicle signals as well as noise in the same and / or different frequency bands, so denoising or signal enhancement is required. This may include applying machine learning (ML), bandpass filtering, median filtering, RMS filtering, automatic gain control (AGC), trace balancing, wavelet transform, stacking signals in moving time windows, extracting signal envelopes, transforming data using STA / LTA ratio (ratio of short time window to long time window), and convolution with wavelets.

[0166] In step S410, performing signal enhancement on the seismic data may include one or more of the following:

[0167] 1) Remove or attenuate noise;

[0168] 2) Remove or attenuate irregular signals other than vehicle vibration;

[0169] 3) Remove or attenuate the secondary response of the initial vehicle vibration;

[0170] 4) Apply mathematical logarithm operation to traffic data to normalize amplitude;

[0171] 5) Taking arbitrary exponents (including square roots) of traffic data to normalize the amplitude;

[0172] 6) Applying automatic gain control (AGC) to traffic data to normalize the amplitude;

[0173] 7) Apply root mean square (RMS) to traffic data to normalize amplitude;

[0174] 8) Apply median filtering to traffic data to smooth the data and remove outliers;

[0175] 9) Apply mean filtering to traffic data to smooth the data;

[0176] 10) Apply integral filtering to traffic data to amplify weak signals;

[0177] 11) Apply global normalization to traffic data;

[0178] 12) Apply single-station (sensor) normalization to traffic data;

[0179] 13) Apply local window normalization to traffic data;

[0180] 14) Apply the above 4)-13) items to the signal envelope of the traffic data;

[0181] 15) Apply bandpass filtering to traffic data or processed data;

[0182] 16) Apply linear motion correction (LMO) to the data, using a preset vehicle speed relative to the recording station;

[0183] 17) Apply any or all of the combinations of 1)–16).

[0184] In step S420, two-way traffic wave field separation or attenuation may be performed after signal enhancement. Two-way traffic wave field separation or attenuation may include:

[0185] Frequency-wavenumber domain filtering (FK filtering);

[0186] Perform FK filtering on pre-processed seismic data;

[0187] Apply attenuation to preprocessed seismic data;

[0188] Apply machine learning (ML) that takes in two-way traffic data and outputs one-way traffic data.

[0189] In step S430, the seismic data may be balanced using the surface consistent amplitude correction curve. Balancing the seismic data between recording stations equipped with seismic sensors may include:

[0190] Applying surface-consistent amplitude correction to traffic data;

[0191] Apply trace normalization to traffic data.

[0192] Surface Consistent Amplitude Correction - For the same vehicle passing different roadside receivers at the same speed, the amplitude response may vary due to variations in the road structure and receiver-ground coupling. This amplitude amplification or attenuation effect is site- and receiver-specific, and all vehicles passing the same site are affected in the same way. Therefore, it is surface consistent. The amplitude variation can be measured from an existing data set and a response curve interpolated to correct for the amplitude variation. To account for differences in response due to different vehicle weights, multiple correction curves can be generated for different vehicle weight groups.

[0193] Due to the differences in road and roadbed conditions between each site, the seismic data recorded by vehicle vibration need to be corrected.

[0194] Figures 5A-5C This correction process is shown in Figure 1. By multiplying the original recorded data with the station correction function / correction curve, the data after earthquake amplitude correction is obtained for further processing. The vertical axis represents the spatial position of the seismic sensor. The horizontal axis in the earthquake data graph ( Figure 5A and Figure 5C ) indicates the recording time, and Figure 5B The horizontal axis represents the site correction value.

[0195] Figure 5A The original data is shown. Figure 5B The calibration curve is shown. Figure 5CThe plots after surface-consistent amplitude correction are shown.

[0196] In step S440, a seismic data fluctuation map may be obtained based on the balanced seismic data.

[0197] In order to remove some background noise and highlight vehicle events, bandpass filtering of different frequency ranges is required. In one embodiment, after balancing the seismic data, bandpass signal processing can be performed.

[0198] Figure 6 The application of bandpass filtering to the amplitude-corrected data is shown. Figure 6 (a) shows the schematic diagram after the ground surface consistency amplitude correction. Figure 6 (b) in the figure shows a schematic diagram of bandpass signal processing. It can be seen that after bandpass signal processing, the seismic data fluctuation diagram becomes clearer.

[0199] In one embodiment, a seismic data fluctuation map can be obtained based on the seismic data after bandpass signal processing, and then a vehicle speed spectrum can be obtained based on the seismic data fluctuation map, wherein the vehicle speed spectrum is used to intuitively display the motion state of the vehicle.

[0200] Vehicle velocity spectrum - A 2D image spectrum plotted in (time, velocity) or (velocity, distance) coordinates, where bright spots in the image spectrum correspond to moving vehicles, and the amplitude of any image spectrum point is related to the weight of the vehicle. To generate the vehicle velocity spectrum, we iterate over all receivers (locations where seismic traces were recorded) and iterate over a time window within a trace. For each seismic trace recorded at a receiver within a time window, select multiple adjacent receiver traces on one side of the current receiver, and multiple adjacent receiver traces on the other side. For each selected receiver (location where a trace was recorded) on either side of the current receiver, apply a linear motion correction (LMO), where the correction time is calculated by dividing the distance offset between the two receivers by the specified vehicle velocity. Subtracting the calculated correction time from all corresponding seismic traces allows us to stack all corrected seismic traces and generate a single trace of data associated with the specified vehicle velocity. Cycle through all velocities within the specified vehicle velocity sweep range and repeat the above process; this generates a (time, velocity) spectrum associated with the current receiver. Perform the above process for each receiver within a small time window around time t, and place the rotated velocity spectrum (velocity, time) at the corresponding receiver location. The velocity spectra of all receivers at time t are combined in the domain to create a final velocity spectrum (speed, distance) along the road.

[0201] Figure 7 The vehicle speed spectrum generated from the bandpass signal processing data is shown. To obtain the vehicle speed spectrum, a linear moving average (LMO) stack and speed sweep are performed on the pre-processed data. Figure 7(a) in Figure 3 shows the bandpass signal processing data for LMO stacking. Figure 7 (b) in the figure shows the time domain vehicle speed spectrum of the station location plotted together with the signal. The vertical axis of the speed spectrum is the vehicle speed. Each individual bright spot (cloud) in the speed spectrum represents an individual vehicle.

[0202] Generating a vehicle speed profile from signal-enhanced traffic data may include:

[0203] Scan a range of vehicle speeds;

[0204] Apply linear motion overlay for a given scan speed;

[0205] Create vehicle speed spectrum images in the time domain.

[0206] Figure 8 A schematic diagram showing the generation of vehicle speed spectra from a single waveform of signal-enhanced traffic data.

[0207] like Figure 8 As shown, the seismic data fluctuation curve is scanned with a virtual scan line. The intersection of the virtual scan line and the seismic data fluctuation curve is mapped to a point in the vehicle speed spectrum.

[0208] In a specific example, the speed and / or movement path of the vehicle can be inferred by comparing the similarity of seismic data between different seismic recording devices. The similarity between seismic data of different seismic recording devices may include similarity between peak values ​​and / or speeds, etc.

[0209] Vehicle Association - Associate vehicle image points across different vehicle velocity spectra to reveal the movement of the same vehicle and derive a velocity profile over time or distance for each vehicle. This is accomplished by connecting all image points by applying a function that predicts the next vehicle location, or by applying machine learning methods to track the same image point across different vehicle velocity spectra. To accurately associate (or match) vehicles detected from vehicle velocity spectra, the Hungarian method can be used to identify the best match, thereby creating a vehicle velocity profile over time or distance. In addition to deriving vehicle velocity profiles from vehicle velocity spectra, vehicle velocity profiles can also be generated directly from recorded seismic data by utilizing machine learning methods.

[0210] Hungarian Algorithm - is a combinatorial optimization algorithm that solves assignment problems in polynomial time. The algorithm is commonly used to solve linear assignment problems, where the goal is to find the best pairing between elements of two sets while taking into account some cost or weight associated with each pair. For vehicle detection and association using traffic data, the Hungarian algorithm can be used to associate detected vehicles over time and / or stations to optimize the overall assignment of trajectories to vehicles, thereby minimizing some cost function. This cost function can include factors such as speed and amplitude differences between vehicles.

[0211] In one embodiment, determining the vehicle speed and / or the vehicle movement path using geophysical methods may include:

[0212] Utilize the time domain vehicle speed spectrum of adjacent stations;

[0213] Find connections between the same vehicle;

[0214] Connect the same vehicle with lines, including:

[0215] ο Use piecewise linear lines;

[0216] ο Using piecewise polynomial curves;

[0217] o Use a piecewise polynomial curve, smoothed at the recording sites.

[0218] Fig. 9 The vehicle travel paths and the process of extracting vehicle speeds from the speed spectrum are shown. In order to obtain the speed and movement path of each individual vehicle, it is necessary to correlate the bright spots (clouds) between the stations.

[0219] like Fig. 9 As shown in (b), the upward sloping curve formed by connecting multiple points at different stations represents the vehicle's driving path. The vehicle speed at a certain point on the upward sloping curve can be determined based on the slope of the point. Since the upward sloping curve is not a straight line, the vehicle speed will vary between different stations. Based on the slopes of different points, the speed curve of the vehicle during driving can be obtained, such as Fig. 9 As shown in (c), irregular driving, traffic violations, and traffic accident warnings are determined by using the moving path and speed information of individual vehicles.

[0220] In one embodiment, determining the vehicle speed from the vehicle speed spectrum and the movement path may include:

[0221] Use vehicle speeds from the speed spectrum at the recording site;

[0222] interpolate the speed of the same vehicle between recording stations; and

[0223] · Derivatives of individual vehicle movement curves obtained using piecewise polynomial curves and / or using piecewise polynomial curves and smoothed at the recording sites.

[0224] In one embodiment, whether the vehicle is speeding may be determined by comparing the vehicle speed with a preset reference line. Fig.10 A schematic diagram for determining whether a vehicle is speeding is shown. Specifically, when a vehicle speed curve is higher than a reference line, the vehicle is speeding; and when a vehicle speed curve is lower than the reference line, the vehicle is not speeding.

[0225] In one embodiment, the type and weight of the vehicle may be determined based on the peak value of the vehicle on each of a plurality of seismic data fluctuation curves.

[0226] Vehicle Weight - The amplitude response associated with a vehicle in a seismic record is caused by the vehicle's weight, speed, number of axles, road conditions, soil conditions beneath the road, sensor instrumentation response, and sensor-to-ground coupling. The vehicle's weight dominates the signal amplitude. With calibration or isolation of other factors, the vehicle's weight can be estimated from the amplitude response.

[0227] In an implementation example, determining relative vehicle weight and vehicle type by extracting vehicle amplitude and frequency attributes (including data envelope and power spectrum) may include:

[0228] Determine vehicle type from vibration signals;

[0229] Determine vehicle weight from seismic amplitude and phase, vehicle speed, number of axles, data envelope, and power spectrum;

[0230] Create a linear regression empirical relationship between vehicle weight, lane position, number of axles, and the vehicle's seismic response at the recording site; and

[0231] Create nonlinear regression empirical relationships between vehicle weight, lane position, number of axles, and the vehicle's seismic response at the recording site.

[0232] Fig.11 An interpretation of the relationship between vehicle type, speed, and weight and earthquake amplitude is shown. Fig.11 (a) shows the interpretation of vehicle type from earthquake amplitude. Fig.11 (b) in the figure shows the relationship between vehicle type and vehicle weight. Fig.11 (c) in Figure 2 shows the relationship between vehicle weight and earthquake amplitude obtained by linear regression.

[0233] In this embodiment, four types of vehicles are considered (small, medium, medium-large and large). It can be seen that the larger the vehicle size and the heavier the weight, the greater the amplitude.

[0234] Since the recorded seismic data generated by vehicles contains information about vehicle speed, weight, type, and lane, machine learning (ML) can be used to extract this vehicle information. The present disclosure introduces an intermediate step in ML when applying machine learning to improve vehicle detection by converting seismic time series data into road heat maps, thereby generating real-time road condition information. This information can then be used to detect vehicles and determine their location, lane, movement path, speed, and even type and weight.

[0235] In an implementation case, the vehicle type, speed, and weight recognition model may be used to determine the vehicle type, speed, and weight.

[0236] In an implementation case, a vehicle type, speed, and weight recognition model may be trained using multiple known vehicle types, speeds, and weights and corresponding seismic data.

[0237] In an implementation case, the vehicle type, speed, and weight may be determined by inputting the seismic data into a vehicle type, speed, and weight identification model.

[0238] Stopped Vehicles on Highways - Stopped vehicles on highways can pose significant risks and challenges to drivers and overall traffic flow. Sudden stops can occur due to emergencies, breakdowns, or unforeseen circumstances, creating potentially dangerous situations. When a vehicle stops on a busy highway, it disrupts the smooth flow of traffic and increases the likelihood of rear-end collisions. In such situations, timely response from traffic management is critical and warning approaching drivers becomes even more important. When a vehicle stops on a highway, the continuity of the vehicle vibration signal along the road in the recording is interrupted. Applying algorithms or machine learning methods to analyze the data can help detect sudden stops and notify traffic control authorities. Using seismic data to reconstruct traffic flow in real time can provide early warnings to law enforcement and take immediate actions to avoid sudden stops on highways, thereby maintaining traffic safety and reducing the risk of dangerous collisions.

[0239] In an embodiment, whether the vehicle is stopped may be determined based on the vehicle moving path. Fig.12 A schematic diagram illustrating a vehicle stop is shown. Within the elliptical area, it is clear to see that the seismic data at the center of the ellipse disappears at the next sensor location. This indicates that the vehicle stopped suddenly.

[0240] Fig.13 A single vehicle movement path along the road is shown (shown as a curve). This curve is traced in the recorded vehicle seismic data. From this selected vehicle movement path, the vehicle movement speed (slope of the curve) can be easily obtained and the driver's behavior can be analyzed to determine whether the driver is speeding.

[0241] Vehicle Correction Statics - A small time correction used to adjust for signal phase variations of the same vehicle on a particular gather to achieve high quality stacking in velocity spectrum calculations. Signal shifts over several adjacent gathers should be close to linear. Variations can be due to irregular receiver locations, lane changes, and vehicle speed variations. This correction statics can be calculated by correlating the stacked traces with the individual traces before stacking, or extracted directly from the data using machine learning methods.

[0242] Vehicle Warning System - Using traffic flow information obtained from seismic sensor data, the system will automatically trigger a warning for all oncoming vehicles when the overall traffic speed or the speed of some or individual vehicles is found to be drastically reduced near the seismic sensor. This warning can be displayed on a roadside electronic information board, or in the form of a warning light or siren, or on the monitoring screen of the traffic management department. The warning information can also be sent to nearby mobile phones or in-vehicle navigation systems. The vehicle warning system is designed to avoid or minimize traffic accidents, especially in foggy weather or on blind bends on the road.

[0243] like Fig.13 As shown, the curve represents the movement path of a vehicle. From this movement path, the speed of the respective vehicle can be determined. The speed gradually decreases, which may indicate a traffic jam ahead.

[0244] In an implementation case, pedestrians walking on the road can be detected based on seismic data.

[0245] Walking on highways is extremely dangerous due to high-speed traffic and lack of infrastructure suitable for pedestrians. This very risky activity increases the possibility of accidents and injuries to pedestrians and drivers. By analyzing seismic data recorded along highways and using machine learning techniques, early warnings of events can be provided, preventing potentially fatal accidents. When a person walks past a seismic sensor, the footsteps will be recorded by the sensor and identified through real-time analysis.

[0246] Fig.14 The typical response of a vehicle during an earthquake is shown. Fig.15 The image below shows the seismic response when a person is walking on the highway (as shown in this area). By analyzing the type of seismic signal, it is possible to distinguish whether it is a person walking on the road or a vehicle driving on the lane.

[0247] Objects Fallen from Moving Vehicles - Objects fallen from moving vehicles can pose a serious safety hazard on the road. Whether it is unsecured cargo, debris, or accidentally dropped items, these fallen objects have the potential to cause accidents, property damage, or personal injury. Rapid detection of objects fallen from moving vehicles is critical to preventing accidents and ensuring road safety. By leveraging seismic data recorded along the road and using advanced geophysical techniques and machine learning algorithms, real-time traffic monitoring systems can improve road safety and reduce potential accidents by quickly detecting objects fallen from moving vehicles and notifying the relevant authorities.

[0248] Natural Hazard Intensity Maps - Earthquakes, storms, hurricanes or typhoons shake the ground. Using the same roadside seismic sensors, intensity maps of ground vibrations can be revealed and reported in real time. This map can be used by authorities and the public to plan rescues and issue warnings.

[0249] Vibration-triggered video attention - Information from seismic sensors can be used as a supplement to video surveillance. If a seismic sensor detects any abnormal vibration, it can be used to trigger the attention of video cameras nearby at the same location. Such a triggering system can help avoid processing a large amount of camera and video data.

[0250] Just like detecting pedestrians, detecting objects dropped from moving vehicles, natural disaster intensity maps, and vibration-triggered video attention can also be implemented.

[0251] Fig.16 shows the extrapolation of the velocity spectrum from the time domain to the space domain. Fig.16 The vehicle speed spectrum along the road shown in (b) is obtained from Fig.16 (a) is the extrapolation of the time domain vehicle speed spectrum shown. Fig.16 (b) Visually demonstrates the movement of vehicles along the road. Fig.16 The image shown in (b) can provide a more intuitive and convenient view for traffic managers.

[0252] In order to obtain the vehicle speed spectrum along the road, the vehicle speed spectrum generated by LMO superposition in the time domain at each measuring station needs to be extrapolated along the road.

[0253] Generating a vehicle speed spectrum along a road in the spatial domain may include:

[0254] Time domain vehicle speed spectra using multiple recording sites; and

[0255] • Extrapolate the speed spectrum along the road in the spatial domain.

[0256] Predicting the next vehicle position - By using the current vehicle speed spectrum (time, speed) of a receiver within a certain time window, assuming that the vehicle speed remains constant, the vehicle speed spectrum of a virtual receiver near (in front of or behind) the current receiver can be predicted. By using the current vehicle speed spectrum (speed, distance) of multiple receivers at a specific time point, assuming that the vehicle speed is constant, the vehicle speed spectrum at the next time point (in front of or behind) close to that time point can be predicted. Both of the above predictions are achieved by calculating the time or distance offset using the speed values ​​in the speed spectrum.

[0257] exist Fig.16 In (b), the vehicle velocity spectrum at a specific point in time is plotted for a series of receivers along the road. From this graph, the next position of the vehicle can be calculated.

[0258] One embodiment of the present application provides a system for monitoring traffic flow through seismic data. Fig.17 The structure of a system 1700 for monitoring traffic flow based on seismic data is shown, and the system is designed according to an embodiment. Fig.17 As shown, system 1700 may include seismic recording devices 1710 - 1 to 1710 - n , a data processing module 1720 , and an analysis module 1730 .

[0259] In the embodiment, the seismic recording devices 1710-1 to 1710-n may be configured to acquire seismic data. The data processing module 1720 may be configured to acquire a seismic data fluctuation graph based on the seismic data, wherein the seismic data fluctuation graph includes a plurality of seismic data fluctuation curves of each seismic recording device.

[0260] In the embodiment, the analysis module 1730 may be configured to analyze traffic flow based on the earthquake data fluctuation graph, and the analysis module 1730 may also be configured to perform the above operations.

[0261] An embodiment of the present application provides a method for training a vehicle recognition model, which may include:

[0262] Acquire a first image containing a vehicle from the video;

[0263] Converting the first image into a second image representing the vehicle and the lane it is traveling in, wherein the second image serves as a training label;

[0264] Acquire first seismic data corresponding to the first image from the seismic data;

[0265] and training a vehicle recognition model based on the first seismic data and the second image.

[0266] Fig.18The process of converting from a first video image to a second two-color image is shown, where (a) shows a video recording of a road and vehicles are detected using image / video recognition techniques, and (b) shows training labels generated using vehicles detected from the video recording.

[0267] In an embodiment, the vehicle in the first video image may be captured using a YOLO algorithm.

[0268] YOLO algorithm - is a real-time object detection system that can simultaneously identify, classify, and localize multiple objects in an image or video frame. For vehicle detection and association with traffic data through machine learning, the YOLO algorithm can be used to generate labels that accurately capture the true situation of moving vehicles on the road from recorded videos. The most important part of training a machine learning model is to obtain labels. We want to record videos of roads ( Fig.18 (a) in order to get the absolute truth of the passing vehicles. To detect and capture vehicles in the video, the YOLO algorithm can be used on each frame of the video. This can then be converted into a bird's-eye view heat map of the road ( Fig.18 (b) in the figure for simpler analysis.

[0269] In an embodiment, a YOLO algorithm may be used to detect and capture vehicles in recorded videos and convert real video recordings into bird's-eye view heat map cartoons of the road for machine learning vehicle detection.

[0270] In an embodiment, the visualized vehicle data may be used to generate machine learning training labels for a neural network, including:

[0271] Generate labels from the intermediate steps of detecting and capturing vehicles in recorded videos and convert real video recordings into bird’s-eye-view heatmap cartoons of the road;

[0272] Obtain the seismic data corresponding to the label by splitting the seismic data into small window frames.

[0273] Generating heatmaps and detecting vehicles using machine learning may include:

[0274] Generate heatmaps and detect vehicles, types, weights, speeds and lanes using any single component data from single-component and / or multi-component sensors;

[0275] Generate road heatmaps and detect vehicles, types, weights, speeds, and lanes using any combination of multi-component data;

[0276] Generate heat maps and detect vehicles, types, weights, speeds, and lanes using multiple seismic sensors;

[0277] Create heatmaps and detect vehicles, types, weights, speeds, and lanes using any time series generated by vehicles;

[0278] Generate heatmaps and detect vehicles, types, weights, speeds, and lanes using any combination of the above data.

[0279] Fig.19 The figure shows the use of machine learning for vehicle detection and interpretation, where (a) shows the seismic data map used for vehicle detection and interpretation, and (b) shows the generated road heat map showing the predicted vehicle location. In order to obtain the vehicle seismic data corresponding to the label, we split the seismic data into small window frames (such as Fig.19 (a). After preprocessing the seismic data (e.g., normalization, Fourier transform), the data and label pairs are used to train a neural network, which will be used to generate a cartoon-like visual representation of the road (e.g., Fig.19 (b)). In addition, the visual representation of the road is used to analyze the vehicle position and obtain relevant vehicle and road information.

[0280] An embodiment of the present application provides a vehicle identification method, which may include: inputting seismic data to be identified into a vehicle identification model to determine the vehicle and the lane it is traveling on, wherein the vehicle identification model is trained using the above method.

[0281] Another embodiment of the present application provides a method for training a vehicle recognition model. The method may include:

[0282] Select earthquake data in multiple time windows as labels;

[0283] Identify the maximum peak or Gaussian distribution peak for each tag;

[0284] Determine the vehicle position corresponding to the maximum peak or the peak of the Gaussian distribution;

[0285] and training a vehicle recognition model based on the position and the corresponding maximum peak or Gaussian distribution peak. Fig. 20 Demonstrated direct access from ML-spike( Fig. 20 (a)) or ML-curve( Fig. 20 (b)) Obtained vehicle path.

[0286] In addition to deriving the vehicle trajectory from the vehicle velocity spectrum alone, another approach is to extract the vehicle trajectory directly from the recorded vibration data, e.g. Fig. 20 As shown. To obtain vehicle trajectories, the key is to generate ML-spike labels or ML-curve labels using the recorded vehicle vibration data. By training a neural network, an ML-spike model or ML-curve model is established, which can then be used to detect vehicles in seismic data. Fig. 20 In (a), the bright spots represent vehicles identified using the ML-spike model, while Fig. 20 In (b), the curves represent vehicles identified by the ML-curve model. The Hungarian optimal matching method can be used to extract vehicle trajectories based on vehicles detected by the ML-spike or ML-curve models (in Fig. 20 (a) and (b) are both represented as curves).

[0287] In one embodiment, vehicle detection and association of traffic data using machine learning and the Hungarian optimal matching algorithm may include one or more of the following steps:

[0288] 1) Preprocessing traffic data;

[0289] 2) Create ML-spike labels (e.g., peaks) for traffic data, where vehicle signals correspond to spikes at the same time on the recorded trajectory;

[0290] 3) Train the ML-spike model using traffic data as input and spikes as labels;

[0291] 4) Create ML Gaussian curve labels for traffic data, where vehicle signals correspond to Gaussian signals at the same time on the recorded trajectory;

[0292] 5) Train the ML Gaussian curve model using traffic data as input and spikes as labels;

[0293] 6) Create ML curve labels for user-defined curve functions on raw or preprocessed data;

[0294] 7) Train ML curve models for user-defined curve functions on raw or preprocessed data;

[0295] 8) detecting ML peaks by running the model in step 3) and taking a small data window around the corresponding peak in the traffic data or the preprocessed traffic data;

[0296] 9) detecting the ML curve by running the ML model in step 5), and taking a small data window around the corresponding peak of the ML curve in the traffic data or the preprocessed traffic data;

[0297] 10) detecting the ML curve of the user-defined curve by running the ML model in step 7), and taking a small data window around the corresponding peak of the ML curve in the traffic data or the preprocessed traffic data;

[0298] 11) Applying the Hungarian optimal matching algorithm, the vehicles detected in step 8), 9) or 10) are associated between multiple adjacent stations to form vehicle movement curves at these stations.

[0299] An embodiment of the present application provides a vehicle identification method. The method may include:

[0300] Input the earthquake data to be identified into the vehicle identification model to determine the vehicle position;

[0301] and determining the vehicle speed and / or the vehicle movement path based on the position, wherein the vehicle recognition model is trained using the above method.

[0302] The neural network can have additional metadata containing site information. These metadata, such as temperature, road type, etc., can be used to improve vehicle detection results. Other domain adaptation methods and transfer learning techniques can also be used to improve results across sites. Alternatively, if resources allow, we can create a model for each given site.

[0303] Road Monitoring

[0304] The recorded vehicle vibration data can be used not only to reconstruct traffic flows, but also to monitor geophysical conditions on and beneath the road and obtain seismic velocity and structural images of the subsurface. Changes in geophysical conditions beneath the road can be used to provide early warning of potential hazards and prevent them from occurring.

[0305] An embodiment of the present application provides a method for monitoring geological conditions under a road using seismic data, which may include:

[0306] Obtain target seismic data and reference seismic data from seismic recording equipment;

[0307] Generate target Green's function based on target seismic data;

[0308] Generate reference Green's function based on reference seismic data;

[0309] Generate near-surface relative velocity changes based on target Green’s function and reference Green’s function to monitor the geological conditions beneath the road.

[0310] Fig.21 FIG21 is a schematic flow chart 2100 illustrating a method for monitoring geological conditions and structures under a road using seismic data according to an exemplary embodiment of the present disclosure. Fig.21 As shown, flowchart 2100 may include the following operations.

[0311] In operation 2110, target seismic data and reference seismic data may be acquired from a seismic recording device.

[0312] In operation 2120, a target Green's function may be generated based on the target seismic data.

[0313] In operation 2130, a reference Green's function may be generated based on the reference seismic data.

[0314] In operation 2140 , a near-surface relative velocity change may be generated based on the target Green's function and the reference Green's function to monitor geological conditions under the road.

[0315] In one embodiment, the target seismic data and the reference seismic data may be subjected to signal preprocessing, noise removal, resampling, data offset correction, and filtering.

[0316] In one embodiment, the method of using seismic data to monitor geological conditions under a road may include one or more of the following:

[0317] 1) Traffic data preprocessing workflow, including:

[0318] a) Noise suppression;

[0319] b) resampling;

[0320] c) Data drift correction;

[0321] d) Multiple frequency ranges (f 1i ,f 2i ) bandpass filter, where f 0 <f 1i <f 1 ;f 0 <f 2i <

[0322] f 1 ;f 0 and f 1 are the lower and upper limits of the sensor frequency range;

[0323] e) OneBit filtering;

[0324] f) Median filtering;

[0325] g) Root mean square filtering;

[0326] h) smoothing filtering;

[0327] i) Fourier transform of traffic vibration data;

[0328] j) Conjugation of Fourier transformed data;

[0329] k) inverse Fourier transform;

[0330] l) inverse Fourier transform of conjugate Fourier transformed data;

[0331] m) applying a tapered window to the data in the time domain or the frequency domain or both domains;

[0332] 2) Extract body waves, elastic P waves, S waves, SH waves, surface waves, coda waves, Rayleigh waves, and Love waves from traffic data and generate reference and target Green’s functions;

[0333] 3) Generate Green's function from traffic data, including:

[0334] a) using the preprocessed data as described in 1), or any combination of a) to m) in 1);

[0335] b) cross-correlating the pre-processed data between sensor pairs in the time domain or frequency domain to generate a reference Green's function;

[0336] c) cross-correlating the pre-processed data between the sensor pairs in the time domain or frequency domain to generate a target Green's function (or current time Green's function);

[0337] d) Superimpose the cross-correlation results of (b) and (c) in (3);

[0338] e) performing weighted superposition of the cross-correlation results of (b) and (c) in (3);

[0339] 4) Create near-surface relative velocity variations using the ambient noise imaging method, reference and target Green's functions.

[0340] Background Noise Imaging - Seismic sensors mounted on the roadside can record vehicle movement and ambient noise. Especially after midnight, when traffic stops, ambient noise can be well recorded and used to infer the surface wave response between any two receivers. Correlating Green's functions or coda waves over a time interval can yield velocity changes in the near-surface region and help image changes in road conditions.

[0341] Fig. 22 Reference Green's functions between pairs of sites (seismic sensors) are shown.

[0342] Fig.23 A Green's function comparison between the reference and target times is shown.

[0343] Fig.24 An image showing the relative velocity changes of the subsurface medium beneath the road.

[0344] according to Fig.24 , the geological changes beneath the road can be observed.

[0345] In addition to reconstructing real-time traffic flows and monitoring geophysical conditions beneath the road, seismic data recorded from vehicle vibrations can be used to monitor changes in road surface characteristic responses over time. By analyzing road surface characteristic responses, their future changes can be understood and predicted, and early warnings of road conditions can be provided in real time. As a result, authorities can implement maintenance strategies and construction techniques that improve road safety and longevity, ultimately benefiting all road users.

[0346] Road surface characteristic response - refers to the dynamic behavior of a road under various conditions, including its interaction with environmental factors, traffic loads, and overall ability to withstand stress. Since road surfaces are constantly subjected to vehicle loads and are subject to environmental changes, temperature fluctuations, and extreme weather conditions (rain and snow), it is important to understand how roads degrade, wear, or withstand stress over time. Seismic data recorded along roads can be used to monitor road surface conditions in real time and assess road health. By monitoring roads in real time and analyzing seismic data in the time and / or frequency domains, road service agencies can develop timely maintenance plans to ensure a safe and durable infrastructure for travelers.

[0347] Slope Rockfall and Landslides - Rockfall and landslides on roads pose a persistent challenge to safe travel, especially in mountainous or geologically unstable areas. These events, whether sudden rock falls or gradual landslides, pose a serious risk and can result in vehicle damage, personal injury or road blockage. Installing seismic sensors along roadsides and analyzing the collected seismic data can provide real-time monitoring and early warnings to help manage these hazards, ensure safe passage and minimize the impact of rockfall and landslides on road travelers.

[0348] Avalanches - During winter in northern mountainous regions, avalanches can pose a danger of quickly blocking roads, leading to isolation, accidents, and even casualties. Roadside infrastructure designed with seismic sensors can continuously monitor roads. Analyzing seismic data collected from seismic sensors is critical to minimizing the dangers posed by avalanches and ensuring safer travel for all.

[0349] An embodiment of the present application provides a method for monitoring road surface conditions using seismic data, which may include:

[0350] Obtain target seismic data and reference seismic data from seismic recording equipment;

[0351] Calculate a first total energy of a vehicle passing through a first frequency range within a reference time based on the reference seismic data;

[0352] Calculate the second total energy of the vehicle passing through the first frequency range within the target time based on the target seismic data;

[0353] and monitoring a road surface condition based on the first total energy and the second total energy.

[0354] Fig.25 is a schematic flow chart illustrating a method for monitoring road surface conditions using seismic data according to an exemplary embodiment of the present disclosure.

[0355] like Fig.25 As shown, flowchart 2500 may include the following operations.

[0356] In operation 2501, target seismic data and reference seismic data may be acquired from a seismic recording device.

[0357] In operation 2502, a first total energy passing through a vehicle within a first frequency range within a reference time may be calculated based on reference seismic data.

[0358] In operation 2503, a second total energy passing through the vehicle within the first frequency range within the target time may be calculated based on the target seismic data.

[0359] In operation 2504, a road surface condition may be monitored based on the first total energy and the second total energy.

[0360] In one embodiment, a method for monitoring road surface conditions using seismic data may include one or more of the following steps:

[0361] 1) Preprocessing of traffic data, including:

[0362] a. Noise suppression.

[0363] b. Resampling.

[0364] c. Data drift correction.

[0365] d.Multiple frequency ranges (f 1i ,f 2i ) bandpass filter, where f 0 <f 1i <f 1 ;f 0 <f 2i <f 1 ;f 0 and f 1 are the lower and upper limits of the sensor frequency range.

[0366] e.OneBit filtering.

[0367] f. Median filtering.

[0368] g. Root mean square filtering.

[0369] h. Smoothing filtering.

[0370] i. Fourier transform of traffic vibration data.

[0371] j. Conjugation of Fourier transformed data.

[0372] k. Inverse Fourier transform.

[0373] l. Inverse Fourier transform of the conjugate Fourier transformed data.

[0374] m. Apply a tapered window to the data in the time domain or the frequency domain or both domains.

[0375] 2) Using the pre-processed, band-limited time domain data, calculate the total energy of vehicles passing by within the selected frequency range within the reference time.

[0376] 3) Using the pre-processed, band-limited time domain data, calculate the total energy of vehicles passing through the selected frequency range within the target (or current) time.

[0377] 4) Using the pre-processed, band-limited frequency domain data, calculate the total energy of vehicles passing by within the selected frequency range within the reference time.

[0378] 5) Using the pre-processed, band-limited frequency domain data, calculate the total energy of vehicles passing through the selected frequency range within the target (or current) time.

[0379] 6) Calculate the ratio of the total energy of different frequency bands in steps 2) and 3).

[0380] 7) Calculate the ratio of the total energy of different frequency bands in steps 4) and 5).

[0381] 8) Perform weighted addition and / or subtraction on the results of steps 6) and 7).

[0382] Through the above steps, Fig.26 The road surface characteristic response is shown as a function of time.

[0383] According to an embodiment of the present application, a method for predicting a traffic accident is provided, and the method may include:

[0384] Obtain seismic data from seismic recording equipment;

[0385] Obtain information about the vehicle based on seismic data;

[0386] Predict traffic accidents based on vehicle information, current and historical road information, driver information, and weather information.

[0387] Fig. 27 A schematic flow chart is shown to illustrate a traffic accident prediction method according to an example embodiment of the present disclosure.

[0388] like Fig. 27 As shown, flowchart 2700 may include the following operations.

[0389] In operation 2701, seismic data may be acquired from a seismic recording device.

[0390] In operation 2702, information about the vehicle may be acquired based on the seismic data.

[0391] In operation 2703, a traffic accident may be predicted based on vehicle information, current and historical road information, driver information, and weather information.

[0392] According to an embodiment of the present application, a method for training a traffic accident prediction model is provided, and the method may include:

[0393] Obtain earthquake data, vehicle information, road information, driver information and weather information at the time of the accident;

[0394] · And train traffic accident prediction models based on earthquake data and this information.

[0395] According to an embodiment of the present application, a method for predicting a traffic accident is provided, and the method may include:

[0396] Input the earthquake data to be identified into a traffic accident prediction model to predict traffic accidents, thereby obtaining a traffic accident score, wherein the traffic accident prediction model is trained using the above method. According to one embodiment of the present application, a method for training a vehicle weight model is provided, and the method may include:

[0397] Get the peak value of the vehicle from the seismic data;

[0398] A traffic accident prediction model is trained based on the peak value and the weight of the vehicle.

[0399] According to an embodiment of the present application, a method for obtaining vehicle weight is provided, and the method may include:

[0400] Input the earthquake data to be identified into the vehicle weight model to obtain the weight of the vehicle, wherein the vehicle weight model is trained using the above method.

[0401] Using reconstructed digital traffic flow information and historical data, traffic accidents are predicted through machine learning or algorithms. This information may include one or more of the following:

[0402] 1) The driver’s driving experience or driving record;

[0403] 2) The driver’s accident history;

[0404] 3) Abnormal driving records of drivers discovered through vibration data;

[0405] 4) Abnormal road surface conditions;

[0406] 5) Roads under construction or maintenance;

[0407] 6) Roads with a history of accidents;

[0408] 7) Severe weather conditions such as snow, storms, hail, rain, wind and fog;

[0409] 8) Low visibility;

[0410] 9) Traffic flow density;

[0411] 10) Vehicle travel path;

[0412] 11) Vehicle speed and speed change data;

[0413] 12) Vehicle changing lanes;

[0414] 13) Historical road vibration data at the time of the accident. 14)

[0416] By applying the above machine learning models and detected vehicle data, real-time traffic flow can be monitored and driving can be assisted.

[0417] Vehicle movement curves can be generated directly from seismic data using a combination of manual picking, automatic picking, and machine learning picking.

[0418] By directly picking up the moving vehicle curve, the instantaneous vehicle speed can be calculated. The instantaneous vehicle speed is the slope of the vehicle moving curve, that is, v = dx / dt.

[0419] Using the directly picked moving vehicle curve, the average vehicle speed can be calculated, v = X / T, where X is the distance the vehicle moves along the road in the time period T.

[0420] By directly picking up the moving vehicle curve, the average driving speed of the vehicle can be calculated, and the formula is v=X / T, where X represents the distance the vehicle moves along the road in the time period T.

[0421] Providing traffic warnings using reconstructed digital traffic information, which may include:

[0422] 1) Accidents - Provide accident warnings by analyzing traffic data and abnormal behavior of vehicle movement curves or speed spectra.

[0423] 2) Overweight vehicles - Overloaded vehicles are determined by determining the vehicle weight using the above method.

[0424] 3) Speeding vehicle - calculated from the vehicle moving path, v = d / t, where d is the distance between sensors and t is the vehicle moving time between sensors.

[0425] 4) Determine speeding vehicles from the speed spectrum - Based on the speed spectrum, warn vehicles that exceed the speed limit.

[0426] 5) Determine vehicles below the speed limit from the speed spectrum - Based on the speed spectrum, warn vehicles that are very slow or below the speed limit.

[0427] 6) Sudden vehicle stop - find the breakpoint when the vehicle movement curve disappears; find from the vehicle speed spectrum that the speed spectrum (energy) value in the direction of vehicle movement decreases significantly, or directly by processing vibration data.

[0428] 7) Objects dropped from vehicles - using machine learning or analytics to detect objects that have fallen from a moving vehicle and hit the ground through signal patterns from vibration data.

[0429] 8) Irregular Driving - Uses vehicle movement profile, speed and lane information to find irregular lane and speed changes within a time window.

[0430] 9) Frequent lane changes - analyze lane information.

[0431] 10) Detecting people walking on the highway by signal matching between traffic signals and walking signals recorded on the roadside.

[0432] 11) Use machine learning models to detect people walking on the highway.

[0433] 12) Detect abnormal road conditions and underground road conditions by post-processing road surface feature responses and underground structure changes.

[0434] 13) Detect abnormal driving weather conditions by analyzing the road surface feature response and underground relative velocity changes over time.

[0435] According to one embodiment of the present application, an electronic device is provided, which may include: at least one processor; and a memory communicatively connected to the at least one processor, wherein the memory stores instructions that, when executed by the at least one processor, enable the at least one processor to perform the above method.

[0436] Fig.28 An electronic device according to an exemplary embodiment of the present disclosure is shown. Fig.28 As shown, the electronic device may include a memory 2801 and at least one processor 2802. The memory 2801 may be in communication with the at least one processor 2802. The memory 2801 stores instructions, which, when executed by the at least one processor, enable the at least one processor to perform the above method.

[0437] According to one embodiment of the present application, a non-transitory computer-readable storage medium is provided, which stores instructions, and when executed on a computer, causes the computer to perform the above method.

[0438] According to one embodiment of the present application, a computer program product is provided, comprising a computer program having instructions, which, when executed by a computer, causes the computer to perform the above method.

[0439] In an embodiment of this application, a calibration vehicle may be used to calibrate seismic sensors.

[0440] Calibration Vehicle - A vehicle used to calibrate roadside seismic sensors over time. This is the same or similar fixed weight vehicle that passes over the roadside seismic sensors at a specified speed. This process is repeated weekly or monthly or any other time interval and the data is analyzed to check if any changes have occurred. If a systematic amplitude change has occurred, this change will be accounted for in the surface consistency amplitude correction.

[0441] Road and Bridge Quality Assurance - Use a calibrated vehicle at a specified speed to pass over the road or bridge where the seismic sensors are installed and find any changes in the recorded full waveform data over a period of time. If the changes are large, it may indicate that the road or bridge quality has changed.

[0442] Although multiple components are shown in the various block diagrams above, those skilled in the art will appreciate that the embodiments of the present disclosure may be implemented without one or more components, or in combination with certain components. Although the above steps have been described in the order shown in the figures, those skilled in the art will appreciate that these steps may be performed in a different order, or that the embodiments of the present disclosure may be implemented without performing one or more of the above steps.

[0443] As can be appreciated from the foregoing, the electronic components of one or more systems or devices may include, but are not limited to, at least one processing unit, a memory, and a communication bus or communication device that couples various components, including the memory, to the processing unit. The system or device may include or access various device-readable media. The system memory may include device-readable storage media in the form of volatile and / or non-volatile memory (e.g., read-only memory (ROM) and / or random access memory (RAM)). By way of example, but not limitation, the system memory may also include an operating system, application programs, other program modules, and program data.

[0444] The embodiments of the present application may be implemented as a system, method or program product. Therefore, the embodiments may take the form of a full hardware embodiment, or an embodiment including software (including firmware, resident software, microcode, etc.), which may be collectively referred to as "circuit", "module" or "system" herein. In addition, the embodiments may take the form of a program product, which is embodied on at least one device-readable medium, including device-readable program code.

[0445] The present application may use a combination of multiple device-readable storage media. In the context of this document, a device-readable storage medium ("storage medium") may be any tangible, non-signal transmission medium that may contain or store a program consisting of program code that is configured to be used or used in conjunction with an instruction execution system, apparatus, or device. For the purposes of this disclosure, a storage medium or device should be interpreted as non-temporary, i.e., excluding signal or propagation media.

[0446] Although the present invention has been described in conjunction with the inventive concept embodiments shown in the accompanying drawings, it will be appreciated by those skilled in the art that various changes and modifications may be made without departing from the technical spirit and essential features of the inventive concept. It will be appreciated by those skilled in the art that various substitutions, modifications and changes may be made thereto without departing from the scope and spirit of the inventive concept. Therefore, all such modifications are intended to be included within the scope of the present invention, as defined in the claims.

Claims

1. A method for monitoring traffic flow using seismic data, the method comprising: oAcquire seismic data from seismic recording equipment; o obtaining a seismic data fluctuation map based on the seismic data, wherein the seismic data fluctuation map includes a plurality of seismic data fluctuation curves of each seismic recording device; as well as oMonitoring traffic flow based on seismic data and / or seismic data wave patterns.

2. The method according to claim 1, wherein the seismic recording devices are arranged on one side, both sides, in the middle of the road, or any combination of these positions, and the intervals can be fixed or variable.

3. The method according to claim 1, wherein the step of obtaining a seismic data wave map based on seismic data comprises: Perform signal enhancement on seismic data.

4. The method according to claim 3, wherein the step of obtaining a seismic data wave map based on seismic data further comprises: After signal enhancement, two-way traffic wavefield separation or attenuation is performed on the seismic data.

5. The method according to claim 4, wherein the step of obtaining a seismic data wave map based on seismic data further comprises: o Balancing of seismic data by using correction curves after separation or attenuation of the two-way traffic wavefield; o And obtain the seismic data fluctuation map based on the balanced seismic data.

6. The method according to claim 1, wherein the method further comprises: A vehicle speed spectrum is acquired based on the seismic data and / or the seismic data wave pattern, wherein the vehicle speed spectrum is used to visually display the motion state of the vehicle.

7. The method according to claim 6, wherein the method further comprises: Based on similarities between the seismic data of different seismic recording devices, the speed and / or movement path of the vehicle is determined.

8. The method according to claim 7, wherein the method further comprises: Determine whether the vehicle is speeding based on the vehicle speed.

9. The method according to claim 1, wherein the method further comprises: Based on the peak of the vehicle in the multiple seismic data fluctuation curves, the type and weight of the vehicle are determined.

10. The method according to claim 7, wherein the method further comprises: Determine whether the vehicle is stopped based on the vehicle's moving path.

11. The method according to claim 1, wherein the method further comprises: Determine whether there are pedestrians on the road based on seismic data.

12. The method according to claim 1, wherein the method further comprises: Detect if an object has fallen from the vehicle based on seismic data.

13. A system for monitoring traffic flow using seismic data, the system comprising: oSeismic recording equipment configured to acquire seismic data; o a data processing module configured to obtain a seismic data fluctuation map based on the seismic data, wherein the seismic data fluctuation map includes a plurality of seismic data fluctuation curves for each seismic recording device; o and an analysis module configured to analyze traffic flow based on a wave pattern of seismic data.

14. The system of claim 1, wherein the analysis module is further configured to perform the method of any one of claims 2 to 11.

15. A system for monitoring traffic flow using seismic data, the system comprising: oSeismic recording equipment configured to acquire seismic data; oProcessor; oA memory connected to a processor that stores instructions that, when executed, enable the processor to: o obtaining a seismic data fluctuation map based on the seismic data, wherein the seismic data fluctuation map includes a plurality of seismic data fluctuation curves for each seismic recording device; and oMonitoring traffic flow based on seismic data wave patterns.

16. A method for training a vehicle recognition model, comprising: oGet the first image containing the vehicle from the video; o Converting the first image into a second image representing the vehicle and its lane, where the second image is used as a training label; o obtaining a first set of seismic data corresponding to the first image from the seismic data; o and training a vehicle recognition model based on the first set of seismic data and the second image.

17. A vehicle identification method, comprising: o Input the seismic data to be identified into the vehicle identification model to determine the vehicle and the lane it is traveling in, oWherein, the vehicle recognition model is trained using the method of claim 16.

18. A method for training a vehicle recognition model, comprising: o Pick up earthquake data in multiple time windows as labels; oIdentify the maximum peak or Gaussian distribution peak for each label; o determining the vehicle position corresponding to the maximum peak or the peak of the Gaussian distribution; and oTrain a vehicle recognition model based on the location and the corresponding maximum peak or Gaussian distribution peak.

19. A vehicle identification method, comprising: o Input the seismic data to be identified into the vehicle identification model to determine the location of the vehicle; as well as o Determine the speed and / or movement path of the vehicle based on the position, wherein: o The vehicle recognition model is trained using the method of claim 18.

20. A method for monitoring geological conditions beneath a road using seismic data, the method comprising: o Obtain target seismic data and reference seismic data from seismic recording equipment; oGenerate target Green's function based on target seismic data; oGenerate reference Green's functions based on reference seismic data; oGenerate near-surface relative velocity changes based on target and reference Green’s functions to monitor geological conditions beneath the road.

21. The method according to claim 20, further comprising: oPerform signal preprocessing, noise removal, resampling, data offset correction and filtering on target and reference seismic data.

22. The method according to claim 21, wherein the filtering comprises multi-frequency range (f 1i , f 2i ) band-pass filtering, where f0 < f 1i < f1; f0 < f 2i < f1; f0 and f1 are the lower and upper limits of the frequency range of the seismic recording device.

23. The method according to claim 21, further comprising: o extracting one or more body waves, elastic P waves, S waves, SH waves, surface waves, coda waves, Rayleigh waves, and Love waves from each of the target seismic data and the reference seismic data; and oGenerate reference Green's function and target Green's function based on the extracted waves.

24. The method of claim 21, wherein generating the near-surface relative velocity variation comprises: oGenerate near-surface relative velocity changes based on target and reference Green’s functions using the ambient noise imaging method.

25. A method for monitoring road surface conditions using seismic data, the method comprising: o Obtain target seismic data and reference seismic data from seismic recording equipment; o calculating a first total energy of a vehicle passing through a first frequency range within a reference time based on the reference seismic data; oCalculate the second total energy of the vehicle passing through the first frequency range within the target time based on the target seismic data; o and monitoring a road surface condition based on the first total energy and the second total energy.

26. The method according to claim 25, wherein the calculation of the first total energy and the calculation of the second total energy are performed in the frequency domain or the time domain.

27. The method of claim 25, wherein the calculation of the first total energy and the calculation of the second total energy are performed in a plurality of frequency bands.

28. The method of claim 27, wherein monitoring the road surface condition based on the first total energy and the second total energy comprises: o calculating a ratio of first total energies of the plurality of frequency bands and calculating a ratio of second total energies of the plurality of frequency bands; as well as o Perform weighted summation and / or subtraction of the first ratio and the second ratio.

29. A method for predicting a traffic accident, comprising: a. Obtain seismic data from seismic recording equipment; b. Obtaining information about the vehicle based on seismic data; c. Predict traffic accidents based on vehicle information, current and historical road information, driver information, and weather information.

30. A method for training a traffic accident prediction model, comprising: a. Obtain earthquake data, vehicle information, road information, driver information and weather information at the time of the accident; b. And train traffic accident prediction models based on earthquake data and this information.

31. A method for predicting a traffic accident, comprising: a. Input the earthquake data to be identified into the traffic accident prediction model to predict traffic accidents. Therefore, the score of the traffic accident is obtained, wherein the traffic accident prediction model is trained using the method in claim 30.

32. A method for training a vehicle weight model, comprising: a. Get the peak value of the vehicle from the seismic data; as well as b. Train a vehicle weight model based on the peak value and the vehicle's weight.

33. A method for obtaining vehicle weight, comprising: a. Input the earthquake data to be identified into the vehicle weight model to obtain the weight of the vehicle, Wherein the vehicle weight model is trained using the method of claim 32.

34. An electronic device comprising: a. at least one processor; b. and a memory communicatively connected to at least one processor, wherein the memory stores instructions that, when executed by at least one processor, cause the at least one processor to perform the method described in any one of claims 1-11 or 16-33.

35. A non-transitory computer-readable storage medium storing instructions that, when executed on a computer, cause the computer to perform the method of any one of claims 1-11 or 16-33.

36. A computer program product comprising a computer program having instructions which, when executed by a computer, cause the computer to perform the method of any one of claims 1-11 or 16-33.

Images