A Road Congestion Prediction Method and System Based on Spatiotemporal Propagation and Superposition of Vehicle Equivalent Wave Packets

Abstract

This application provides a road congestion prediction method and system based on the spatiotemporal propagation and superposition of vehicle equivalent wave packets. The method converts the number of vehicles acquired by roadside detection equipment at a fixed sampling period into temporally continuous vehicle equivalent wave packets, and determines the wave packet translation speed based on vehicle speed. For road sections lacking detection equipment, complete wave packets are generated using traffic state interpolation from adjacent detection equipment. All wave packets are driven to propagate along the road direction in the road's spatiotemporal domain, with vehicle numbers added or subtracted at road boundaries. All wave packets that have propagated to the target location and still contain vehicles are superimposed to obtain the total traffic flow waveform. Congestion is determined based on the waveform peak characteristics, equivalent propagation speed, and duration. This application transforms discrete detection data into continuous wave packets with propagation attributes, achieving a unified expression of the entire road's traffic flow status and congestion prediction.

Description

Technical Field

[0001] This application relates to the field of road congestion prediction technology, and more specifically, to a road congestion prediction method and system based on the spatiotemporal propagation superposition of vehicle equivalent wave packets. Background Technology

[0002] Data acquired by existing roadside cameras, millimeter-wave radar, and speed probes is typically used as independent point data, making it difficult to directly represent the continuous process of traffic flow moving, merging, and separating along the route over time. For blind spots lacking detection equipment, existing completion methods easily lead to duplicate vehicle counting and inflated totals, affecting prediction accuracy. During prediction, if new detection data is continuously accumulated, the same batch of vehicles passing through multiple detection points will be counted repeatedly; furthermore, vehicle merging and departure at entrances and exits are treated as boundary events, making it difficult to integrate them with traffic flow movement.

[0003] Therefore, how to transform the discrete detection data across the entire line into a unified and propagable continuous representation, while simultaneously achieving blind spot state completion, boundary event processing, and spatiotemporal overlay prediction of traffic flow, has become an urgent technical problem to be solved. Summary of the Invention

[0004] This application provides a road congestion prediction method and system based on the spatiotemporal propagation superposition of vehicle equivalent wave packets, which solves the problem that existing road point detection data are independent and difficult to continuously express the spatiotemporal propagation process of traffic flow, and realizes unified and continuous modeling of the traffic state of the entire road and accurate congestion prediction.

[0005] To solve the above-mentioned technical problems, the technical solution of this application is as follows: Firstly, this application provides a road congestion prediction method based on the spatiotemporal propagation and superposition of vehicle equivalent wave packets, including: Based on the number of vehicles and vehicle speeds obtained by roadside detection equipment at fixed sampling periods, the number of vehicles in each sampling period is converted into time-continuous vehicle equivalent wave packets, and the translation speed of each wave packet along the road direction is determined according to the vehicle speed of the corresponding period to generate the initial wave packet. For road sections lacking detection equipment, interpolation is performed using the traffic conditions of adjacent detection equipment to generate complete wave packets and their corresponding translation velocities; The initial wave packet and the completed wave packet are driven to propagate along the road direction in the road spatiotemporal domain at their respective corresponding translation speeds. At the road boundary position, the initial wave packet and the completed wave packet are processed to add or subtract vehicles based on the number of vehicles detected at the boundary. Wave packets that still contain vehicles after the vehicle addition or subtraction processing continue to propagate. At the target location to be predicted and its corresponding prediction time, all wave packets that have been propagated and still contain vehicles after the vehicle number has been increased or decreased are superimposed to obtain the total traffic flow waveform. Based on the peak characteristics of the total traffic flow waveform, the equivalent propagation speed of the wave packet, and the duration of the corresponding state, it is determined whether congestion has occurred, and the congestion prediction result is output.

[0006] Furthermore, the step of converting the number of vehicles in each sampling period into time-continuous vehicle equivalent wave packets includes: The number of vehicles detected in each sampling period is used as the amplitude of the discrete signal to construct a periodic discrete signal centered at the sampling time. The periodic discrete signal is then convolved using a Gaussian kernel function to convert it into a time-continuous vehicle equivalent wave packet.

[0007] Furthermore, the step of interpolating the traffic conditions of adjacent detection devices to generate complete wave packets and their corresponding translation velocities for road segments lacking detection devices includes: Based on the vehicle density and vehicle speed of the nearest upstream and downstream detection points of the road segment lacking detection equipment, spatial linear interpolation is performed to obtain the interpolation density of each spatial unit within the road segment. The allocation weight of each spatial unit to the total number of vehicles on the road segment is determined based on the interpolation density and unit length of each spatial unit. Based on the allocation weights, the estimated total number of vehicles on the road segment is allocated to the corresponding spatial units, and the complete wave packet and its corresponding translation velocity are generated for each spatial unit.

[0008] Furthermore, the road boundary location includes road entrances and road exits; the vehicle quantity addition or reduction processing includes: At the road entrance, the number of vehicles detected at the entrance is used to generate a new wave packet, which is then assigned a translational speed and added to the main line for propagation. At road exits, based on the number of vehicles detected at the exit, the wave packets that propagate to the exit are deducted according to the source matching or diversion ratio. The number of vehicles deducted does not exceed the number of vehicles currently contained in the wave packet, and wave packets that still contain vehicles after deduction continue to propagate at the original translation speed.

[0009] Furthermore, the step of assigning translation speed includes: within the sampling period for generating the new wave packet, if V2X vehicle speed data exists, the average V2X vehicle speed within that period is taken as the translation speed of the new wave packet; if V2X vehicle speed data does not exist, the historical average speed corresponding to that sampling period is taken as the translation speed of the new wave packet.

[0010] Furthermore, the vehicle quantity deduction process also includes: If the number of vehicles leaving the road exit is greater than the sum of the number of vehicles currently contained in all wave packets that have propagated to the exit, then the number of vehicles currently contained in all wave packets is reduced to zero, and the excess is recorded as a measurement deviation. Wave packets containing a negative number of vehicles are not generated.

[0011] Furthermore, the process of superimposing all propagating wave packets that still contain vehicles after the addition or subtraction of the number of vehicles to obtain the total traffic flow waveform is as follows: The intensity of all wave packets arriving at the same target location at the same prediction time is summed to obtain the total traffic waveform of that target location at that prediction time.

[0012] Furthermore, based on the peak characteristics of the total traffic flow waveform, the equivalent propagation speed of the wave packet, and the duration of the corresponding state, it is determined whether congestion has occurred, including: if the ratio of the peak value of the total traffic flow waveform to the traffic capacity of the road segment reaches a preset threshold, the equivalent propagation speed of the wave packet is lower than a preset speed threshold, and the number of sampling periods in which the state is continuously maintained reaches a preset number of periods, then congestion is determined to have occurred; otherwise, congestion is determined not to have occurred.

[0013] Furthermore, it also includes: when the interval from the last refresh time reaches a preset refresh cycle, regenerating the initial wave packet and the complete wave packet based on the latest detection data, and replacing the currently propagating initial wave packet and the complete wave packet with the regenerated initial wave packet and the complete wave packet as a whole.

[0014] Secondly, this application provides a road congestion prediction system based on the spatiotemporal propagation and superposition of vehicle equivalent wave packets, including: The wave packet initialization module is used to convert the number of vehicles in each sampling period into time-continuous vehicle equivalent wave packets based on the number of vehicles and vehicle speeds obtained by the roadside detection equipment at a fixed sampling period, and to determine the translation speed of each wave packet along the road direction according to the vehicle speed of the corresponding period, thereby generating the initial wave packet. The blind spot completion module is used to interpolate the traffic conditions of adjacent detection devices for road sections with missing detection devices, and generate completion wave packets and their corresponding translation speeds. The wave packet propagation and boundary processing module is used to drive the initial wave packet and the completed wave packet to propagate along the road direction in the road spatiotemporal domain at their respective corresponding translation speeds, and to perform vehicle number addition or subtraction processing on the initial wave packet and the completed wave packet at the road boundary position according to the number of vehicles detected at the boundary. Wave packets that still contain vehicles after the vehicle number addition or subtraction processing continue to propagate. The wave packet superposition module is used to superimpose all wave packets that have been propagated to and passed through the vehicle number addition or subtraction processing at the target location to be predicted and its corresponding prediction time to obtain the total traffic flow waveform. The congestion determination module is used to determine whether congestion has occurred based on the peak characteristics of the total traffic flow waveform, the equivalent propagation speed of the wave packet, and the duration of the corresponding state, and outputs the congestion prediction result.

[0015] Compared with the prior art, the beneficial effects of the technical solution of this application are: This application uses wave packet translation velocity to drive traffic flow along the road direction, which can directly express the physical process of vehicle batches moving downstream, merging and separating over time; by interpolating blind spot sections, it achieves full-line state coverage while ensuring the conservation of total vehicle volume; by adding and subtracting boundary conditions, it ensures vehicle conservation constraints; by obtaining the total traffic flow waveform through multi-wave packet spatiotemporal superposition, and by combining peak characteristics, equivalent speed and duration to jointly determine congestion, it improves prediction accuracy. Attached Figure Description

[0016] Figure 1 A flowchart illustrating a road congestion prediction method based on the spatiotemporal propagation and superposition of vehicle equivalent wave packets, provided in this application embodiment; Figure 2 This is a schematic diagram of the structure of a road congestion prediction system based on the spatiotemporal propagation and superposition of vehicle equivalent wave packets, provided in an embodiment of this application. Detailed Implementation

[0017] To make the objectives and advantages of this application clearer, the application will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining this application and are not intended to limit this application.

[0018] The prefixes such as "first" and "second" used in this application embodiment are merely for distinguishing different descriptive objects and do not limit the position, order, priority, quantity, or content of the described objects. The use of ordinal numbers and other prefixes used to distinguish descriptive objects in this application embodiment does not constitute a restriction on the described objects. The description of the described objects is given in the context of the embodiments, and the use of such prefixes should not constitute unnecessary restrictions.

[0019] Preferred embodiments of this application will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of this application and are not intended to limit the scope of protection of this application.

[0020] Please see Figure 1 This is a flowchart illustrating a road congestion prediction method based on the spatiotemporal propagation and superposition of vehicle equivalent wave packets, provided in an embodiment of this application. The embodiment of this application provides a road congestion prediction method based on the spatiotemporal propagation and superposition of vehicle equivalent wave packets, including: Step S1: Based on the number of vehicles and vehicle speeds obtained by the roadside detection equipment at a fixed sampling period, the number of vehicles in each sampling period is converted into time-continuous vehicle equivalent wave packets, and the translation speed of each wave packet along the road direction is determined according to the vehicle speed of the corresponding period to generate the initial wave packet. Step S2: For road sections lacking detection equipment, interpolation is performed using the traffic conditions of adjacent detection equipment to generate complete wave packets and their corresponding translation speeds. Step S3: Drive the initial wave packet and the completed wave packet to propagate along the road direction in the road spatiotemporal domain at their respective corresponding translation speeds, and perform vehicle number addition or subtraction processing on the initial wave packet and the completed wave packet according to the number of vehicles detected at the road boundary. Wave packets that still contain vehicles after the vehicle number addition or subtraction processing continue to propagate. Step S4: At the target location to be predicted and its corresponding prediction time, all the wave packets that have been propagated and still contain vehicles after the vehicle number addition or subtraction processing are superimposed to obtain the total traffic flow waveform. Step S5: Based on the peak characteristics of the total traffic flow waveform, the equivalent propagation speed of the wave packet, and the duration of the corresponding state, determine whether congestion has occurred and output the congestion prediction result.

[0021] In one embodiment, converting the number of vehicles in each sampling period into temporally continuous vehicle equivalent wave packets includes: The number of vehicles detected in each sampling period is used as the amplitude of the discrete signal to construct a periodic discrete signal centered at the sampling time. The periodic discrete signal is then convolved using a Gaussian kernel function to convert it into a time-continuous vehicle equivalent wave packet.

[0022] Furthermore, by constructing a periodic discrete signal using the number of vehicles within the sampling period as the discrete signal amplitude, and then performing convolution processing with a Gaussian kernel function, lossless conversion of discrete detection data to continuous wave packets is achieved. During the conversion process, each wave packet completely retains the vehicle quantity information of the corresponding sampling period, avoiding information distortion in the conversion from discretization to continuous processing, and making the wave packets generated by different detection points and different sampling periods have unified physical dimensions and comparability.

[0023] In one embodiment, the step of interpolating the traffic conditions of adjacent detection devices to generate complete wave packets and their corresponding translation velocities for road segments lacking detection devices includes: Based on the vehicle density and vehicle speed of the nearest upstream and downstream detection points of the road segment lacking detection equipment, spatial linear interpolation is performed to obtain the interpolation density of each spatial unit within the road segment. The allocation weight of each spatial unit to the total number of vehicles on the road segment is determined based on the interpolation density and unit length of each spatial unit. Based on the allocation weights, the estimated total number of vehicles on the road segment is allocated to the corresponding spatial units, and the complete wave packet and its corresponding translation velocity are generated for each spatial unit.

[0024] Furthermore, for road sections lacking detection equipment, spatial linear interpolation of vehicle density and speed at upstream and downstream detection points is used, and the normalized allocation weight is determined by the product of the interpolation density and the unit length. This achieves a reasonable distribution of the total number of vehicles among the spatial units in the blind spot road section. This ensures the conservation of the total number of vehicles before and after completion, avoids the amplification of the number of vehicles caused by simple replication, and assigns an independent translation speed to each completed spatial unit, so that the completed wave packet can truly reflect the differences in traffic conditions in the blind spot road section.

[0025] In one embodiment, the road boundary location includes a road entrance and a road exit; the vehicle quantity addition or reduction processing includes: At the road entrance, the number of vehicles detected at the entrance is used to generate a new wave packet, which is then assigned a translational speed and added to the main line for propagation. At road exits, based on the number of vehicles detected at the exit, the wave packets that propagate to the exit are deducted according to the source matching or diversion ratio. The number of vehicles deducted does not exceed the number of vehicles currently contained in the wave packet, and wave packets that still contain vehicles after deduction continue to propagate at the original translation speed.

[0026] Furthermore, the road boundary processing is clearly defined into two operations: entry addition and exit deduction. The entry addition wave packet can reflect the impact of vehicles merging into the main line on downstream traffic. The exit deduction achieves accurate deduction of the number of vehicles through source matching or diversion ratio, and the deduction amount does not exceed the number of vehicles currently contained in the wave packet, ensuring vehicle conservation and avoiding negative traffic flow at the exit, so that the impact of boundary events on main line traffic can be accurately quantified.

[0027] In one embodiment, the step of assigning translation speed includes: within the sampling period for generating the new wave packet, if V2X vehicle speed data exists, then the average V2X vehicle speed within that period is taken as the translation speed of the new wave packet; if V2X vehicle speed data does not exist, then the historical average speed corresponding to that sampling period is taken as the translation speed of the new wave packet.

[0028] Furthermore, for the translation speed of newly added wave packets at the entrance, real-time speed data of V2X vehicles is used first. When V2X data is missing, the historical average speed of the same period is used as a substitute, which ensures the real-time performance and reliability of the translation speed of newly added wave packets. The more comprehensive the V2X vehicle coverage, the more the assigned translation speed can reflect the actual operating status of the batch of vehicles, thereby improving the accuracy of congestion prediction.

[0029] In one embodiment, the vehicle quantity deduction process further includes: If the number of vehicles leaving the road exit is greater than the sum of the number of vehicles currently contained in all wave packets that have propagated to the exit, then the number of vehicles currently contained in all wave packets is reduced to zero, and the excess is recorded as a measurement deviation. Wave packets containing a negative number of vehicles are not generated.

[0030] In one embodiment, the step of superimposing all wave packets that have reached and passed through the vehicle number addition or subtraction processing and still contain vehicles to obtain the total traffic flow waveform specifically involves: The intensity of all wave packets arriving at the same target location at the same prediction time is summed to obtain the total traffic waveform of that target location at that prediction time.

[0031] Furthermore, the intensity of all effective wave packets arriving at the same target location at the same predicted time is summed to obtain the total traffic flow waveform. The superposition method is simple and clear. When multiple high-amplitude wave packets arrive at the same time, a higher peak value is formed. When wave packets arrive at a dispersed location, the waveform is relatively flat, which can intuitively reflect the vehicle density at the target location and provide a reliable waveform basis for subsequent congestion judgment.

[0032] In one embodiment, determining whether congestion has occurred based on the peak characteristics of the total traffic flow waveform, the equivalent propagation speed of the wave packet, and the duration of the corresponding state includes: if the ratio of the peak value of the total traffic flow waveform to the traffic capacity of the road segment reaches a preset threshold, the equivalent propagation speed of the wave packet is lower than a preset speed threshold, and the number of sampling periods in which the state is continuously maintained reaches a preset number of periods, then congestion is determined to have occurred; otherwise, no congestion is determined to have occurred.

[0033] Furthermore, three indicators—peak saturation, equivalent propagation speed, and duration—are used to jointly determine congestion, and situations where the conditions are not met are explicitly judged as not having congestion, forming a complete two-way judgment logic. The joint judgment of multiple indicators can effectively distinguish between real congestion and short-term high-flow or high-speed operation, reduce false alarm rate, and improve the accuracy and robustness of congestion judgment.

[0034] In one embodiment, the method further includes: when the interval from the last refresh time reaches a preset refresh period, regenerating the initial wave packet and the complete wave packet based on the latest detection data, and replacing the currently propagating initial wave packet and the complete wave packet with the regenerated initial wave packet and the complete wave packet as a whole.

[0035] Furthermore, a refresh mechanism is set up to replace the current wave packet field with the latest detection data when the preset refresh cycle is reached, which avoids the continuous accumulation of detection data during the prediction process, which would cause the wave packet to be counted repeatedly. Periodic refresh enables the prediction results to correct deviations in a timely manner, keep them synchronized with the actual traffic conditions, and maintain a high prediction accuracy in the long term.

[0036] In one embodiment, this application takes a section of highway as an example. Roadside cameras and speed detectors are deployed along this section, covering most of the road. Some sections, however, lack detection equipment due to geographical limitations. The section has one highway entrance and one highway exit. Some vehicles have V2X communication capabilities and can upload their speed in real time.

[0037] Establish a one-dimensional road marker coordinate system along the highway's driving direction, and denot any road location as... System startup or The refresh time is recorded as The fixed sampling period is denoted as This unified cycle will be used in all subsequent iterations. The start time of each sampling period and end time They are respectively:

[0038]

[0039] In the formula, Number of sampling periods; This refers to the time when the system starts up or refreshes. For a fixed sampling period; For the first At time s of each sampling period; For the first At time e of each sampling period.

[0040] No. Each sampling period from Beginning, to End. The sampling period is determined based on the minimum stable statistical duration of the vehicle detection equipment, the frequency of road traffic flow fluctuations, and the real-time prediction requirements. Wave packets can be reconstructed using historical data across multiple candidate periods, with the period exhibiting the lowest prediction location error, congestion time error, and false alarm rate being selected as the sampling period for the target road.

[0041] In the same round of prediction, roadside baseline data is only used for initialization during the startup or refresh phase. Within the prediction interval, the baseline vehicle equivalent wave packet moves downstream according to the determined vehicle batch translation speed, and new data from roadside cameras and speed sensors are no longer continuously added to the current prediction; after reaching the next refresh time, the latest measured data completely replaces the previous round of prediction results.

[0042] Let the first The road markers for the roadside monitoring points are: In the Within each sampling period, a total of [number] samples were detected. A vehicle. The first The vehicle equivalent factor of a vehicle is denoted as All vehicles .

[0043] No. The detection point at the first The discrete signal amplitude within a sampling period is defined as the total number of vehicle equivalents in that period:

[0044] In the formula, The amplitude is a discrete signal; the more vehicles within a fixed sampling period, the higher the amplitude. The larger the value, the higher the corresponding continuous waveform.

[0045] When different devices use different sampling periods, the amplitude can be converted into the average vehicle equivalent flow rate. The calculation formula is:

[0046] In this embodiment, the same sampling period is used throughout, and the data is directly used. As discrete amplitude values, to avoid repeated conversions.

[0047] Conversion of discrete-time signals to continuous waveforms: A Gaussian kernel function is used to convert periodic discrete signals into continuous waveforms. Defined as:

[0048] In the formula, The waveform smoothing bandwidth controls the time width of a waveform segment corresponding to one sampling period; t is the corresponding time within the sampling period.

[0049] The continuous basic waveform at the m-th detection point for:

[0050] in Let r be the center time of the r-th sampling period. The continuous waveform is the superposition of vehicle equivalent wave packets from multiple sampling periods on the time axis. The periodic discrete signal refers to the sequence of the total number of vehicle equivalents formed at the same detection point within multiple sampling periods, with the total number of vehicle equivalents in each sampling period representing the amplitude of a unit impact signal. The continuous waveform is not defined directly out of thin air, but is obtained by convolving the periodic discrete signal with a Gaussian kernel function.

[0051] Specifically, no. Periodic discrete signal at each detection point It can be represented as:

[0052] In the formula, For the first The total number of vehicle equivalents at each detection point in the r-th sampling cycle; For the first The center moment of each sampling period; The unit impact function.

[0053] The periodic discrete signal and the Gaussian kernel function After convolution, a continuous basic waveform is obtained. The continuous basic waveform is then calculated. The formula is:

[0054] In the formula, Let represent a single vehicle equivalent wave packet formed in the r-th sampling period. Vehicle equivalent wave packets from multiple sampling periods are superimposed on the time axis to form a continuous basic waveform. Since the integral of the Gaussian kernel function is... Therefore, the integral of a single wave packet is equal to the total number of vehicle equivalents in that sampling period. This ensures that the vehicle equivalent is not amplified or reduced after the discrete signal is made continuous.

[0055] because The integral is 1, therefore the area of ​​each period waveform segment is equal to the vehicle equivalent number for that period, as shown in the specific formula:

[0056] Peak value of a single periodic waveform segment at its center time for:

[0057] When all detection points use the same bandwidth At this time, the peak height is proportional to the number of vehicles in the cycle, thus giving the waveforms at different detection points and different sampling periods a unified and comparable physical meaning.

[0058] bandwidth It can be set according to the sampling period. Preferably, it can be set to... to To ensure comparability of peak values ​​across the entire prediction task, waveforms of the same type within the same prediction task use the same bandwidth.

[0059] Determination of the basic vehicle equivalent wave packet translation velocity: The roadside detection points simultaneously acquire the speed of each vehicle within the sampling period. The average speed of the vehicle batch at the m-th detection point in the r-th period. The formula is:

[0060] In the formula, The detection speed of the nth vehicle within the rth sampling period at the mth detection point; This represents the number of vehicles inspected within the current period. This is the average speed of the vehicle batch within this cycle. The cycle-averaged speed is assigned to the base vehicle equivalent wave packet formed in this cycle, serving as the translational speed of the vehicle batch within that wave packet in the current prediction round. This speed only represents the average displacement speed of the corresponding vehicle batch along the mainline travel direction and is not considered as the traffic shock wave speed determined by the difference in upstream and downstream flow and density states.

[0061] When no vehicle is detected in a certain period, the discrete amplitude of that period is 0, and no new vehicle equivalent wave packet is generated. If it is necessary to maintain the continuity of the batch translation velocity field, the velocity of the previous effective period or the historical average velocity of the same period can be used as the velocity background value at that position, but the velocity background value does not generate additional vehicle amplitude.

[0062] Suppose vehicle data is collected at a detection point on the road, with sampling periods of 10:00, 10:02, 10:04, and 10:06. At 10:00, vehicle equivalent wave packet 'a' is detected with an average speed of Va; at 10:02, vehicle equivalent wave packet 'b' is detected with an average speed of Vb; no vehicle is detected at 10:04; and at 10:06, vehicle equivalent wave packet 'c' is detected with an average speed of Vc. For the period at 10:04 where no vehicle is detected, to maintain the continuity of the translational speed of each wave packet, the average speed of the vehicle equivalent wave packet corresponding to the sampling period 10:04 can be taken as the speed of the previous valid period, i.e., the average speed Vb of vehicle equivalent wave packet 'b' corresponding to the sampling period 10:02.

[0063] For road sections without roadside equipment, the basic condition is completed. For sections without cameras or speed sensors, the nearest upstream detection point u and the nearest downstream detection point d are selected, and the section is divided into M spatial units according to the distance between adjacent detection devices. For the first... For each spatial unit, first calculate the spatial interpolation coefficients based on the unit's center position. And perform linear interpolation on the densities with the same period at both ends;

[0064] In the formula, For the first The center position of each spatial unit; This refers to the location of the nearest upstream monitoring point; The location is the nearest downstream detection point; the spatial interpolation coefficients represent the positional relationship between each spatial unit and the upstream and downstream detection points.

[0065] The density of detection points at both ends refers to the number of vehicles equivalent per unit road length at the nearest upstream detection point u and the nearest downstream detection point d on a road section without equipment within the same sampling period r, denoted as […]. and .

[0066] If the detection equipment can directly obtain the number of vehicles within the detection range, then the density of the i-th detection point in the r-th sampling period is... It can be represented as:

[0067] In the formula, Vehicle equivalent density; The total number of vehicle equivalents detected during this period; This represents the effective detection section length corresponding to this detection point.

[0068] If the detection equipment can only obtain the number of vehicles passing through and the average speed, then the vehicle equivalent flow rate can be calculated first. The calculation formula is as follows:

[0069] Then estimate the density based on the relationship between flow rate, velocity, and density. ,density The formula is:

[0070] In the formula, The sampling period length; This represents the average speed of the vehicles during that period. To prevent small positive numbers with a denominator of zero.

[0071] After obtaining the densities at both ends, the density of the first section of road without equipment can be determined. spatial units The specific formula for interpolation is as follows:

[0072] In the formula, These are spatial interpolation coefficients, used to represent the first... The relative positions of each spatial unit between upstream detection point u and downstream detection point d.

[0073] The density of detection points at both ends is the spatial density of vehicle equivalents calculated from data such as the total number of vehicle equivalents, detection length, and average speed of upstream and downstream detection points in the same sampling period. It is used to interpolate and estimate the vehicle distribution and the total number of vehicle equivalents in each spatial unit of the road section without equipment.

[0074] Simultaneously, the density at both ends is calculated based on the camera or probe. With interval length Estimate the total number of vehicle equivalents in the interval during the r-th sampling period. :

[0075] In the formula, For the first The interpolation density of a spatial cell in the r-th period; This refers to the density of upstream detection points; To reduce the downstream detection point density and avoid amplifying the total number of vehicles by replicating the average waveforms from both ends to multiple spatial units, a normalized weighting is constructed based on the product of the interpolation density and the unit length of each spatial unit. This uniquely allocates the total number of vehicles in the interval to each spatial unit. Each unit generates a vehicle equivalent wave packet accordingly, and the translational velocity of the vehicle batch is obtained by spatial interpolation of the average velocity of the periods at both ends.

[0076] In the formula, To be allocated to the first The vehicle equivalent of a spatial unit; j is the spatial unit index in the normalized summation; The continuous vehicle equivalent wave packet generated for this unit; This represents the batch translation speed of the vehicles in this wave packet; The average speed of the upstream detection points over a period of time; The average speed at downstream detection points is the periodic average speed. The above normalized allocation satisfies that the sum of the vehicle equivalents in each spatial unit equals the estimated total number of vehicle equivalents in the unequipped section. Each spatial unit retains only one unique wave packet and one vehicle batch translation speed. The completion is only used to reconstruct the initial spatial distribution and does not generate new vehicle sources, thereby avoiding the amplification of the total number of vehicles or the duplicate counting of the same batch of vehicles due to the replication of the average waveforms at both ends in each unit in unequipped road sections.

[0077] In the formula, This indicates that the total number of vehicle equivalents has been obtained; Indicates the first The interpolation density of a spatial cell in the r-th sampling period; Indicates the first The length of each spatial unit; This represents the total number of spatial units obtained by dividing the unequipped interval.

[0078] In the formula, j is the spatial cell index in the normalized summation, used to traverse the i-th cell in the unequipped interval. The first to the second One spatial unit; This indicates that the j-th spatial unit is in the... Interpolation density for each sampling period, This represents the length of the j-th spatial unit. This represents the sum of the interpolation density × cell length of all spatial cells within the unequipped interval, used to achieve normalized allocation.

[0079] For the r-th sampling period, the first... A continuous vehicle equivalent wave packet generated by a spatial cell, whose vehicle equivalent is allocated by that spatial cell. The expression obtained by Gaussian kernel continuum is the local wave packet expression for a single spatial cell and a single sampling period.

[0080] The establishment and movement of the initial basic wave packet of the entire line is as follows: at the time of system startup or refresh, based on the measured wave packets of all equipment locations and the conserved space interpolation and normalization allocation results of the equipment-free intervals, the initial basic vehicle equivalent wave packet field of the entire line is established, and the corresponding vehicle batch translation velocity field is established, and then the autonomous translation stage is entered.

[0081] At system startup or refresh time, the wave packet is completed based on the measured wave packet of each sampling period used for this round of initialization and the deviceless spatial unit, and the initial wave packet field of the entire line is organized according to the road location; the initial wave packet field is not re-established before the next refresh time.

[0082] For any road space cell, its initial position and the vehicle batch translation velocity of the periodic wave packet are determined by the spatial coordinates and velocity state of that cell, respectively; the translation velocity is maintained with the corresponding wave packet. The center position of the wave packet at the prediction time is:

[0083] In the formula, This represents the initial position plus the translation speed multiplied by the elapsed time. The initial position of the wave packet was detected for the detected road segment; The measured velocity of the wave packet vehicle batch detected in this test section is t; the predicted time is t. This represents the moment the wave packet is formed. This formula only describes the movement of the wave packet center downstream of the road and does not represent the addition of new vehicles or repeated generation of wave packets.

[0084] For a target position X located downstream of the vehicle equivalent wave packet, the vehicle batch translation time of the basic wave packet segment for:

[0085] When the target location is upstream of the initial position of the wave packet, the wave packet of this vehicle batch does not shift upstream and does not participate in the forward traffic flow calculation at the target location.

[0086] The waveform of the basic vehicle equivalent wave packet formed from the initial position c in the r-th sampling period after reaching the target position X is expressed as follows:

[0087] in, This represents a single original vehicle equivalent wave packet formed at position c in the r-th sampling period. The base predicted waveform for the target position consists of all vehicle equivalent wave packets that can reach that position at the corresponding time. Each wave packet is only shifted in position and is not copied or retained at the original position.

[0088] The waveform of the basic vehicle equivalent wave packet formed from the initial position c in the r-th sampling period after reaching the target position is represented by the predicted waveform of each vehicle equivalent wave packet reaching the target position. Based on the center position of the vehicle equivalent wave packet (total number of vehicle equivalents) corresponding to each detection point on the road in each sampling period and the average velocity of the vehicle equivalent wave packet, the waveform of each vehicle equivalent wave packet reaching the target position is determined.

[0089] In the formula, This represents the predicted waveform after the wave packet reaches the target position x; This represents the original continuous waveform formed by the wave packet at its initial position; This represents the propagation time or time offset required for the wave packet to translate from its initial position c to its target position x. The formula represents translating the original waveform according to the time required to reach the target position. This formula describes the predicted waveform after a single wave packet reaches the target position, not the total superposition result of all wave packets.

[0090] Generation of entry and exit periodic events: After the system starts or refreshes, vehicles continue to be counted at each highway entrance and exit according to the sampling period.

[0091] After the system starts or refreshes, each highway entrance and exit continues to count vehicles according to the sampling period. This can be understood as the following: at the exit detection point, the remaining vehicles after the vehicle equivalent wave packet from the previous detection point is reduced at the exit detection point are counted; and at the entrance detection point, the newly added vehicle equivalent wave packet is counted. Let the number of entrance / exit vehicles in the r-th period be q. and export vehicle equivalent They are respectively:

[0092] In the formula, The number of vehicles entering; This refers to the number of vehicles exported. Let n be the vehicle equivalent coefficient for the nth vehicle entering the gate; Let be the vehicle equivalent coefficient for the nth exported vehicle.

[0093] The continuous wave packet of the q-th inlet / outlet during the r-th period. Deducting demand from export vehicle equivalents Represented as:

[0094] The entry waveform packet represents vehicles merging into the mainline, joining the mainline as a new positive vehicle batch. The exit waveform does not generate a globally negative waveform propagating downstream; instead, it serves as a vehicle equivalent sink, only performing source matching and deduction on the actual vehicle batches arriving at that exit. When V2X, ETC, or vehicle identification is available, matching is performed directly by vehicle or vehicle batch; when direct matching is not possible, the diversion weight is determined based on the vehicle equivalent proportion of each arriving batch. For additional disturbances near the exit caused by lane changes, deceleration, or queuing, a separate positive disturbance waveform can be constructed.

[0095] The initial basic wave packet already includes vehicles that entered the road before the refresh time and are still running on the road. Therefore, the entry addition event and exit deduction event only count vehicle events generated after the refresh time, avoiding duplicate calculation of the initial basic wave packet and boundary events.

[0096] Determination of the translation speed of vehicle batch wave packets at entrances and exits: Suppose that N vehicles with valid V2X speed data are detected at the i-th entrance / exit in the r-th period. When N is greater than 0, the translation speed of the vehicle batch of newly added wave packets at the entrance in that period is taken as the average V2X vehicle speed. for:

[0097] In the formula, The number of vehicles with valid V2X speed data at the i-th entrance / exit during the r-th cycle; Let be the speed of the j-th V2X vehicle. When the real-time V2X speed cannot be obtained directly in this cycle, the average speed of the entrance / exit, the same driving direction, and the historical same period is used as the substitute vehicle batch translation speed for the new wave packet at the entrance in this cycle.

[0098] The uniform expression for the wave packet translation speed of vehicles entering and exiting the entrance / exit is as follows:

[0099] in, This represents the average speed for the same period in history. Once this velocity is assigned to a new wave packet at the current cycle ingress, it remains consistent with the vehicle batch in this round of prediction and is not repeatedly updated due to subsequent camera re-detection. The more comprehensive the V2X vehicle coverage, the more accurately the cycle-averaged velocity reflects the true operating status of that batch of vehicles. Exit deduction events only occur at the exit location and do not propagate downstream as negative wave packets.

[0100] The specific details of the inlet wave packet translation and outlet deduction positions are as follows: Let the position of the qth inlet / outlet be the corresponding road station number. For a newly added inlet wave packet, when the target position X is downstream of the inlet, the inlet wave packet in the rth period is translated to the target position. time The formula is:

[0101] In the formula, This is the entrance location; Add wave packet velocity to the inlet.

[0102] To represent the local amplification effect caused by inlet merging and outlet divergence, lane changing, and deceleration, the spatial influence gain can be set:

[0103] in, d represents the directed distance of the target point relative to the highway entrance / exit node; a is the nonlinear attenuation coefficient, preferably a=2, after which... This means no further amplification will occur. d < 0 indicates that the target point is upstream of the entrance / exit node, and d > 0 indicates that the target point is downstream of the entrance / exit node.

[0104] When an export deduction event occurs, the wave packet that actually reaches the q-th export in the corresponding period and still has remaining vehicle equivalents is determined as the set of deductible vehicle batches. For the b-th vehicle batch, the export deduction vehicle equivalents and the remaining vehicle equivalents after deduction are as follows:

[0105] in, The available vehicle equivalent for the b-th vehicle batch before the exit deduction; in the formula, the available vehicle equivalent before reaching the exit represents the vehicle equivalent of the b-th vehicle batch before it enters the exit influence position and has not yet been deducted; the source matching or diversion weights satisfy the requirement that the sum of the weights of each valid batch is 1. When the source can be directly identified, the deduction amount is allocated to the matching batch; when the source cannot be directly identified, it is allocated according to the proportion of vehicle equivalent of each arriving batch. The remaining wave packet after deduction continues to shift downstream; if the number of vehicles measured at the exit exceeds the available vehicle equivalent, the excess is only recorded as a measurement deviation or a quantity to be corrected, and must not form a negative wave packet.

[0106] The spatiotemporal superposition of all wave packets results in the total traffic flow waveform at the target location X at prediction time t. This waveform is composed of all vehicle batch wave packets arriving at that location and still having remaining vehicle equivalents after additions at the entrance and deductions at the exit, as well as optional disturbance waveforms.

[0107] In the formula, The waveform of the total traffic flow at target location X at predicted time t; b is the time it takes for the wave packet to propagate to the target location x; b is the wave packet number of the vehicle batch. This indicates the formation time or center time of the wave packet of the b-th vehicle batch; This represents the set of effective vehicle batch wave packets that can reach or influence the target location X at the predicted time t. This represents the remaining vehicle equivalent after deducting the output values ​​along the route for the wave packet of the b-th vehicle batch. is a Gaussian kernel function used to represent the continuous shape of the vehicle equivalent wave packet; This represents the waveform of the b-th wave packet when it reaches the target position X after a time shift; ) represents other additional perturbations not characterized by the spatial influence gain G(d), and For example, parameters can be freely set according to the actual situation, such as traffic accidents, road construction, and temporary road closures.

[0108] The summation set represents the effective batch of vehicles that can reach the target location X at the predicted time. The vehicle equivalent in the summation term represents the remaining vehicle equivalent of the corresponding batch after deducting the number of vehicles at each exit along the route. The time offset represents the time it takes for the batch to move to the target location. Additional terms represent optional exit lane changes, sudden deceleration, accident, or construction disturbance waveforms. The number of vehicles at exits is no longer directly superimposed as a global negative waveform.

[0109] Vehicle equivalent wave packets only overlap when they arrive at the same location and at the same time. When multiple high-amplitude wave packets with positive residual vehicle equivalent arrive at the target location simultaneously, a higher convergence peak is formed; when the wave packets arrive at different times, the overall waveform is relatively flat.

[0110] To ensure that the exit deduction conforms to vehicle conservation and avoids negative traffic flow, the waveforms of the exit deduction and total traffic flow should meet the following requirements:

[0111] The sum of vehicle equivalents of each batch of vehicles before export deductions Equal to the remaining vehicle equivalent after deduction Compared with the actual deduction of vehicle equivalent The sum of the total traffic flow waveforms at the target location must be equal to or equal to 0; meanwhile, the total traffic flow waveform at the target location must not be less than 0. Subsequent congestion determinations will use the total traffic flow waveform that satisfies the above vehicle equivalent conservation and non-negativity constraints.

[0112] Predicted speed and congestion assessment: For each effective forward vehicle batch wave packet arriving at the target location, the equivalent predicted speed can be calculated based on the wave packet intensity. Its formula is:

[0113] Where, q p (X,t) represents the intensity of the wave packet of the p-th effective base or entrance forward vehicle batch at the target location; u p This represents the translational speed of the vehicle batch; ε is a small positive number to prevent the denominator from being zero.

[0114] Let the traffic capacity of the road segment where the target location is located be... Define peak saturation Its formula is:

[0115] The waveform represents the total traffic flow at the target location. Effective traffic capacity is determined based on the number of effective lanes, basic single-lane traffic capacity, and current road operating conditions of the road segment where the target location is located. Basic single-lane traffic capacity can be obtained from road design data, preset traffic capacity parameters corresponding to the road type, or historical traffic data of the target location and its adjacent similar road segments. The spatial impact gain G(d) does not change the actual vehicle equivalent corresponding to the vehicle batch wave packet; it is only used to correct the peak saturation near highway entrances and exits to characterize the local impact of factors such as entrance merging, exit divergence, lane changes, and deceleration, which make the same traffic flow intensity more prone to congestion.

[0116] When the peak saturation reaches the threshold, the equivalent predicted velocity is lower than the velocity threshold, and this condition persists for a preset number of cycles, congestion is determined to exist.

[0117]

[0118]

[0119] in, This refers to the number of sampling periods during which the above state is maintained continuously; The preset peak saturation threshold is used; The preset speed threshold; A minimum number of duration cycles is preset. By combining the judgment of high peak, low speed, and duration, it is possible to avoid misjudging short-term high flow or high-speed operation as congestion.

[0120] The system searches for locations that meet congestion conditions along road mileage and in the future, and outputs the expected congestion location, expected time of occurrence, peak value, predicted speed, duration, and the source of the main contributing vehicle batch wave packets.

[0121] Refresh and Repeat Count Control: Assume the refresh period is... The next refresh time The formula is:

[0122] From the refresh time Start, until the next refresh time Previously, new data from roadside sensors and cameras were no longer added to the current base waveform; only newly generated entrance / exit cycle waveforms after the update were allowed as new boundary inputs. Therefore, when an entrance vehicle passes subsequent sensors, it will not be converted back to the base waveform and repeatedly superimposed.

[0123] arrive Afterwards, the system rereads all line-side equipment data, generates a new basic waveform field, and stops using the previous round of prediction states. The refresh uses a global replacement relationship:

[0124] Instead of adding the new measured state to the old prediction results.

[0125] In the formula, This represents the state of the predicted wave packet used within the k-th refresh cycle, up to the next refresh time. This represents the wave packet state reconstructed based on the latest full-line side detection data after reaching the (k+1)th refresh time.

[0126] Please see Figure 2 This application provides a schematic diagram of the structure of a road congestion prediction system based on the spatiotemporal propagation and superposition of vehicle equivalent wave packets, which is used to implement the road congestion prediction method provided in the above embodiments, including: The wave packet initialization module is used to convert the number of vehicles in each sampling period into time-continuous vehicle equivalent wave packets based on the number of vehicles and vehicle speeds obtained by the roadside detection equipment at a fixed sampling period, and to determine the translation speed of each wave packet along the road direction according to the vehicle speed of the corresponding period, thereby generating the initial wave packet. The blind spot completion module is used to interpolate the traffic conditions of adjacent detection devices for road sections with missing detection devices, and generate completion wave packets and their corresponding translation speeds. The wave packet propagation and boundary processing module is used to drive the initial wave packet and the completed wave packet to propagate along the road direction in the road spatiotemporal domain at their respective corresponding translation speeds, and to perform vehicle number addition or subtraction processing on the initial wave packet and the completed wave packet at the road boundary position according to the number of vehicles detected at the boundary. Wave packets that still contain vehicles after the vehicle number addition or subtraction processing continue to propagate. The wave packet superposition module is used to superimpose all wave packets that have been propagated to and passed through the vehicle number addition or subtraction processing at the target location to be predicted and its corresponding prediction time to obtain the total traffic flow waveform. The congestion determination module is used to determine whether congestion has occurred based on the peak characteristics of the total traffic flow waveform, the equivalent propagation speed of the wave packet, and the duration of the corresponding state, and outputs the congestion prediction result.

[0127] The technical solutions of this application have been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of this application is obviously not limited to these specific embodiments. Without departing from the principles of this application, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of this application.

[0128] The above are merely specific embodiments of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions within the technical scope disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A road congestion prediction method based on the spatiotemporal propagation and superposition of vehicle equivalent wave packets, characterized in that, include: Based on the number of vehicles and vehicle speeds obtained by roadside detection equipment at fixed sampling periods, the number of vehicles in each sampling period is converted into time-continuous vehicle equivalent wave packets, and the translation speed of each wave packet along the road direction is determined according to the vehicle speed of the corresponding period to generate the initial wave packet. For road sections lacking detection equipment, interpolation is performed using the traffic conditions of adjacent detection equipment to generate complete wave packets and their corresponding translation velocities; The initial wave packet and the completed wave packet are driven to propagate along the road direction in the road spatiotemporal domain at their respective corresponding translation speeds. At the road boundary position, the initial wave packet and the completed wave packet are processed to add or subtract vehicles based on the number of vehicles detected at the boundary. Wave packets that still contain vehicles after the vehicle addition or subtraction processing continue to propagate. At the target location to be predicted and its corresponding prediction time, all wave packets that have been propagated and still contain vehicles after the vehicle number has been increased or decreased are superimposed to obtain the total traffic flow waveform. Based on the peak characteristics of the total traffic flow waveform, the equivalent propagation speed of the wave packet, and the duration of the corresponding state, it is determined whether congestion has occurred, and the congestion prediction result is output.

2. The method according to claim 1, characterized in that, The step of converting the number of vehicles in each sampling period into time-continuous vehicle equivalent wave packets includes: The number of vehicles detected in each sampling period is used as the amplitude of the discrete signal to construct a periodic discrete signal centered at the sampling time. The periodic discrete signal is then convolved using a Gaussian kernel function to convert it into a time-continuous vehicle equivalent wave packet.

3. The method according to claim 1, characterized in that, The method of interpolating and generating complete wave packets and their corresponding translation velocities for road sections lacking detection equipment using the traffic conditions of adjacent detection equipment includes: Based on the vehicle density and vehicle speed of the nearest upstream and downstream detection points of the road segment lacking detection equipment, spatial linear interpolation is performed to obtain the interpolation density of each spatial unit within the road segment. The allocation weight of each spatial unit to the total number of vehicles on the road segment is determined based on the interpolation density and unit length of each spatial unit. Based on the allocation weights, the estimated total number of vehicles on the road segment is allocated to the corresponding spatial units, and the complete wave packet and its corresponding translation velocity are generated for each spatial unit.

4. The method according to claim 1, characterized in that, The road boundary location includes the road entrance and the road exit; the process of adding or deducting the number of vehicles includes: At the road entrance, the number of vehicles detected at the entrance is used to generate a new wave packet, which is then assigned a translational speed and added to the main line for propagation. At road exits, based on the number of vehicles detected at the exit, the wave packets that propagate to the exit are deducted according to the source matching or diversion ratio. The number of vehicles deducted does not exceed the number of vehicles currently contained in the wave packet, and wave packets that still contain vehicles after deduction continue to propagate at the original translation speed.

5. The method according to claim 4, characterized in that, The method of assigning translation speed includes: within the sampling period for generating the new wave packet, if V2X vehicle speed data exists, the average V2X vehicle speed within that period is taken as the translation speed of the new wave packet; if V2X vehicle speed data does not exist, the historical average speed corresponding to that sampling period is taken as the translation speed of the new wave packet.

6. The method according to claim 4, characterized in that, The vehicle quantity deduction process also includes: If the number of vehicles leaving the road exit is greater than the sum of the number of vehicles currently contained in all wave packets that have propagated to the exit, then the number of vehicles currently contained in all wave packets is reduced to zero, and the excess is recorded as a measurement deviation. Wave packets containing a negative number of vehicles are not generated.

7. The method according to claim 1, characterized in that, The process of superimposing all wave packets that have reached and passed through the vehicle count after the addition or subtraction of vehicle numbers to obtain the total traffic flow waveform is as follows: The intensity of all wave packets arriving at the same target location at the same prediction time is summed to obtain the total traffic waveform of that target location at that prediction time.

8. The method according to claim 1, characterized in that, The step of determining whether congestion has occurred based on the peak characteristics of the total traffic flow waveform, the equivalent propagation speed of the wave packet, and the duration of the corresponding state includes: if the ratio of the peak value of the total traffic flow waveform to the traffic capacity of the road segment reaches a preset threshold, the equivalent propagation speed of the wave packet is lower than a preset speed threshold, and the number of sampling periods in which the state is continuously maintained reaches a preset number of periods, then congestion is determined to have occurred; otherwise, congestion is determined not to have occurred.

9. The method according to claim 1, characterized in that, Also includes: When the interval since the last refresh reaches a preset refresh cycle, the initial wave packet and the complete wave packet are regenerated based on the latest detection data, and the currently propagating initial wave packet and the complete wave packet are replaced by the regenerated initial wave packet and the complete wave packet as a whole.

10. A road congestion prediction system based on the spatiotemporal propagation and superposition of vehicle equivalent wave packets, characterized in that, include: The wave packet initialization module is used to convert the number of vehicles in each sampling period into time-continuous vehicle equivalent wave packets based on the number of vehicles and vehicle speeds obtained by the roadside detection equipment at a fixed sampling period, and to determine the translation speed of each wave packet along the road direction according to the vehicle speed of the corresponding period, thereby generating the initial wave packet. The blind spot completion module is used to interpolate the traffic conditions of adjacent detection devices for road sections with missing detection devices, and generate completion wave packets and their corresponding translation speeds. The wave packet propagation and boundary processing module is used to drive the initial wave packet and the completed wave packet to propagate along the road direction in the road spatiotemporal domain at their respective corresponding translation speeds, and to perform vehicle number addition or subtraction processing on the initial wave packet and the completed wave packet at the road boundary position according to the number of vehicles detected at the boundary. Wave packets that still contain vehicles after the vehicle number addition or subtraction processing continue to propagate. The wave packet superposition module is used to superimpose all wave packets that have been propagated to and passed through the vehicle number addition or subtraction processing at the target location to be predicted and its corresponding prediction time to obtain the total traffic flow waveform. The congestion determination module is used to determine whether congestion has occurred based on the peak characteristics of the total traffic flow waveform, the equivalent propagation speed of the wave packet, and the duration of the corresponding state, and outputs the congestion prediction result.

Images