CN114049764A - Traffic simulation method and system based on convolution long-time and short-time memory neural network - Google Patents
Traffic simulation method and system based on convolution long-time and short-time memory neural network Download PDFInfo
- Publication number
- CN114049764A CN114049764A CN202111264070.4A CN202111264070A CN114049764A CN 114049764 A CN114049764 A CN 114049764A CN 202111264070 A CN202111264070 A CN 202111264070A CN 114049764 A CN114049764 A CN 114049764A
- Authority
- CN
- China
- Prior art keywords
- traffic
- neural network
- memory neural
- vehicle
- time
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Images
Classifications
-
- G—PHYSICS
- G08—SIGNALLING
- G08G—TRAFFIC CONTROL SYSTEMS
- G08G1/00—Traffic control systems for road vehicles
- G08G1/01—Detecting movement of traffic to be counted or controlled
- G08G1/0104—Measuring and analyzing of parameters relative to traffic conditions
- G08G1/0125—Traffic data processing
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/044—Recurrent networks, e.g. Hopfield networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Engineering & Computer Science (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Biomedical Technology (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Chemical & Material Sciences (AREA)
- Analytical Chemistry (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a traffic simulation method based on a convolution long-time memory neural network, which comprises the following steps: acquiring a traffic flow video in the simulation area, and acquiring track data of all vehicles in the simulation area from the traffic flow video by an image identification method; preprocessing vehicle track data to obtain a traffic state grid map of a simulation area; taking the traffic state grid diagram as the input of the convolution long-time and short-time memory neural network, and carrying out parameter training on the convolution long-time and short-time memory neural network to obtain a trained convolution long-time and short-time memory neural network; acquiring a traffic state grid diagram based on the simulation area, and inputting the traffic state grid diagram into a trained convolution long-time memory neural network to acquire a traffic simulation result; and acquiring a traffic state grid map based on the simulation scene, and inputting the sampled traffic state grid map into a neural network to acquire a simulation result. Thereby obviously improving the precision of the microscopic traffic simulation.
Description
Technical Field
The invention relates to the technical field of intelligent traffic simulation, in particular to a traffic simulation method and system based on a convolution long-time memory neural network.
Background
With the vigorous development of economy in China, motor vehicles gradually enter families of common people, and the year-by-year explosive growth is reflected on the quantity keeping indexes of automobiles in everyone. The problem of urban traffic congestion is also becoming more and more significant. However, due to the large number of city components and the large number of individual influencing factors, the traffic characteristics are difficult to accurately describe by a centralized method. With the development of data science and the popularization of microcomputers, traffic simulation increasingly becomes an important technology for modeling urban traffic characteristics. The traffic simulation can effectively simulate the expected traffic running state in a given environment to find a traffic jam point in advance, and can also be used for screening a planning scheme through visual simulation data comparison, so that the urban traffic policy is powerfully promoted to be formulated. The microscopic simulation is used as a simulation means for simulating the microscopic running process of the traffic individuals in a non-centralized manner, and has good modeling and depicting capabilities aiming at problems occurring in local traffic scenes. The microscopic simulation simulates the traffic running state of non-integrated individuals so as to reproduce the traffic problems in the actual traffic scene, and observes the corresponding control measure effect by assuming the traffic policy so as to realize traffic characterization and quantitative evaluation of the traffic policy.
The existing traffic microscopic simulation software, including Paramics, VISSIM and the like, adopts an analytical model simulation method, and simulates the microscopic driving behavior of vehicles in a real scene by arranging a following model and a lane changing model in each vehicle individual. However, because each individual vehicle has differences, the high nonlinearity in the operation state decision cannot be accurately depicted through a traditional mathematical analysis model, the method often causes that the real traffic problem cannot be accurately reproduced, the simulation result has partial differences from the actual situation on the microscopic representation, and the solution made based on the microscopic simulation result is difficult to accurately take effect in reality. Similarly, the invention patents CN202110418191.3 and CN202110281113.3 disclose methods for calibrating mathematical analysis models in two traffic scenarios, namely, a circular intersection and an express intersection, in microscopic simulation, and the invention patent CN202010665035.2 discloses a microscopic traffic simulation method for updating states by applying different mathematical analysis models in heterogeneous traffic flow. Generally speaking, the existing research focuses on the use of a more accurate mathematical analysis model to perform microscopic traffic simulation, but neglects the problem that it is fundamentally difficult to accurately reproduce the microscopic traffic problems occurring in the actual traffic scene. Therefore, the adaptability of the prior art to the micro traffic simulation in practical application is considered to be insufficient, and a new method capable of accurately completing the micro traffic simulation is not slow.
Disclosure of Invention
The invention aims to overcome the defects of the prior art and provides a traffic simulation method and system based on a convolution long-time and short-time memory neural network, which are oriented to the background of traffic big data, apply the advantages of artificial intelligence based on the basis of mass data and utilize the convolution long-time and short-time memory neural network to deeply mine the highly nonlinear characteristics in traffic characteristics, thereby realizing accurate simulation of real traffic problems.
The invention adopts the following technical scheme for solving the technical problems:
the invention provides a traffic simulation method based on a convolution long-time memory neural network, which comprises the following steps:
s1, obtaining a traffic flow video in the simulation area, and obtaining track data of all vehicles in the simulation area from the traffic flow video through an image recognition method;
step S2, preprocessing the vehicle track data to obtain a traffic state grid map of the simulation area;
step S3, taking the traffic state grid diagram as the input of the convolution long-time and short-time memory neural network, and carrying out parameter training on the convolution long-time and short-time memory neural network to obtain a trained convolution long-time and short-time memory neural network;
and S4, acquiring a traffic state grid map by adopting the methods of the steps S1-S2 based on the simulation area, and inputting the traffic state grid map into the trained convolutional long-time memory neural network to acquire a traffic simulation result.
As a further optimization scheme of the traffic simulation method based on the convolution long-time memory neural network, in step S1, a traffic flow video is shot by an unmanned aerial vehicle.
As a further optimization scheme of the traffic simulation method based on the convolution length-time memory neural network, in step S1, the vehicle trajectory data refers to a current timestamp, a number of a road section where the vehicle enters the sample, a number of a lane where the vehicle is currently located, a longitudinal position where the vehicle is currently located, and a length of the vehicle.
As a further optimization scheme of the traffic simulation method based on the convolution duration memory neural network, in step S2, the vehicle trajectory data preprocessing method includes dividing vehicle trajectory data into a plurality of data sets based on current timestamp information, dividing a simulation area into grids of 2 m × 1 lanes, and dividing vehicle trajectory data in each data set into corresponding grids based on the serial number of the lane where the vehicle is currently located, the current longitudinal position of the vehicle relative to the starting point of the traffic flow video, and the length of the vehicle, so as to obtain a traffic state grid map.
As a further optimization scheme of the traffic simulation method based on the convolution long-and-short term memory neural network, in step S3, the process of performing parameter training on the convolution long-and-short term memory neural network includes:
dividing each M continuous-time traffic state grid graphs into one round based on time continuity, wherein M grids are selected from the M grids as learning samples, n grids are selected as result calibration real value graphs, and M is n + M;
presetting parameters of a convolutional long-time memory neural network;
inputting m traffic state grid graphs with continuous time as a convolution long-and-short time memory neural network, and performing feed-forward prediction through the convolution long-and-short time memory neural network to obtain n future traffic state grid graphs;
and calculating a loss function of the n predicted traffic state grid graphs and the result calibration real value graph, and performing iterative optimization on the parameters through a back propagation algorithm until the parameters are trained repeatedly for a set number of times or the calculated loss value reaches a set convergence condition, namely completing parameter training.
As a further optimization scheme of the traffic simulation method based on the convolution long-time memory neural network, M is 25, M is 20, and n is 5.
As a further optimization scheme of the traffic simulation method based on the convolution long-time and short-time memory neural network, a loss function expression designed in the convolution long-time and short-time memory neural network is as follows:
C1＝(K1L)2
C2＝(K2L)2
wherein Los (x, y) represents a loss function; n represents all pixel points in the traffic state grid diagram; x is the number ofiRepresenting the pixel value of the ith point in the traffic state grid diagram; w is aiRepresenting the weight of the ith point pixel value in the traffic state grid diagram; mu.sxRepresenting the average value of all pixels in the result calibration real value map; mu.syRepresenting the average value of all the pixels in the traffic grid map obtained by prediction; sigmaxRepresenting the variance of the pixel values in the resulting calibration true value map; sigmayRepresenting the variance of pixel values in the traffic grid map obtained by prediction; sigmaxyRepresenting the covariance of the pixel values of the resulting calibration true value map and the predicted traffic grid map; k1Weights representing the resulting calibration true value map; k2Weights representing the traffic grid map obtained by prediction; l is the gray value dynamic range; c1Representation avoidance
A correction term of 0; c2Representation avoidance
A correction term of 0; α represents a first weight; β represents a second weight; p represents the result prediction probability value output by the convolution long-time and short-time memory neural network; softmax (p)iAnd representing a softmax activation function for converting the prediction probability value of the pixel value of the ith pixel point in the traffic state grid graph into a prediction result.
K, a further optimization scheme of the traffic simulation method based on the convolution long-time memory neural network1Taking 0.01, K20.03 is taken.
Based on the traffic simulation system based on the convolution long-time memory neural network, the traffic simulation system comprises:
the vehicle track module is used for acquiring vehicle track data;
and the data preprocessing module is used for dividing the vehicle track data into a plurality of data groups based on the current timestamp information, dividing the simulation area into grids of 2 m by 1 lanes, and dividing the vehicle track data in each data group into corresponding grids based on the serial number of the lane where the vehicle is currently located, the current longitudinal position of the vehicle relative to the starting point of the traffic flow video and the length of the vehicle, so as to obtain the traffic state grid map.
The training module is used for training the convolution duration memory neural network and performing iterative optimization on parameters; the training module comprises a feedforward prediction sub-module and a reverse optimization sub-module;
and the simulation module is used for memorizing the neural network by utilizing the convolution duration obtained by the training module and carrying out traffic simulation prediction based on the input sampled traffic state grid diagram.
Compared with the prior art, the invention adopting the technical scheme has the following technical effects:
(1) the traffic state is innovatively converted into a grid diagram form, perception of vehicles in a traditional model to surrounding vehicles is used as a convolution kernel, and therefore the traffic state evolution is effectively predicted by applying the advantages of the existing developed deep learning convolution neural network technology;
(2) secondly, a forgetting and memorizing structure in the network structure is utilized to realize long-time memory structure and avoid the problems of gradient explosion and gradient disappearance in the traditional neural network, thereby greatly improving the prediction accuracy of the neural network;
(3) under the background of big data, the traffic simulation method based on the convolution long-time memory neural network under the support of mass traffic data can obviously further explore the advantages of deep learning in a big data environment, and has obvious effect on improving the precision of microscopic traffic simulation.
Drawings
Fig. 1 is a flow chart of a traffic simulation method of a convolution long-and-short-term memory neural network according to an embodiment of the present invention.
Fig. 2 is a cut-out picture in a traffic flow video obtained by unmanned aerial vehicle shooting in an example.
FIG. 3 is an exemplary traffic state grid map derived by a vehicle trajectory preprocessing method based on FIG. 2.
Fig. 4 is a schematic diagram of an apparatus of the present invention in an example.
Detailed Description
In order to make the objects, technical solutions and advantages of the present invention more apparent, the present invention will be described in detail with reference to the accompanying drawings and specific embodiments.
It should be noted that the terms "upper", "lower", "left", "right", "front", "back", etc. used in the present invention are for clarity of description only, and are not intended to limit the scope of the present invention, and the relative relationship between the terms and the terms is not limited by the technical contents of the essential changes.
Fig. 1 is a schematic flow chart of a traffic simulation method of a convolution long-and-short-term memory neural network according to an embodiment of the present invention. The embodiment can be used for the case of implementing traffic simulation of the convolutional long-term memory neural network through a device such as a server, and the method can be executed by a traffic simulation system of the convolutional long-term memory neural network, and the system can be implemented in a software and/or hardware manner and can be integrated in an electronic device, for example, an integrated server device.
Referring to fig. 1, the traffic simulation method includes:
s1, obtaining a traffic flow video in the simulation area based on the shooting of the unmanned aerial vehicle, and obtaining track data of all vehicles in the simulation area from the traffic flow video through an image recognition method;
in step S1, the vehicle trajectory data refers to a current timestamp; the number of the vehicle entering the sampling road section; the number of the current lane of the vehicle; the current longitudinal position of the vehicle and the length of the vehicle.
Taking fig. 2 as an example, fig. 2 includes a road section with a total length of 160 meters of one-way three lanes, 5 vehicles currently exist on the road section, and corresponding vehicle trajectory data is obtained by an image recognition method, where the time is as shown in table 1:
TABLE 1 summary of vehicle trajectory data at a given moment
And S2, preprocessing the vehicle track data to obtain a corresponding traffic state grid map.
In step S2, the vehicle trajectory data preprocessing method is to divide the vehicle trajectory data into a plurality of data sets based on the current timestamp information, divide the simulation area into grids of 2 m × 1 lanes, and divide the vehicle trajectory data in each data set into corresponding grids based on the number of lanes where the vehicle is currently located, the current longitudinal position of the vehicle relative to the starting point of the traffic flow video, and the length of the vehicle, so as to obtain the traffic state grid map. .
Fig. 3 is an example, and a traffic state grid diagram obtained by preprocessing vehicle trajectory data at a certain time included in table 1 by a vehicle trajectory data preprocessing method is shown in fig. 3.
And S3, taking the traffic state grid diagram as the input of the convolution long-time and short-time memory neural network, and performing parameter training on the convolution long-time and short-time memory neural network to obtain the trained convolution long-time and short-time memory neural network.
In step S3, the process of performing parameter training on the convolutional long-and-short term memory neural network includes:
dividing 25 continuous-time traffic state grid graphs into one round based on time continuity, wherein 20 are learning samples, and 5 are result calibration true values;
inputting 20 continuous traffic state grid graphs as a convolution long-time and short-time memory neural network, and predicting 5 traffic state grid graphs after the convolution long-time and short-time memory neural network is used for predicting;
and calculating a loss function of the 5 predicted traffic state grid graphs and the 5 calibrated real values serving as results, and performing iterative optimization on parameters through a back propagation algorithm.
And finishing the parameter training until the parameter training is finished after the set times of iterative training or the calculated loss value reaches the set convergence condition.
In step S3, the loss function expression designed in the convolutional long-term memory neural network is as follows:
C1＝(K1L)2
C2＝(K2L)2
wherein Los (x, y) represents a loss function; n represents all pixel points in the traffic state grid diagram; x is the number ofiRepresenting the pixel value of the ith point in the traffic state grid diagram; w is aiRepresenting the weight of the ith point pixel value in the traffic state grid diagram; mu.sxRepresenting the average value of all pixels in the result calibration real value map; mu.syRepresenting the average value of all the pixels in the traffic grid map obtained by prediction; sigmaxRepresenting the variance of the pixel values in the resulting calibration true value map; sigmayRepresenting the variance of pixel values in the traffic grid map obtained by prediction; sigmaxyRepresenting the covariance of the pixel values of the resulting calibration true value map and the predicted traffic grid map; k1Weights representing the resulting calibration true value map; k2Weights representing the traffic grid map obtained by prediction; l is the gray value dynamic range; c1Representation avoidance
A correction term of 0; c2Representation avoidance
A correction term of 0; α represents a first weight; β represents a second weight; p represents the result prediction probability value output by the convolution long-time and short-time memory neural network; softmax (p)iAnd representing a softmax activation function for converting the prediction probability value of the pixel value of the ith pixel point in the traffic state grid graph into a prediction result.
And S4, acquiring a traffic state grid map by adopting the steps of S1-S2 based on the simulation area, and inputting the sampled traffic state grid map into a neural network to acquire a simulation result.
The pair of the traffic simulation method based on the convolution long-time memory neural network and the traffic simulation method based on the traditional mathematical analysis model on the fitting degree of the microscopic simulation track is shown in the following table 2:
TABLE 2 comparison of simulation effect indexes of traffic simulation method based on convolution time memory neural network and traffic simulation method of traditional mathematical analysis model
The traffic simulation method based on the convolutional long-term memory neural network is designed, traffic states are innovatively converted into a grid diagram form, perception of vehicles around the vehicles in a traditional model is used as a convolution kernel, and therefore the traffic state evolution is effectively predicted by applying the advantages of the existing developed deep learning convolutional neural network technology; and secondly, a forgetting and memorizing structure in the network structure is utilized to realize long-time memory structure and avoid the problems of gradient explosion and gradient disappearance in the traditional neural network, thereby greatly improving the prediction accuracy of the neural network. Under the background of big data, the traffic simulation method based on the convolution long-time memory neural network under the support of mass traffic data can obviously further explore the advantages of deep learning in a big data environment, and has obvious effect on improving the precision of microscopic traffic simulation.
The embodiment of the invention provides a traffic simulation system with a convolution long-time memory neural network.
The vehicle track module is used for acquiring vehicle track data, and the vehicle track data comprises a current timestamp; the number of the vehicle entering the sampling road section; the number of the current lane of the vehicle; the current longitudinal position of the vehicle and the length of the vehicle;
the data preprocessing module is used for dividing the vehicle track data into a plurality of data groups based on the current timestamp information; the sampling area is divided into a grid of 2 meters of X1 lanes; dividing the data in each data group into corresponding grids based on the number of the lane where the vehicle is located at present and the longitudinal position where the vehicle is located at present;
the training module is used for training a convolution long-time memory neural network and carrying out iterative optimization on parameters and comprises a feedforward prediction submodule; a reverse optimization submodule;
and the simulation module is used for memorizing the neural network by utilizing the convolution duration obtained by the training module and carrying out traffic simulation prediction based on the input sampled traffic state grid diagram.
According to the traffic simulation system based on the convolution long-and-short-term memory neural network, the transmission object is determined by establishing the data containing relation of the whole application, and the aim of the traffic simulation method of the convolution long-and-short-term memory neural network is achieved. The microscopic simulation system provided by the embodiment of the invention can execute the traffic simulation method for memorizing the neural network by convolution time provided by any embodiment of the invention, and has corresponding functional modules and beneficial effects of the execution method. Fig. 4 shows a schematic diagram of a traffic simulation system based on a convolutional long-and-short-term memory neural network.
The embodiment of the application provides an electronic device, which comprises a processor, a memory, an input device and an output device; in the electronic device, the number of the processors can be one or more; the processor, memory, input devices, and output devices in the electronic device may be connected by a bus or other means.
The memory, which is a computer-readable storage medium, may be used for storing software programs, computer-executable programs, and modules, such as program instructions/modules corresponding to the detection method in the embodiments of the present invention. The processor executes various functional applications and data processing of the electronic device by operating the software program, the instructions and the modules stored in the memory, namely, the traffic simulation method of the convolution duration memory neural network provided by the embodiment of the invention is realized.
The memory can mainly comprise a program storage area and a data storage area, wherein the program storage area can store an operating system and an application program required by at least one function; the storage data area may store data created according to the use of the terminal, and the like. Further, the memory may include high speed random access memory, and may also include non-volatile memory, such as at least one magnetic disk storage device, flash memory device, or other non-volatile solid state storage device. In some examples, the memory may further include memory located remotely from the processor, and these remote memories may be connected to the electronic device through a network. Examples of such networks include, but are not limited to, the internet, intranets, local area networks, mobile communication networks, and combinations thereof.
The input device may be used to receive input numeric or character information and generate key signal inputs related to user settings and function control of the electronic device, and may include a keyboard, a mouse, and the like. The output device may include a display device such as a display screen.
Embodiments of the present application provide a computer-readable storage medium, on which a computer program is stored, and when the program is executed by a processor, the method for traffic simulation of a convolutional long-and-short-term memory neural network is implemented.
Of course, the storage medium provided by the embodiment of the present invention includes computer-executable instructions, and the computer-executable instructions are not limited to the operations of the method described above, and may also perform related operations in the traffic simulation method of the convolutional long-term memory neural network provided by any embodiment of the present invention.
The above description is only for the specific embodiment of the present invention, but the scope of the present invention is not limited thereto, and any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope of the present invention are included in the scope of the present invention.
Claims (9)
1. A traffic simulation method based on a convolution long-time memory neural network is characterized by comprising the following steps:
s1, obtaining a traffic flow video in the simulation area, and obtaining track data of all vehicles in the simulation area from the traffic flow video through an image recognition method;
step S2, preprocessing the vehicle track data to obtain a traffic state grid map of the simulation area;
step S3, taking the traffic state grid diagram as the input of the convolution long-time and short-time memory neural network, and carrying out parameter training on the convolution long-time and short-time memory neural network to obtain a trained convolution long-time and short-time memory neural network;
and S4, acquiring a traffic state grid map by adopting the methods of the steps S1-S2 based on the simulation area, and inputting the traffic state grid map into the trained convolutional long-time memory neural network to acquire a traffic simulation result.
2. The traffic simulation method based on the convolutional neural network for long-and-short term memory according to claim 1, wherein in step S1, the traffic flow video is photographed by an unmanned aerial vehicle.
3. The traffic simulation method based on the convolutional neural network for memorizing long and short term as claimed in claim 1, wherein in step S1, the vehicle trajectory data refers to a current timestamp, a number of a vehicle entering a sampling road segment, a number of a lane where the vehicle is currently located, a longitudinal position where the vehicle is currently located, and a length of the vehicle.
4. The traffic simulation method based on the convolution duration memory neural network as claimed in claim 1, wherein in step S2, the vehicle trajectory data preprocessing method is to divide the vehicle trajectory data into a plurality of data sets based on current timestamp information, divide the simulation area into grids of 2 m × 1 lanes, and divide the vehicle trajectory data in each data set into corresponding grids based on the number of the lane where the vehicle is currently located, the current longitudinal position of the vehicle relative to the starting point of the traffic flow video, and the length of the vehicle, so as to obtain the traffic state grid map.
5. The traffic simulation method based on the convolutional long-short term memory neural network as claimed in claim 1, wherein in step S3, the process of performing parameter training on the convolutional long-short term memory neural network includes:
dividing each M continuous-time traffic state grid graphs into one round based on time continuity, wherein M grids are selected from the M grids as learning samples, n grids are selected as result calibration real value graphs, and M is n + M;
presetting parameters of a convolutional long-time memory neural network;
inputting m traffic state grid graphs with continuous time as a convolution long-and-short time memory neural network, and performing feed-forward prediction through the convolution long-and-short time memory neural network to obtain n future traffic state grid graphs;
and calculating a loss function of the n predicted traffic state grid graphs and the result calibration real value graph, and performing iterative optimization on the parameters through a back propagation algorithm until the parameters are trained repeatedly for a set number of times or the calculated loss value reaches a set convergence condition, namely completing parameter training.
6. The traffic simulation method based on the convolution duration memory neural network as claimed in claim 5, wherein M is 25, M is 20, and n is 5.
7. The traffic simulation method based on the convolution long-and-short-term memory neural network as claimed in claim 5, wherein the loss function expression designed in the convolution long-and-short-term memory neural network is as follows:
C1＝(K1L)2
C2＝(K2L)2
in the formula, Loss (x, y) represents a Loss function; n represents all pixel points in the traffic state grid diagram; x is the number ofiRepresenting the pixel value of the ith point in the traffic state grid diagram; w is aiRepresenting the weight of the ith point pixel value in the traffic state grid diagram; mu.sxRepresenting the average value of all pixels in the result calibration real value map; mu.syRepresenting the average value of all the pixels in the traffic grid map obtained by prediction; sigmaxRepresenting the variance of the pixel values in the resulting calibration true value map; sigmayRepresenting the variance of pixel values in the traffic grid map obtained by prediction; sigmaxyRepresenting the covariance of the pixel values of the resulting calibration true value map and the predicted traffic grid map; k1Weights representing the resulting calibration true value map; k2Weights representing the traffic grid map obtained by prediction; l is the gray value dynamic range; c1Representation avoidance
A correction term of 0; c2Representation avoidance
A correction term of 0; α represents a first weight; β represents a second weight; p represents the result prediction probability value output by the convolution long-time and short-time memory neural network; softmax (p)iAnd representing a softmax activation function for converting the prediction probability value of the pixel value of the ith pixel point in the traffic state grid graph into a prediction result.
8. The traffic simulation system based on the convolutional neural network for long and short term memory according to claim 7, wherein K is1Taking 0.01, K20.03 is taken.
9. The traffic simulation system based on the convolution long-time and short-time memory neural network as claimed in claim 1, comprising:
the vehicle track module is used for acquiring vehicle track data;
and the data preprocessing module is used for dividing the vehicle track data into a plurality of data groups based on the current timestamp information, dividing the simulation area into grids of 2 m by 1 lanes, and dividing the vehicle track data in each data group into corresponding grids based on the serial number of the lane where the vehicle is currently located, the current longitudinal position of the vehicle relative to the starting point of the traffic flow video and the length of the vehicle, so as to obtain the traffic state grid map.
The training module is used for training the convolution duration memory neural network and performing iterative optimization on parameters; the training module comprises a feedforward prediction sub-module and a reverse optimization sub-module;
and the simulation module is used for memorizing the neural network by utilizing the convolution duration obtained by the training module and carrying out traffic simulation prediction based on the input sampled traffic state grid diagram.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111264070.4A CN114049764B (en) | 2021-10-27 | 2021-10-27 | Traffic simulation method and system based on convolution long-time and short-time memory neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111264070.4A CN114049764B (en) | 2021-10-27 | 2021-10-27 | Traffic simulation method and system based on convolution long-time and short-time memory neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN114049764A true CN114049764A (en) | 2022-02-15 |
| CN114049764B CN114049764B (en) | 2022-09-23 |
Family
ID=80206320
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111264070.4A Active CN114049764B (en) | 2021-10-27 | 2021-10-27 | Traffic simulation method and system based on convolution long-time and short-time memory neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114049764B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117935561A (en) * | 2024-03-20 | 2024-04-26 | 山东万博科技股份有限公司 | Intelligent traffic flow analysis method based on Beidou data |
Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3410404A1 (en) * | 2017-05-29 | 2018-12-05 | Cognata Ltd. | Method and system for creating and simulating a realistic 3d virtual world |
| CN109871876A (en) * | 2019-01-22 | 2019-06-11 | 东南大学 | A kind of Freeway Conditions identification and prediction technique based on floating car data |
| US20200184027A1 (en) * | 2018-12-07 | 2020-06-11 | Zoox, Inc. | System and method for modeling physical objects in a simulation |
| CN112100163A (en) * | 2020-08-19 | 2020-12-18 | 北京航空航天大学 | Road network state space-time prediction method based on three-dimensional convolutional neural network |
| CN112989568A (en) * | 2021-02-06 | 2021-06-18 | 武汉光庭信息技术股份有限公司 | Simulation scene three-dimensional road automatic construction method and device |
| CN113094875A (en) * | 2021-03-16 | 2021-07-09 | 东南大学 | Method and device for calibrating micro traffic simulation system in urban expressway intersection area |
| CN113268855A (en) * | 2021-04-19 | 2021-08-17 | 东南大学 | Calibration method of microscopic traffic simulation model of ring intersection |
| CN113312760A (en) * | 2021-05-14 | 2021-08-27 | 东南大学 | Traffic simulation-based networked motor vehicle right turn trajectory planning method and device |
-
2021
- 2021-10-27 CN CN202111264070.4A patent/CN114049764B/en active Active
Patent Citations (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3410404A1 (en) * | 2017-05-29 | 2018-12-05 | Cognata Ltd. | Method and system for creating and simulating a realistic 3d virtual world |
| US20200184027A1 (en) * | 2018-12-07 | 2020-06-11 | Zoox, Inc. | System and method for modeling physical objects in a simulation |
| CN109871876A (en) * | 2019-01-22 | 2019-06-11 | 东南大学 | A kind of Freeway Conditions identification and prediction technique based on floating car data |
| CN112100163A (en) * | 2020-08-19 | 2020-12-18 | 北京航空航天大学 | Road network state space-time prediction method based on three-dimensional convolutional neural network |
| CN112989568A (en) * | 2021-02-06 | 2021-06-18 | 武汉光庭信息技术股份有限公司 | Simulation scene three-dimensional road automatic construction method and device |
| CN113094875A (en) * | 2021-03-16 | 2021-07-09 | 东南大学 | Method and device for calibrating micro traffic simulation system in urban expressway intersection area |
| CN113268855A (en) * | 2021-04-19 | 2021-08-17 | 东南大学 | Calibration method of microscopic traffic simulation model of ring intersection |
| CN113312760A (en) * | 2021-05-14 | 2021-08-27 | 东南大学 | Traffic simulation-based networked motor vehicle right turn trajectory planning method and device |
Non-Patent Citations (2)
| Title |
|---|
| 孙腾达等: "一种基于路网网格化的微观交通仿真模型", 《***仿真学报》 * |
| 曹渝: "基于小波神经网络的交通状态短时预测", 《移动通信》 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117935561A (en) * | 2024-03-20 | 2024-04-26 | 山东万博科技股份有限公司 | Intelligent traffic flow analysis method based on Beidou data |
| CN117935561B (en) * | 2024-03-20 | 2024-05-31 | 山东万博科技股份有限公司 | Intelligent traffic flow analysis method based on Beidou data |
Also Published As
| Publication number | Publication date |
|---|---|
| CN114049764B (en) | 2022-09-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108921200B (en) | Method, apparatus, device and medium for classifying driving scene data | |
| CN112052787B (en) | Target detection method and device based on artificial intelligence and electronic equipment | |
| CN110929577A (en) | Improved target identification method based on YOLOv3 lightweight framework | |
| WO2021238262A1 (en) | Vehicle recognition method and apparatus, device, and storage medium | |
| CN109816032B (en) | Unbiased mapping zero sample classification method and device based on generative countermeasure network | |
| CN111741330B (en) | Video content evaluation method and device, storage medium and computer equipment | |
| CN110766038B (en) | Unsupervised landform classification model training and landform image construction method | |
| CN115616937B (en) | Automatic driving simulation test method, device, equipment and computer readable medium | |
| CN113505854A (en) | Method, device, equipment and medium for constructing facial image quality evaluation model | |
| CN114049764B (en) | Traffic simulation method and system based on convolution long-time and short-time memory neural network | |
| CN115187772A (en) | Training method, device and equipment of target detection network and target detection method, device and equipment | |
| CN113895460A (en) | Pedestrian trajectory prediction method, device and storage medium | |
| CN115561834A (en) | Meteorological short-term and temporary forecasting all-in-one machine based on artificial intelligence | |
| CN115018039A (en) | Neural network distillation method, target detection method and device | |
| CN116486624A (en) | Traffic flow prediction method and system based on space-time diagram convolutional neural network | |
| CN117152513A (en) | Vehicle boundary positioning method for night scene | |
| CN113409393B (en) | Method and device for identifying traffic sign | |
| CN117217280A (en) | Neural network model optimization method and device and computing equipment | |
| CN114882307A (en) | Classification model training and image feature extraction method and device | |
| CN117787585A (en) | Method and device for predicting number of charging piles driven by data | |
| CN116166642A (en) | Spatio-temporal data filling method, system, equipment and medium based on guide information | |
| CN116091964A (en) | High-order video scene analysis method and system | |
| CN114187759B (en) | Road side unit driving assistance method and device based on data driving model | |
| CN115576990A (en) | Method, device, equipment and medium for evaluating visual truth value data and perception data | |
| CN114048886A (en) | Effective berthage combination prediction method based on neural network, terminal and storage medium |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |