Traffic state reconstruction method and system based on sparse microscopic vehicle trajectory data
By using an adaptive smoothing spatiotemporal interpolation algorithm based on sparse microscopic vehicle trajectory data, the problems of high algorithm complexity and poor real-time performance in existing technologies are solved, achieving efficient reconstruction and real-time monitoring of traffic conditions, and improving the accuracy and generalization ability of traffic condition recognition.
Patent Information
- Application Number
- CN202410963339.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-18
- Publication Date
- 2024-11-12
AI Technical Summary
Existing traffic condition reconstruction technologies have high algorithm complexity, large computational resource requirements, poor real-time performance and generalization performance, making it difficult to achieve efficient traffic condition identification and monitoring.
An adaptive smoothing spatiotemporal interpolation algorithm based on sparse microscopic vehicle trajectory data is adopted. The algorithm is constructed by using discrete convolution function and weighted kernel function to calculate the velocity fields of free traffic and congested traffic respectively, and to calculate the degree of traffic congestion. The average speed is calculated by superimposing the final velocity fields to reconstruct the traffic state.
It improves the real-time performance and accuracy of traffic condition reconstruction, reduces computational complexity, has good generalization ability, is suitable for real-time traffic condition monitoring, and reduces maintenance costs.
Smart Images

Figure CN118942244A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of traffic state reconstruction, and in particular to a traffic state reconstruction method and system based on sparse microscopic vehicle trajectory data. Background Art
[0002] Traffic state recognition is a prerequisite for the implementation of traffic management measures. With the current increase in the number of motor vehicles and frequent traffic accidents, efficient and accurate traffic state recognition is of great significance to improving road capacity, alleviating traffic congestion and reducing traffic accidents. In today's traffic management and control, fixed detectors have been maturely and widely used in traffic state recognition, but they have problems such as limited coverage, high installation and maintenance costs, and cannot perceive the nonlinear problem of uneven temporal and spatial distribution of traffic flow. In recent years, mobile traffic flow detection technology has developed rapidly, and floating vehicle data can make real-time judgments on traffic status, but the accuracy and comprehensiveness of the collected information still have much room for improvement.
[0003] In the existing technology, there are methods of reconstructing traffic conditions at different locations and times by using spatiotemporal interpolation methods based on the vehicle speed detected by a stationary detector, or by integrating a multi-lane traffic flow model into the data assimilation framework of an extended Kalman filter. Although these two methods have improved processing accuracy, the calculations are too complex, and the real-time performance and response speed are limited. Some people have proposed a trajectory prediction model based on a bidirectional long short-term memory network. Although this model has high real-time performance, it requires a large amount of training data and is prone to overfitting problems. The cost of using high-performance computing equipment for a multi-sensor system is too high, and the maintenance is complex, making it difficult to promote and apply on a large scale. Summary of the invention
[0004] To this end, the technical problem to be solved by the present invention is to overcome the problems of high algorithm complexity, large computing resource requirements, poor real-time and generalization performance in the existing traffic state reconstruction technology, and provide a traffic state reconstruction method and system based on sparse microscopic vehicle trajectory data, which significantly improves the generalization ability and real-time performance of traffic state estimation.
[0005] In a first aspect, in order to solve the above problems, the present invention provides a traffic state reconstruction method based on sparse microscopic vehicle trajectory data, comprising:
[0006] S1. Extracting trajectory data in traffic flow through the ngsim dataset, traversing the trajectory data until interruption or missing occurs;
[0007] S2. construct an adaptive smooth spatiotemporal interpolation algorithm, divide the traffic flow into two traffic states based on the adaptive smooth spatiotemporal interpolation algorithm, the two traffic states include free traffic and congested traffic, and calculate the speed field of the free traffic and congested traffic; the basis of the adaptive smooth spatiotemporal interpolation algorithm for reconstructing the traffic state is a discrete convolution function, and the discrete convolution function includes a weighted kernel function;
[0008] S3. Calculate the degree of traffic congestion according to the speed field of the two traffic states;
[0009] S4. Calculate the average speed of the traffic flow according to the superposition of the speed field and the degree of traffic congestion, predict the traffic flow information of the traffic flow at different time and space positions through the average speed, and complete the traffic state reconstruction process.
[0010] In one embodiment of the present invention, the adaptive smoothing spatiotemporal interpolation algorithm performs spatiotemporal interpolation operation and adaptive smoothing operation through the discrete convolution function, wherein:
[0011] The weighted kernel function is defined based on the logarithmic function, and its formula is:
[0012]
[0013] Among them, the is the weighted kernel function, x is the spatial interval value, t is the time interval value, σ is the smoothing width in the spatial coordinate, τ is the smoothing width in the time coordinate, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i;
[0014] The weighted kernel function is normalized to obtain the sum of the discrete weights of all weighted kernel functions. The normalization formula is:
[0015]
[0016] Where N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i.
[0017] In one embodiment of the present invention, the weighted kernel function includes the following properties:
[0018] When the absolute values of the road section interval and the time interval are close to the boundary of the local area where the weighted kernel function has an impact, the value of the weighted kernel approaches zero;
[0019] When the road section interval and the time interval are at the origin of the local area where the weighted kernel function has an impact, the impact range of the weighted kernel is the largest;
[0020] The value of the weighted kernel decreases monotonically with the absolute value of the road section interval and the time interval.
[0021] In one embodiment of the present invention, calculating the velocity fields of two traffic states includes:
[0022] Construct velocity field calculation model:
[0023]
[0024] Wherein, V1 represents the space-time average velocity field; v i represents the velocity field corresponding to data point i; N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i;
[0025] The free traffic speed field is calculated based on the speed field model, and the formula is:
[0026]
[0027] The congested traffic speed field is calculated based on the speed field model, and the formula is:
[0028]
[0029] In the above two formulas, V free is the velocity field of the free traffic, V cong is the velocity field of the congested traffic, V free is the discrete weight sum in the free traffic, N cong is the discrete weight sum in the congested traffic, c free is the propagation velocity field in the free traffic, c cong is the propagation velocity field in the congested traffic, N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i, v i Represents the velocity field corresponding to data point i.
[0030] In one embodiment of the present invention, the velocity field calculation model performs smoothing processing on the interruption or missing part based on an exponential function low-pass filter, and the range of the smoothing processing includes:
[0031] For the interpolation of detector data with an aggregation interval of 5s, the time coordinate smoothing width is smoothed over the range of observations within 2.5s-5s;
[0032] Time fluctuations that are smaller than the smoothing width of the time coordinate;
[0033] Spatial fluctuations that are smaller than the smoothing width of the spatial coordinates.
[0034] In one embodiment of the present invention, the traffic congestion degree is calculated based on an s-shaped nonlinear function, and the formula is:
[0035]
[0036] Wherein, w is the degree of traffic congestion, V c is the threshold between the free traffic and the congested traffic, ΔV is the transition width between the free traffic and the congested traffic, V * is the predicted value of free traffic and congested traffic, x is the spatial interval value, and t is the time interval value.
[0037] In one embodiment of the present invention, the value of the traffic congestion degree is calculated based on the speed fields of the two traffic states, wherein the predicted values V of the free traffic and the congested traffic are * (x, t) = min[V free , V cong ]; and the value of the traffic congestion level is 0-1;
[0038] In one embodiment of the present invention, the average speed is calculated based on a nonlinear adaptive speed filter, and the formula is:
[0039] V(x,t)=w(x,t)V cong (x,t)+[1-w(x,t)]V free (x,t)
[0040] Wherein, V is the average speed, V free is the velocity field of the free traffic, V cong is the speed field of the congested traffic, W is the degree of traffic congestion, x is the spatial interval value, and t is the time interval value.
[0041] On the other hand, the present invention also provides a traffic state reconstruction system based on sparse microscopic vehicle trajectory data, comprising:
[0042] A trajectory data acquisition module extracts trajectory data in the traffic flow through the ngsim dataset and traverses the trajectory data until interruption or missing occurs;
[0043] A velocity field calculation module is used to construct an adaptive smooth space-time interpolation algorithm, and to divide the traffic flow into two traffic states based on the adaptive smooth space-time interpolation algorithm, wherein the two traffic states include free traffic and congested traffic, and to calculate the velocity fields of the free traffic and congested traffic; the basis for the adaptive smooth space-time interpolation algorithm to perform the traffic state reconstruction process is a discrete convolution function, and the discrete convolution function includes a weighted kernel function;
[0044] A traffic state classification module, configured to calculate the degree of traffic congestion according to the speed fields of the two traffic states;
[0045] The average speed calculation module is configured to calculate the average speed of the traffic flow according to the superposition of the speed field and the traffic congestion degree, predict the traffic flow information of the traffic flow at different time and space positions through the average speed, and complete the traffic state reconstruction process.
[0046] In one embodiment of the present invention, the velocity field calculation model is constructed as follows:
[0047]
[0048] Wherein, V1 represents the space-time average velocity field; v i represents the velocity field corresponding to data point i; N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i;
[0049] The free traffic speed field is calculated based on the speed field model, and the formula is:
[0050]
[0051] The congested traffic speed field is calculated based on the speed field model, and the formula is:
[0052]
[0053] In the above two formulas, V free is the velocity field of the free traffic, V cong is the velocity field of the congested traffic, N free is the discrete weight sum in the free traffic, N cong is the discrete weight sum in the congested traffic, c freeis the propagation velocity field in the free traffic, c cong is the propagation velocity field in the congested traffic, N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i, v i Represents the velocity field corresponding to data point i.
[0054] The above technical solution of the present invention has the following beneficial effects compared with the prior art:
[0055] The traffic state reconstruction method and system based on sparse microscopic vehicle trajectory data of the present invention respectively calculate the final velocity fields of free traffic and congested traffic states and calculate the degree of traffic congestion through an adaptive smoothing space-time interpolation algorithm, and calculate the average speed through the superposition of the final velocity fields to reconstruct the traffic state; the application of the adaptive smoothing space-time interpolation algorithm enables the accurate prediction of the traffic state even when the trajectory data contains interruptions / missing data, and maintains a high prediction accuracy; the adaptive smoothing method included in the adaptive smoothing space-time interpolation algorithm considers the traffic characteristics of the two traffic states of free traffic and congested traffic, and can accurately calculate the traffic flow data under different traffic states, and has good generalization ability; the adaptive smoothing space-time interpolation algorithm can perform fast interpolation and smoothing processing in time and space, is suitable for real-time traffic state monitoring, and effectively improves the real-time performance of traffic state reconstruction. BRIEF DESCRIPTION OF THE DRAWINGS
[0056] In order to make the content of the present invention more clearly understood, the present invention is further described in detail below according to specific embodiments of the present invention in conjunction with the accompanying drawings, wherein
[0057] Figure 1 It is a flow chart of a traffic state reconstruction method based on sparse microscopic vehicle trajectory data in a preferred embodiment of the present invention;
[0058] Figure 2 A definition diagram of sparse microscopic trajectories in a traffic state reconstruction method based on sparse microscopic vehicle trajectory data in a preferred embodiment of the present invention;
[0059] Figure 3 A vehicle spatiotemporal trajectory diagram in a traffic state reconstruction method based on sparse microscopic vehicle trajectory data in a preferred embodiment of the present invention;
[0060] Figure 4 A schematic diagram of a weighted kernel function in a preferred embodiment of the present invention;
[0061] Figure 5It is a module diagram of a traffic state reconstruction system based on sparse microscopic vehicle trajectory data in a preferred embodiment of the present invention;
[0062] Figure 6 This is a thermal comparison diagram of the real lane data and predicted data of the I-80 dataset using the traffic state reconstruction method based on sparse microscopic vehicle trajectory data in the preferred embodiment of the present invention.
[0063] Description of the accompanying drawings in the specification: 100, trajectory data acquisition module; 200, speed field calculation module; 300, traffic state classification module; 400, average speed calculation module. DETAILED DESCRIPTION
[0064] The present invention is further described below in conjunction with the accompanying drawings and specific embodiments so that those skilled in the art can better understand the present invention and implement it, but the embodiments are not intended to limit the present invention.
[0065] Example 1
[0066] Reference Figure 1 As shown, the present application provides a traffic state reconstruction method based on sparse microscopic vehicle trajectory data, and the specific steps are as follows:
[0067] S1. Extract the trajectory data in the traffic flow through the ngsim (Next Generation Simulation) data set, and check whether there are interruptions or omissions in the trajectory data, the trajectory data at least includes vehicle number, data frame number, vehicle length, vehicle width, vehicle speed, vehicle acceleration, headway and time interval; if there are no interruptions or omissions, continue to check the trajectory data in the traffic flow through the ngsim data set until the interruptions or omissions are found; after finding the interruptions or omissions, perform the subsequent steps of traffic state reconstruction; the ngsim data set contains rich detailed road data, the collection time for each road is up to 45 minutes, and the data sampling interval is 0.1 seconds, so the data is rich and can provide detailed traffic flow information; and the data format in the ngsim data set is standardized and easy to process and analyze.
[0068] like Figure 2 As shown, the yellow vehicle trajectory and the black vehicle trajectory represent sparse microscopic vehicle trajectories, the yellow vehicle trajectory represents an interrupted trajectory, the black vehicle trajectory represents a missing trajectory, and the green vehicle trajectory represents a complete trajectory.
[0069] like Figure 3 As shown, the vehicle trajectory is represented by the blue curve, which includes the complete vehicle trajectory, and some vehicle trajectories are interrupted or missing.
[0070] S2. Construct an adaptive smooth space-time interpolation algorithm. The basis of the adaptive smooth space-time interpolation algorithm for the traffic state reconstruction processing is a discrete convolution function, and the discrete convolution function includes a weighted kernel function. The adaptive smooth space-time interpolation algorithm can fill the interruptions / missing parts in the trajectory data, making the trajectory data more complete, which is helpful for comprehensive traffic state prediction, and the adaptive smooth space-time interpolation algorithm can adjust its parameters and models according to data characteristics to adapt to different application scenarios to improve generalization ability.
[0071] S3. Calculate the degree of traffic congestion according to the speed field of the two traffic states;
[0072] S4. Calculate the average speed of the traffic flow according to the superposition of the speed field and the degree of traffic congestion, predict the traffic flow information of the traffic flow at different time and space positions through the average speed, and complete the traffic state reconstruction process.
[0073] The traffic state reconstruction method based on micro-trajectory data described in the present invention can perform fast interpolation and smoothing in time and space, is suitable for real-time traffic state monitoring, and significantly improves the accuracy and real-time performance of traffic state reconstruction. By dividing the traffic state into free traffic and congested traffic, the model has good generalization ability. In addition, the model of this method is simple and the computational complexity is low, which effectively improves the efficiency of traffic state reconstruction and reduces maintenance costs.
[0074] Specifically, the adaptive smoothing space-time interpolation algorithm is constructed based on the space-time interpolation method and the adaptive smoothing process, and the adaptive smoothing space-time interpolation algorithm performs traffic state reconstruction processing by performing convolution operation through the discrete convolution function;
[0075] The weighted kernel function is defined based on the logarithmic function, and its formula is:
[0076]
[0077] Among them, the is the weighted kernel function, x is the spatial interval value, t is the time interval value, σ is the smoothing width in the spatial coordinate, τ is the smoothing width in the time coordinate, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i
[0078] like Figure 4As shown in the figure, it shows how to reflect the spatiotemporal changes of traffic status through different weighted kernel functions in the spatiotemporal coordinate system. φ0(free) represents the weighted kernel function under free traffic status; φ0(congested) represents the weighted kernel function under congested traffic status; t0 represents the spatiotemporal interval value at the origin; x0, the spatial interval value at the origin; 2σ, the spatial smoothing width; 2τ, the temporal smoothing width; x i+1 Indicates the spatial interval value when the data point is i+1; x i Represents the spatial interval value when the data point is i; t, the time interval value; the green triangle and yellow dot in the figure represent the locations of the observed data at different points in time and space, respectively; it can be seen that the two weighted kernel functions have different influence ranges in time and space. In the free traffic state, the influence range of the weighted kernel is larger, indicating that in the free traffic state, vehicles propagate at a higher speed and have a wider influence range; the influence range in the congested traffic state is smaller, indicating that in the congested traffic state, vehicles propagate at a lower speed and have a narrower influence range.
[0079] The weighted kernel is normalized to obtain the sum of the discrete weights of all weighted kernels, and the weighted kernel is normalized to determine the sum of the weights of all weighted kernels to be 1, so that the relative proportion and stability of the data can be maintained when performing weighted operations. In the adaptive smoothing spatiotemporal interpolation algorithm, normalizing the weighted kernel can make the algorithm more stable and reliable when processing data of different scales and densities, so as to improve the stability of the algorithm. The normalization formula is:
[0080]
[0081] Where N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i.
[0082] Furthermore, the weighted kernel function has locality, including that when the absolute value of the road section interval and time interval is close to the boundary of the local area where the weighted kernel function has an impact, the value of the weighted kernel approaches zero; when the road section interval and time interval are at the origin of the local area where the weighted kernel function has an impact, the weighted kernel has a maximum value; the value of the weighted kernel decreases monotonically with the absolute value of the road section interval and time interval. Due to the locality of the weighted kernel, the weighted kernel takes effect in the area near the data points measured by the weighted kernel, so only these data points need to be considered during calculation, which effectively reduces the amount of calculation and significantly improves the efficiency of the algorithm in processing trajectory data.
[0083] Specifically, a velocity field calculation model is constructed. The velocity field calculation model is constructed by a low-pass filter. The low-pass filter can smooth the signal and make the signal more continuous and stable by removing high-frequency noise and rapidly changing parts. The use of a low-pass filter can effectively smooth the fluctuations in space and time, and extract the basic trends and long-term changes in traffic flow, so as to more accurately calculate the final velocity field value.
[0084] Furthermore, the velocity field calculation model smoothes the interruption or missing parts based on an exponential function low-pass filter, and the range of the smoothing includes, for the interpolation of detector data with an aggregation interval of 5s, the time coordinate smoothing width is smoothed within the range of 2.5s-5s for observation values; time fluctuations smaller than the time coordinate smoothing width; and spatial fluctuations smaller than the space coordinate smoothing width.
[0085] Specifically, the velocity field calculation model is:
[0086]
[0087] Wherein, V1 represents the space-time average velocity field; v i represents the velocity field corresponding to data point i; N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i;
[0088] The free traffic speed field is calculated based on the speed field model, and the formula is:
[0089]
[0090] The congested traffic speed field is calculated based on the speed field model, and the formula is:
[0091]
[0092] In the above two formulas, V free is the velocity field of the free traffic, V cong is the velocity field of the congested traffic, N free is the discrete weight sum in the free traffic, N cong is the discrete weight sum in the congested traffic, c free is the propagation velocity field in the free traffic, c cong is the propagation velocity field in the congested traffic, N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x iis the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i, v i Represents the velocity field corresponding to data point i.
[0093] Furthermore, the degree of traffic congestion is calculated based on the Sigmoid nonlinear function. The Sigmoid function has a smooth curve characteristic, which enables it to map subtle changes in the input to subtle changes in the output. In traffic status prediction, this smoothness helps to improve the accuracy and robustness of the model, allowing the model to better capture small fluctuations and changes in traffic flow, thereby achieving more accurate calculation of traffic status.
[0094] The formula for calculating the traffic congestion degree is:
[0095]
[0096] Wherein, w is the degree of traffic congestion, V c is the threshold between the free traffic and the congested traffic, ΔV is the transition width between the free traffic and the congested traffic, V * is the predicted value of free traffic and congested traffic, x is the spatial interval value, and t is the time interval value.
[0097] The value of the traffic congestion degree is calculated based on the speed fields of the two traffic states, wherein the predicted values V of the free traffic and the congested traffic are * (x, t) = min[V free , V cong ]; and the value of the traffic congestion degree is 0-1. When the traffic congestion degree is 0, the traffic flow interference speed is lower than the local speed of the vehicle flow, the traffic flow interference is less, and the proximity between the driving trajectories of vehicles is low. Therefore, the impact between road users is small and the traffic congestion degree is weak; when the congestion degree is 1, the traffic flow interference propagates in reverse, the traffic flow interference increases significantly, the driving trajectories of road users frequently intersect or approach each other, resulting in enhanced mutual influence and interaction, the traffic flow shows chaotic and unstable characteristics, and the local vehicle flow speed is slow or stagnant, and it is judged that the traffic congestion degree is strong.
[0098] Furthermore, the average speed is calculated based on a nonlinear adaptive speed filter; the average speed is one of the important parameters for describing the state of traffic flow. By calculating the average speed of the traffic flow, the traffic conditions in the area can be intuitively reflected to reconstruct the traffic state;
[0099] The formula is:
[0100] V(x,t)=w(x,t)V cong (x,t)+[1-w(x,t)]Vfree (x,t)
[0101] Wherein, V is the average speed, V free is the velocity field of the free traffic, V cong is the speed field of the congested traffic, w is the degree of traffic congestion, x is the spatial interval value, and t is the time interval value.
[0102] Example 2
[0103] like Figure 5 As shown, the present application provides a traffic state reconstruction system based on sparse microscopic vehicle trajectory data, comprising:
[0104] The trajectory data acquisition module 100 extracts the trajectory data in the traffic flow through the ngsim data set, and traverses the trajectory data until interruption or missing occurs;
[0105] The velocity field calculation module 200 constructs an adaptive smooth spatiotemporal interpolation algorithm, divides the traffic flow into two traffic states based on the adaptive smooth spatiotemporal interpolation algorithm, and calculates the velocity fields of the free traffic and the congested traffic; the basis for the adaptive smooth spatiotemporal interpolation algorithm to perform the traffic state reconstruction process is a discrete convolution function, and the discrete convolution function includes a weighted kernel function;
[0106] A traffic state classification module 300, which is configured to calculate the degree of traffic congestion according to the speed fields of the two traffic states;
[0107] The average speed calculation module 400 is configured to calculate the average speed of the traffic flow according to the superposition of the speed field and the traffic congestion degree, predict the traffic flow information of the traffic flow at different time and space positions through the average speed, and complete the traffic state reconstruction process.
[0108] Specifically, the average speed calculated by the adaptive smoothing spatiotemporal interpolation algorithm is the superposition of the average speed of free traffic and the average speed of congested traffic, and the formula is:
[0109] V(x,t)=w(x,t)V cong (x,t)+[1-w(x,t)]V free (x,t)
[0110] In the above formula, w is the degree of traffic congestion, V c is the threshold between the free traffic and the congested traffic, ΔV is the transition width between the free traffic and the congested traffic; wherein ΔV determines V c The transition width around the vehicle is set to 10 km / h, V c The parameter is 50km / h.
[0111] The formula for calculating the degree of traffic congestion is:
[0112]
[0113] Wherein, w is the degree of traffic congestion, V c is the threshold between the free traffic and the congested traffic, ΔV is the transition width between the free traffic and the congested traffic, V * is the predicted value of free traffic and congested traffic, x is the spatial interval value, and t is the time interval value.
[0114] Further,
[0115] The free traffic velocity field is calculated as follows:
[0116]
[0117] The congested traffic speed field is calculated as follows:
[0118]
[0119] In the above two formulas, V frree is the velocity field of the free traffic, V cong is the velocity field of the congested traffic, N free is the discrete weight sum in the free traffic, N cong is the discrete weight sum in the congested traffic, c free is the propagation velocity field in the free traffic, c cong is the propagation velocity field in the congested traffic, N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i, v i Represents the velocity field corresponding to data point i.
[0120] In which, set c free The parameters are 70km / h; c cong The parameter is -13km / h.
[0121] Test Example 1
[0122] like Figure 6As shown, the traffic adaptive smoothing method spatiotemporal interpolation algorithm proposed in the present invention is verified using the I-80 highway section data of the ngsim (Next Generation Simulation) data set. It can be seen that the adaptive smoothing method spatiotemporal interpolation algorithm can more accurately predict the traffic status of highways, and can clearly identify the process of propagation and change of traffic congestion and smoothness. The heat map of the prediction results is close to the heat map presented by the real data, which fully proves the generalization ability and prediction accuracy of the method for estimating the traffic status of interrupted / missing vehicle trajectories.
[0123] The traffic state reconstruction method and system based on sparse microscopic vehicle trajectory data of the present invention respectively calculate the final velocity fields of free traffic and congested traffic states and calculate the degree of traffic congestion through an adaptive smoothing space-time interpolation algorithm, and calculate the average speed through the superposition of the final velocity fields to reconstruct the traffic state; the application of the adaptive smoothing space-time interpolation algorithm enables the accurate prediction of the traffic state even when the trajectory data contains interruptions / missing data, and maintains a high prediction accuracy; the adaptive smoothing method included in the adaptive smoothing space-time interpolation algorithm considers the traffic characteristics of the two traffic states of free traffic and congested traffic, and can accurately calculate the traffic flow data under different traffic states, and has good generalization ability; the adaptive smoothing space-time interpolation algorithm can perform fast interpolation and smoothing processing in time and space, is suitable for real-time traffic state monitoring, and effectively improves the real-time performance of traffic state reconstruction.
[0124] Those skilled in the art will appreciate that the embodiments of the present application may be provided as methods, systems, or computer program products. Therefore, the present application may adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment in combination with software and hardware. Moreover, the present application may adopt the form of a computer program product implemented in one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that include computer-usable program code.
[0125] The present application is described with reference to the flowcharts and / or block diagrams of the methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or box in the flowchart and / or block diagram, as well as the combination of the processes and / or boxes in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.
[0126] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.
[0127] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.
[0128] Obviously, the above embodiments are merely examples for clear explanation and are not intended to limit the implementation methods. For those skilled in the art, other different forms of changes or modifications can be made based on the above description. It is not necessary and impossible to list all the implementation methods here. The obvious changes or modifications derived from these are still within the protection scope of the invention.
Claims
1. A traffic state reconstruction method based on sparse microscopic vehicle trajectory data, characterized in that: The following steps are included: S1. Extracting trajectory data in traffic flow through the ngsim dataset, traversing the trajectory data until interruption or missing occurs; S2. construct an adaptive smooth spatiotemporal interpolation algorithm, divide the traffic flow into two traffic states based on the adaptive smooth spatiotemporal interpolation algorithm, the two traffic states include free traffic and congested traffic, and calculate the speed field of the free traffic and congested traffic; the basis of the adaptive smooth spatiotemporal interpolation algorithm for reconstructing the traffic state is a discrete convolution function, and the discrete convolution function includes a weighted kernel function; S3. Calculate the degree of traffic congestion according to the speed field of the two traffic states; S4. Calculate the average speed of the traffic flow according to the superposition of the speed field and the degree of traffic congestion, predict the traffic flow information of the traffic flow at different time and space positions through the average speed, and complete the traffic state reconstruction process.
2. The traffic state reconstruction method based on sparse microscopic vehicle trajectory data according to claim 1 is characterized in that: Constructing the adaptive smoothing spatiotemporal interpolation algorithm includes: The adaptive smoothing spatiotemporal interpolation algorithm performs spatiotemporal interpolation and adaptive smoothing operations through the discrete convolution function, wherein: The weighted kernel function is defined based on the logarithmic function, and its formula is: in, is the weighted kernel function, x is the spatial interval value, t is the time interval value, σ is the smoothing width in the spatial coordinate, τ is the smoothing width in the time coordinate, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i The weighted kernel function is normalized to obtain the sum of the discrete weights of all weighted kernel functions. The normalization formula is: Where N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i.
3. The traffic state reconstruction method based on sparse microscopic vehicle trajectory data according to claim 1 is characterized in that: The weighted kernel includes the following properties: When the absolute values of the road section interval and the time interval are close to the boundary of the local area where the weighted kernel function has an impact, the value of the weighted kernel approaches zero; When the road section interval and the time interval are at the origin of the local area where the weighted kernel function has an impact, the impact range of the weighted kernel is the largest; The value of the weighted kernel decreases monotonically with the absolute value of the road section interval and the time interval.
4. The traffic state reconstruction method based on sparse microscopic vehicle trajectory data according to claim 1 is characterized in that: The calculating the speed field of the free traffic and the congested traffic comprises: Construct velocity field calculation model: Wherein, V1 represents the space-time average velocity field; v i represents the velocity field corresponding to data point i; N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i; The free traffic speed field is calculated based on the speed field model, and the formula is: The congested traffic speed field is calculated based on the speed field model, and the formula is: In the above two formulas, V free is the velocity field of the free traffic, V cong is the velocity field of the congested traffic, N free is the discrete weight sum in the free traffic, N cong is the discrete weight sum in the congested traffic, c free is the propagation velocity field in the free traffic, c cong is the propagation velocity field in the congested traffic, N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i, v i Represents the velocity field corresponding to data point i.
5. The traffic state reconstruction method based on sparse microscopic vehicle trajectory data according to claim 4 is characterized in that: The velocity field calculation model comprises: The velocity field calculation model performs smoothing processing on the interruption or missing part based on an exponential function low-pass filter, and the range of the smoothing processing includes: For the interpolation of detector data with an aggregation interval of 5s, the time coordinate smoothing width is smoothed over the range of observations within 2.5s-5s; Time fluctuations that are smaller than the smoothing width of the time coordinate; Spatial fluctuations that are smaller than the smoothing width of the spatial coordinates.
6. The traffic state reconstruction method based on sparse microscopic vehicle trajectory data according to claim 1 is characterized in that: The calculation of traffic congestion degree includes: The traffic congestion degree is calculated based on the s-type nonlinear function, and the formula is: Wherein, w is the degree of traffic congestion, V c is the threshold between the free traffic and the congested traffic, ΔV is the transition width between the free traffic and the congested traffic, V * is the predicted value of free traffic and congested traffic, x is the spatial interval value, and t is the time interval value.
7. The traffic state reconstruction method based on sparse microscopic vehicle trajectory data according to claim 6 is characterized in that: The degree of traffic congestion, include: Among them, the predicted values V of free traffic and congested traffic * (x, t) = min[V free , V cong ]; and the value of the traffic congestion level is 0-1.
8. The traffic state reconstruction method based on sparse microscopic vehicle trajectory data according to claim 1 is characterized in that: The average speed includes: The average speed is calculated based on the nonlinear adaptive speed filter, and the formula is: V(x,t)=w(x,t)V cong (x,t)+[1-w(x,t)]V free (x,t) Wherein, V is the average speed, V free is the velocity field of the free traffic, V cong is the speed field of the congested traffic, w is the degree of traffic congestion, x is the spatial interval value, and t is the time interval value.
9. A traffic state reconstruction system based on sparse microscopic vehicle trajectory data, characterized in that: include: A trajectory data acquisition module extracts trajectory data in the traffic flow through the ngsim dataset and traverses the trajectory data until interruption or missing occurs; A velocity field calculation module is used to construct an adaptive smooth space-time interpolation algorithm, and to divide the traffic flow into two traffic states based on the adaptive smooth space-time interpolation algorithm, wherein the two traffic states include free traffic and congested traffic, and to calculate the velocity fields of the free traffic and congested traffic; the basis for the adaptive smooth space-time interpolation algorithm to perform the traffic state reconstruction process is a discrete convolution function, and the discrete convolution function includes a weighted kernel function; A traffic state classification module, configured to calculate the degree of traffic congestion according to the speed fields of the two traffic states; The average speed calculation module is configured to calculate the average speed of the traffic flow according to the superposition of the speed field and the traffic congestion degree, predict the traffic flow information of the traffic flow at different time and space positions through the average speed, and complete the traffic state reconstruction process.
10. The traffic state reconstruction system based on sparse microscopic vehicle trajectory data according to claim 9, characterized in that: The velocity field calculation module comprises: Construct velocity field calculation model: Wherein, V1 represents the space-time average velocity field; v i represents the velocity field corresponding to data point i; N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i; The free traffic speed field is calculated based on the speed field model, and the formula is: The congested traffic speed field is calculated based on the speed field model, and the formula is: In the above two formulas, V free is the velocity field of the free traffic, V cong is the velocity field of the congested traffic, N free is the discrete weight sum in the free traffic, N cong is the discrete weight sum in the congested traffic, c free is the propagation velocity field in the free traffic, c cong is the propagation velocity field in the congested traffic, N is the discrete weight sum of all weighted kernels, is the weighted kernel function, x is the spatial interval value, t is the time interval value, x i is the spatial interval value corresponding to data point i, t i is the time interval value corresponding to data point i, v i Represents the velocity field corresponding to data point i.
Citation Information
Cited By
Construction method and system of territorial space planning database and medium
CN120994756A
