CN114399107A - Prediction method and system of traffic state perception information - Google Patents
Prediction method and system of traffic state perception information Download PDFInfo
- Publication number
- CN114399107A CN114399107A CN202210036552.2A CN202210036552A CN114399107A CN 114399107 A CN114399107 A CN 114399107A CN 202210036552 A CN202210036552 A CN 202210036552A CN 114399107 A CN114399107 A CN 114399107A
- Authority
- CN
- China
- Prior art keywords
- vehicle
- track data
- model
- sampling
- traffic state
- 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.)
- Pending
Links
- 230000008447 perception Effects 0.000 title claims abstract description 100
- 238000000034 method Methods 0.000 title claims abstract description 56
- 238000005070 sampling Methods 0.000 claims abstract description 102
- 238000012549 training Methods 0.000 claims abstract description 98
- 230000006870 function Effects 0.000 claims description 27
- 230000015654 memory Effects 0.000 claims description 25
- 238000012545 processing Methods 0.000 claims description 5
- 238000013528 artificial neural network Methods 0.000 claims description 4
- 238000004590 computer program Methods 0.000 claims description 4
- 238000011478 gradient descent method Methods 0.000 claims description 4
- 230000006403 short-term memory Effects 0.000 claims description 2
- 230000007774 longterm Effects 0.000 claims 1
- 238000005457 optimization Methods 0.000 abstract description 4
- 230000008569 process Effects 0.000 description 21
- 238000010586 diagram Methods 0.000 description 10
- 238000004422 calculation algorithm Methods 0.000 description 7
- 238000003062 neural network model Methods 0.000 description 7
- 230000006399 behavior Effects 0.000 description 4
- 230000002123 temporal effect Effects 0.000 description 4
- 238000011144 upstream manufacturing Methods 0.000 description 4
- 239000013598 vector Substances 0.000 description 4
- 238000013473 artificial intelligence Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 238000007726 management method Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 230000000670 limiting effect Effects 0.000 description 2
- 210000002569 neuron Anatomy 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000036961 partial effect Effects 0.000 description 2
- 230000002829 reductive effect Effects 0.000 description 2
- 239000000126 substance Substances 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000001133 acceleration Effects 0.000 description 1
- 230000002238 attenuated effect Effects 0.000 description 1
- 230000001413 cellular effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006757 chemical reactions by type Methods 0.000 description 1
- 238000001514 detection method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000007599 discharging Methods 0.000 description 1
- 230000006698 induction Effects 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 230000003304 psychophysiological effect Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 230000002441 reversible effect Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 230000000638 stimulation Effects 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/40—Business processes related to the transportation industry
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Health & Medical Sciences (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Software Systems (AREA)
- Computational Linguistics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Mathematical Physics (AREA)
- Data Mining & Analysis (AREA)
- Marketing (AREA)
- Biophysics (AREA)
- General Business, Economics & Management (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Tourism & Hospitality (AREA)
- Game Theory and Decision Science (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Development Economics (AREA)
- Primary Health Care (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention discloses a method and a system for predicting traffic state perception information, wherein a traffic state perception model is obtained by training and is composed of vehicle following models of a plurality of vehicles, track data output by the vehicle following model of a front vehicle is input into the vehicle following model of a rear vehicle, the vehicle following models of all vehicles in the model are trained simultaneously during training, and the track data input by the vehicle following model of each vehicle is determined by sampling; and inputting the acquired front vehicle track data and the acquired rear vehicle track data into a vehicle following model corresponding to the rear vehicle in the traffic state perception model, and outputting to obtain traffic state perception information. Therefore, the embodiment of the invention realizes the optimization of the traffic state perception model and reduces the prediction error of the traffic state perception information by changing the structure of the vehicle following model and determining the input parameters of the model by adopting plan sampling during training.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a method and a system for predicting traffic state perception information.
Background
With the rapid development of artificial intelligence technology and the internet of things, intelligent traffic systems have appeared. The intelligent traffic system can adopt a neural network model obtained by training to predict traffic state perception information such as traffic flow data or vehicle running track data, and the prediction result is applied to an unmanned vehicle to make a control driving decision.
And under different traffic scenes, predicting the traffic state perception information by adopting different neural network models.
Under the scene of sparse time dimension, a neural network model based on the time dimension is obtained through training to predict traffic flow data. The sparse track data in the time dimension means that: the case where the trajectory data of a continuous fleet is sparsely sampled in the time dimension. Such a scenario often occurs in ramp control and other problems, such as: after the running track data of the vehicles are acquired by a small number of induction coils arranged in the upstream ramp, a ramp signal lamp processor arranged on the downstream ramp is informed, so that the ramp signal lamp processor can obtain the traffic state of the upstream ramp in time, and the number of the vehicles running on the ramp is adjusted by controlling the ramp signal lamps. Under the scene, the trained neural network model fills up the missing vehicle track observation points in a track filling mode.
Under the scene of sparse spatial latitude, a vehicle following model is obtained based on the neural network training of the spatial latitude, and the vehicle driving track data is identified through the vehicle following model. The sparse spatial latitude of the trajectory data means that: a situation where only the trajectory data of a very small number of vehicles can be completely observed within the preset space. Such a scenario often occurs in the problem of cooperative automatic driving of the vehicle and the road, such as: through a small amount of vehicle running track data acquired by the networked automobiles, the unmanned vehicle can obtain downstream traffic state data in time for controlling a driving decision, and under the scene, the trained vehicle following model generates a complete running track of an upstream vehicle in an iterative manner on the basis of the observed vehicle running track data in a track generation mode.
The prediction technology of the fine-grained traffic state perception information driven by the sparse track data enables fine and efficient traffic control to be possible, and has important significance for further relieving urban traffic jam and improving urban traffic operation efficiency. However, when the internet-connected vehicles are not popularized, the problem of sparse track data makes it difficult to obtain accurate fine-grained traffic state data, which increases prediction errors of fine-grained traffic perception information. Therefore, how to reduce the prediction error, especially how to reduce the prediction error when predicting the traffic state perception information according to the set neural network model based on the sparse trajectory data, and predicting the traffic state perception information according to the set neural network model based on the sparse trajectory data at the spatial latitude, is an urgent problem to be solved.
Disclosure of Invention
In view of this, embodiments of the present invention provide a method for predicting traffic state awareness information, which can reduce a prediction error of the traffic state awareness information.
The embodiment of the invention also provides a system for predicting the traffic state perception information, which can reduce the prediction error of the traffic state perception information.
The embodiment of the invention is realized as follows:
a method of predicting traffic awareness status information, the method comprising:
forming a traffic state perception model by vehicle-following models of a plurality of vehicles, wherein track data output by the vehicle-following model of the front vehicle is input into the vehicle-following model of the rear vehicle;
simultaneously training the vehicle following models of all vehicles in the traffic state perception model, wherein each vehicle is used as a rear vehicle, and front vehicle track data output at the current moment based on the vehicle following model of the front vehicle and the front vehicle track data collected at the current moment are sampled to obtain first sampling track data; sampling rear track data obtained by calculating a vehicle following model of a rear vehicle at the previous moment and rear track data acquired at the current moment to obtain second sampling track data; inputting the first sampling track data and the second sampling track data into a car following model of the rear car for training at the current moment, and adjusting a weight value in the car following model of the rear car according to the rear car track data output at the current moment until the set training period is finished;
when the traffic state perception model is applied, the acquired front vehicle track data and the acquired rear vehicle track data are input into a vehicle following model of a corresponding rear vehicle in the traffic state perception model, and the traffic state perception information of the rear vehicle is output and acquired.
Optionally, the vehicle-following model of the vehicle employs a long-short-term memory artificial neural network structure.
Optionally, the sampling rear track data obtained by calculating a vehicle-following model of the rear vehicle at a previous time and rear track data acquired at a current time to obtain second sampling track data includes:
the vehicle following model of the rear vehicle calculates and obtains rear vehicle track data at the previous moment according to the front vehicle track data acquired at the previous moment and the acquired rear vehicle track data;
and acquiring second sampling track data of the rear vehicle at the current moment according to the rear vehicle track data at the previous moment and the rear vehicle track data acquired at the current moment.
Optionally, the obtaining, according to the rear vehicle trajectory data at the previous time and the rear vehicle trajectory data acquired at the current time, second sampling trajectory data of the rear vehicle at the current time includes:
setting a first adjustment probability value aiming at the rear vehicle track data acquired at the current moment;
extracting the rear vehicle track data at the previous moment according to the ratio of the unoccupied first adjustment probability value; extracting rear vehicle track data acquired at the current moment according to the ratio of the first adjustment probability value, and combining the extracted vehicle track data to obtain a value serving as second sampling track data of the rear vehicle at the current moment;
the first adjustment probability value is obtained by adopting a set attenuation function to calculate, and gradually attenuates to be a set threshold value based on time in a training period of a car-following model of the rear car.
Optionally, the obtaining of the first sampling trajectory data includes:
calculating to obtain the front vehicle track data at the current moment by adopting a vehicle following model of the front vehicle;
acquiring track data of a front vehicle at the current moment;
setting a second adjustment probability value aiming at the previous vehicle track data at the current moment, and extracting and calculating the previous vehicle track data at the current moment according to the unoccupied rate of the second adjustment probability value; and extracting and acquiring the front vehicle track data of the current moment according to the ratio occupied by the second adjustment probability value, and combining the extracted front vehicle track data to obtain a value serving as the obtained first sampling track data.
Optionally, the inputting the first sampling trajectory data and the second sampling trajectory data into a car-following model of the rear car for training at the current time, and adjusting a weight value in the car-following model of the rear car according to the rear car trajectory data output at the current time until the set training period is over includes:
adjusting the weight value of the car-following model of the rear car according to the loss function calculated by adopting the car-following model of the rear car during the training at the previous moment by adopting a gradient descent method, and performing the training at the current moment until the set training period is finished;
and the calculated loss function value is a value obtained by calculating the mean square error of the rear vehicle track data output by the vehicle following model of the rear vehicle at the previous moment.
A system for predicting traffic awareness status information, comprising: a setting unit of the traffic state perception model, a training unit of the traffic state perception model and an application unit of the traffic state perception model, wherein,
the device comprises a setting unit of a traffic state perception model, a processing unit and a control unit, wherein the setting unit is used for forming a traffic state perception model by vehicle-following models of a plurality of single vehicles, and track data output by the vehicle-following model of a front vehicle is input into the vehicle-following model of a rear vehicle;
the training unit of the traffic state perception model is used for simultaneously training the following models of the single vehicles in the traffic state perception model, wherein each vehicle is used as a rear vehicle, and front vehicle track data output at the current moment based on the vehicle following model of the front vehicle and the front vehicle track data collected at the current moment are sampled to obtain first sampling track data; sampling rear track data obtained by calculating a vehicle following model of a rear vehicle at the previous moment and rear track data acquired at the current moment to obtain second sampling track data; inputting the first sampling track data and the second sampling track data into a car following model of the rear car for training at the current moment, and adjusting a weight value in the car following model of the rear car according to the rear car track data output at the current moment until the set training period is finished;
and the application unit of the traffic state perception model is used for inputting the acquired front vehicle track data and the acquired rear vehicle track data into the vehicle following model of the corresponding rear vehicle in the traffic state perception model and outputting the acquired traffic state perception information of the rear vehicle.
A prediction apparatus of traffic perception state information, comprising: a memory; and a processor coupled to the memory, the processor configured to perform any of the above described methods of predicting traffic awareness status information based on instructions stored in the memory.
A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements any of the above-mentioned methods for predicting traffic awareness status information.
As can be seen from the above, the embodiment of the present invention trains and obtains a traffic state perception model, which is composed of vehicle-following models of a plurality of vehicles, wherein, the track data output by the vehicle-following model of the front vehicle is input into the vehicle-following model of the rear vehicle, the vehicle-following models of the vehicles in the model are trained simultaneously during training, and the track data input by the vehicle-following model of each vehicle is determined by sampling; and inputting the acquired front vehicle track data and the acquired rear vehicle track data into a vehicle following model corresponding to the rear vehicle in the traffic state perception model, and outputting to obtain traffic state perception information. Therefore, the embodiment of the invention realizes the optimization of the traffic state perception model and reduces the prediction error of the traffic state perception information by changing the structure of the vehicle following model and determining the input parameters of the model by adopting plan sampling during training.
Drawings
Fig. 1 is a flowchart of a method for predicting traffic status awareness information according to an embodiment of the present invention;
fig. 2 is a schematic structural diagram of a system for predicting traffic status awareness information according to an embodiment of the present invention;
fig. 3 is a schematic structural diagram of a specific implementation of a traffic status awareness model according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of a vehicle-following model adopting a planned sampling strategy according to an embodiment of the present invention.
Detailed Description
The technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the drawings in the embodiments of the present application, and it is obvious that the described embodiments are only a part of the embodiments of the present application, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present application.
The terms "first," "second," "third," "fourth," and the like in the description and in the claims, as well as in the drawings, if any, are used for distinguishing between similar elements and not necessarily for describing a particular sequential or chronological order. It is to be understood that the data so used is interchangeable under appropriate circumstances such that the embodiments of the invention described herein are, for example, capable of operation in sequences other than those illustrated or otherwise described herein. Furthermore, the terms "comprising" and "having," as well as any variations thereof, are intended to cover non-exclusive inclusions. For example, a process, method, system, article, or apparatus that comprises a list of steps or elements is not necessarily limited to those steps or elements explicitly listed, but may include other steps or elements not explicitly listed or inherent to such process, method, article, or apparatus.
The technical solution of the present invention will be described in detail with specific examples. Several of the following embodiments may be combined with each other and some details of the same or similar concepts or processes may not be repeated in some embodiments.
From the background, it can be seen that the sparseness of the trajectory data in the spatial latitude is one of typical scenes of the prediction research of the fine-grained traffic state perception information, and the trajectory data can be applied to the control driving decision when the vehicle is not driven by people. The inaccurate fine-grained traffic state perception result leads to an incorrect prediction result, and simultaneously, the vehicle-road cooperative automatic driving vehicle cannot respond to the change of the downstream traffic condition in advance, so that the vehicle-road cooperative automatic driving vehicle is degraded into a common automatic driving vehicle. In extreme conditions, abnormal driving behavior affecting the operation of the traffic flow, such as emergency braking due to misjudgment, may also occur due to a misprediction of future traffic conditions. In the background art, an accurate model capable of directly describing the mapping relationship between single vehicle track data and traffic state perception information is not provided. Therefore, on the basis of sparse track data on the spatial latitude, the traffic state perception information is predicted according to the set neural network model, and how to reduce prediction errors is also a technical problem to be solved.
In the background art, when an unmanned vehicle makes a decision to control a vehicle, a vehicle-following model, which is a behavior model describing how a driven vehicle travels with a preceding vehicle, is generally used to predict vehicle travel track data. The types into which the car-following model is mainly divided are: stimulus-response class, safe distance class, psycho-physiological class, cellular automaton class, and artificial intelligence class. The stimulation-reaction type car-following model assumes that the driving behavior of a driver is related to a current vehicle speed parameter, a front-rear vehicle distance parameter and a front-rear vehicle relative speed parameter, and simulates the driver to perform 'reflection' on the 'stimulation' of the parameters to obtain a prediction result of the traffic state perception information. The safe distance type car following model aims at avoiding collision, and simulates the condition that a driver expects to keep a certain safe distance from a front car to obtain a prediction result of traffic state perception information.
The problem of error propagation is ubiquitous in the problem of track data generation by adopting the existing vehicle-following model for a single vehicle in the background art, namely: an error exists between the obtained traffic perception state and the actual traffic perception state, and the error can be continuously accumulated and spread in time and space latitude. Particularly, in the case of traffic flow disturbance, that is, frequent acceleration and deceleration of vehicles due to congestion, the phenomenon of error propagation is more remarkable. Therefore, the problem of error propagation in the generation process of the traffic state perception information is a problem that fine-grained traffic state perception based on a spatial sparse track needs to be solved.
In order to solve the above problems, in the embodiment of the present invention, a traffic state perception model is obtained by training, where the model is composed of a plurality of vehicle-following models for single vehicles, track data output by a vehicle-following model of a front vehicle is input into a vehicle-following model of a rear vehicle, the vehicle-following models of vehicles in the model are trained simultaneously during training, and track data input by the vehicle-following model of each vehicle is determined by sampling; when the method is applied, the acquired front vehicle track data and the acquired rear vehicle track data are input into a vehicle following model of a corresponding rear vehicle in the traffic state perception model, and the traffic state perception information is output. Therefore, the embodiment of the invention realizes the optimization of the traffic state perception model and reduces the prediction error of the traffic state perception information by changing the structure of the vehicle following model and adopting the plan sampling to determine the input parameters during training.
That is to say, the embodiment of the present invention trains the traffic state awareness model by using the overall training strategy, and determines the input after the vehicle-following models of the vehicles in the traffic state awareness model perform the planned sampling of the input parameters by using the planned sampling strategy in the training process, thereby implementing the optimization of the traffic state awareness model obtained by training, and reducing the prediction errors of the traffic state awareness information in the time latitude and the space latitude.
Specifically, the method for using the planned sampling strategy in the training process by the vehicle-following model of each vehicle comprises the following steps: taking each vehicle as a rear vehicle, and sampling front vehicle track data output at the current moment based on a vehicle following model of the front vehicle and the front vehicle track data acquired at the current moment to obtain first sampling track data; sampling rear track data obtained by calculating a vehicle following model of a rear vehicle at the previous moment and rear track data acquired at the current moment to obtain second sampling track data; and inputting the first sampling track data and the second sampling track data into a car-following model of the rear car for training at the current moment.
In the embodiment of the invention, in order to solve the error propagation problem of the time latitude and the error propagation problem of the space latitude, the method is realized by a planned sampling strategy, namely, the method is realized by sampling and determining the input track data in the training process of the car-following model of each vehicle in the traffic state perception model; and the method is realized through an integral training strategy, namely the model consists of a plurality of vehicle-following models aiming at the single vehicle, and the track data output by the vehicle-following model of the front vehicle is input into the vehicle-following model of the rear vehicle and is trained simultaneously during training. The model provided by the embodiment of the invention is an improved model of a car-following model, and a planning sampling module is arranged at the input end of the car-following model of each vehicle, so that input track data of the model is determined by sampling and is input into the model for training.
Fig. 1 is a flowchart of a method for predicting traffic status awareness information according to an embodiment of the present invention, which includes the following specific steps:
In this embodiment, the vehicle-following model of the vehicle employs a long short-term memory artificial neural network (LSTM) model.
In this embodiment, the traffic state sensing information of the rear vehicle obtained by the output is actually the rear vehicle trajectory data of the rear vehicle, and may describe the traffic state around the rear vehicle at the current time.
In this embodiment, sampling the rear-vehicle track data obtained by calculating the vehicle-following model of the rear vehicle at the previous time and the rear-vehicle track data acquired at the current time, and obtaining the second sampling track data includes:
the vehicle following model of the rear vehicle calculates and obtains rear track data at the previous moment according to the front track data acquired at the previous moment and the acquired rear track data;
and acquiring second sampling track data of the rear vehicle at the current moment according to the rear vehicle track data at the previous moment and the rear vehicle track data acquired at the current moment.
That is, the trajectory data for training input in the vehicle-following model of the rear vehicle is composed of two parts, and the occupation ratio of the two parts is calculated according to the set first adjustment probability value. One of the two parts is the calculated rear vehicle track data at the current moment, and the other part is the rear vehicle track data acquired at the current moment, so that errors caused in the time dimension in the training process can be offset.
Particularly, the obtaining of the second sampling trajectory data of the rear vehicle at the current time according to the rear vehicle trajectory data at the previous time and the rear vehicle trajectory data acquired at the current time includes:
setting a first adjustment probability value aiming at the rear vehicle track data acquired at the current moment;
extracting rear vehicle track data at the previous moment according to the ratio of the unoccupied first adjustment probability value; and extracting the rear vehicle track data acquired at the current moment according to the ratio of the first adjustment probability value, and combining the extracted vehicle track data to obtain a value serving as second sampling track data of the rear vehicle at the current moment.
The first adjustment probability value is determined here using a set decay function, with a gradual decay as a function of the time of day in the training period of the vehicle-following model of the following vehicle up to a set threshold. The set threshold value can be 0, in this case, in the later stage of the training period, the vehicle following model of the rear vehicle is trained only according to the rear vehicle track data extracted at the previous moment, the rear vehicle track data does not need to be acquired at the current moment, and the trained vehicle following model of the rear vehicle is more accurate in subsequent prediction. The process of attenuating the first adjustment probability value to 0 can be implemented by setting the length of the training period.
That is, in the training period of the traffic state perception model, the first adjustment probability value is different at each training time, and is attenuated as the training time elapses. This is because, as the training progresses, the error of the traffic state perception model in the time latitude becomes smaller and smaller, and therefore, only the rear vehicle trajectory data of the current time calculated at the previous time may be trained as the trajectory data to be input subsequently.
Optionally, the obtaining of the first sampling trajectory data includes:
calculating to obtain the front vehicle track data at the current moment by adopting a vehicle following model of the front vehicle;
acquiring track data of a front vehicle at the current moment;
setting a second adjustment probability value aiming at the previous vehicle track data at the current moment, and extracting and calculating the previous vehicle track data at the current moment according to the unoccupied rate of the second adjustment probability value; and extracting and acquiring the front vehicle track data of the current moment according to the ratio occupied by the second adjustment probability value, and combining the extracted front vehicle track data to obtain a value serving as the obtained first sampling track data.
That is to say, when the vehicle-following model of each rear vehicle is trained, the input front-vehicle track data is composed of two parts, and the occupation proportion of the two parts is calculated according to the set second adjustment probability value. In this way, errors caused in the spatial latitude during the training of the vehicle-following model can be counteracted.
In this embodiment, inputting the first sampling trajectory data and the second sampling trajectory data into a car-following model of the rear car to train at the current time, and adjusting a weight value in the car-following model of the rear car according to the rear-car trajectory data output at the current time until the set training period is finished includes:
adjusting the weight value of the car following model of the rear car according to the calculated loss function value by adopting a gradient descent method according to the loss function of the car following model of the rear car in the training process at the previous moment, and performing the training at the current moment until the set training period is finished;
the calculated loss function value is a value obtained by calculating the mean square error of rear track data output by a vehicle following model of the rear vehicle at the previous moment.
In this way, it is ensured that the vehicle-following models of the vehicles in the traffic state perception model can be trained simultaneously.
In this embodiment, the predicted rear-vehicle trajectory data is output as the traffic state perception information, and a control decision can be made according to the rear-vehicle trajectory data when the vehicle is unmanned according to the predicted rear-vehicle trajectory data.
Fig. 2 is a schematic structural diagram of a system for predicting traffic state perception information according to an embodiment of the present invention, including: a setting unit of the traffic state perception model, a training unit of the traffic state perception model and an application unit of the traffic state perception model, wherein,
the device comprises a setting unit of a traffic state perception model, a processing unit and a control unit, wherein the setting unit is used for forming a traffic state perception model by vehicle-following models of a plurality of single vehicles, and track data output by the vehicle-following model of a front vehicle is input into the vehicle-following model of a rear vehicle;
the training unit of the traffic state perception model is used for simultaneously training the following models of the single vehicles in the traffic state perception model, wherein each vehicle is used as a rear vehicle, and front vehicle track data output at the current moment based on the vehicle following model of the front vehicle and the front vehicle track data collected at the current moment are sampled to obtain first sampling track data; sampling rear track data obtained by calculating a vehicle following model of a rear vehicle at the previous moment and rear track data acquired at the current moment to obtain second sampling track data; inputting the first sampling track data and the second sampling track data into a car following model of the rear car for training at the current moment, and adjusting a weight value in the car following model of the rear car according to the rear car track data output at the current moment until the set training period is finished;
and the application unit of the traffic state perception model is used for inputting the acquired front vehicle track data and the acquired rear vehicle track data into the vehicle following model of the corresponding rear vehicle in the traffic state perception model and outputting the acquired traffic state perception information of the rear vehicle.
In this system, the vehicle-following model of the vehicle is an LSTM model.
In the system, the training unit of the traffic state perception model further comprises a plan sampling module, and the plan sampling module is further used for sampling to obtain first sampling track data and sampling to obtain second sampling track data.
Wherein, the planning sampling module is further configured to sample to obtain second sampling trajectory data, including: the vehicle following model of the rear vehicle calculates and obtains rear track data at the previous moment according to the front track data acquired at the previous moment and the acquired rear track data; and acquiring second sampling track data of the rear vehicle at the current moment according to the rear vehicle track data at the previous moment and the rear vehicle track data acquired at the current moment.
The plan sampling module is further configured to obtain second sampling trajectory data of the rear vehicle at the current time according to the rear vehicle trajectory data at the previous time and the rear vehicle trajectory data acquired at the current time, where the second sampling trajectory data of the rear vehicle at the current time includes: setting a first adjustment probability value aiming at the rear vehicle track data acquired at the current moment; extracting rear vehicle track data at the previous moment according to the ratio of the unoccupied first adjustment probability value; and extracting the rear vehicle track data acquired at the current moment according to the ratio of the first adjustment probability value, and combining the extracted vehicle track data to obtain a value serving as second sampling track data of the rear vehicle at the current moment.
Here, the first adjustment probability value is calculated using a set attenuation function, and the time-based gradual attenuation is set as a set threshold value in a training period of a car-following model of a following car. The set threshold may be 0.
In the system, the planning sampling module is further configured to sample second sampling trajectory data, including: calculating to obtain the front vehicle track data at the current moment by adopting a vehicle following model of the front vehicle; acquiring track data of a front vehicle at the current moment; setting a second adjustment probability value aiming at the previous vehicle track data at the current moment, and extracting and calculating the previous vehicle track data at the current moment according to the unoccupied rate of the second adjustment probability value; and extracting and acquiring the front vehicle track data of the current moment according to the ratio occupied by the second adjustment probability value, and combining the extracted front vehicle track data to obtain a value serving as the obtained first sampling track data.
In the system, a training unit of the traffic state perception model is further used for adjusting the weight value of the car following model of the rear car according to the loss function of the car following model of the rear car in the training process at the previous moment by adopting a gradient descent method and the loss function value obtained by calculation, and training at the current moment is carried out until the set training period is finished; the calculated loss function value is a value obtained by calculating the mean square error of rear-vehicle track data output by a rear-vehicle following model at the previous moment.
The following provides a detailed description of examples of the present invention.
Fig. 3 is a schematic diagram of a specific implementation structure of the traffic state perception model provided in the embodiment of the present invention, and as shown in the figure, the model is obtained from a Vehicle-following model of two vehicles, including a Vehicle-following model of Vehicle 2(Vehicle2) and Vehicle 3(Vehicle 3). Fig. 3 shows three vehicles, namely Vehicle 1(Vehicle1), Vehicle2, and Vehicle3, whose trajectory data are collected in a spatial dimension (spatial dimension) according to a time sequence in the temporal dimension (temporal dimension). For each car-following model (the car-following model of Vehicle2 includes in the figure two boxes at the bottom right and a box labeled "SS", and the car-following model of Vehicle3 includes in the figure two boxes at the top right and a box labeled "SS"), a planning sampling module, denoted "SS", is provided for inputting the track data samples. Here, for the vehicle-following model of the vehicle2, the data sample input at time t1 of the temporal dimension is the vehicle trajectory data(s) of the vehicle1, which is the preceding vehicle1(1) And trajectory data(s) of the following vehicle, i.e., vehicle22(1) ); the data sample input at time t2 is the vehicle trajectory data(s) of the preceding vehicle, i.e., vehicle11(2) And vehicle trajectory data(s) at time t2 calculated at time t12(2)). For the vehicle-following model of vehicle3, at time t1The input data sample is the track data(s) of the rear vehicle (vehicle3)3(1) Track data of the preceding vehicle (vehicle2)
The trajectory data of the preceding vehicle is the vehicle trajectory data at time t 1(s) calculated at time t1 based on the vehicle following model of the vehicle22(1) With the collected vehicle data of the vehicle2 collected at the time t1, the sampled value is calculated
The data sample input at time t2 is vehicle trajectory data(s) of the preceding vehicle, i.e., vehicle22(2) And vehicle trajectory data at time t2 calculated at time t1, the vehicle trajectory data of the vehicle2 being based on the vehicle trajectory data at time t2 calculated at time t2 in the vehicle-following model of the vehicle2 and the collected vehicle data of the vehicle2 collected at time t 2.
As can be seen from fig. 3, the plan sampling module mainly solves the problem of prediction error propagation in the time dimension when predicting traffic state perception information, and vehicle trajectory data of the model is sampled and determined in the training process of the model and is input to the model for training.
Fig. 4 is a schematic structural diagram of a vehicle-following model adopting a planned sampling strategy according to an embodiment of the present invention. The dashed box in the figure is a Car-following Model implementation module, and is represented by "Car-following Model" in the figure. The vehicle following model implementation module adopts an LSTM model structure, the input of the vehicle following model implementation module is respectively from vehicle track data of a front vehicle and a rear vehicle, h (t-1) in the implementation module represents a hidden vector calculated by the LSTM model structure at the t-1 moment, and h (t-1) passes through a fully-connected Output Layer (represented as Output Layer in the figure) and calculates the vehicle track data of the next moment t through the calculation of a Loss function (represented as Loss in the figure)
At time t-1, the plan Sampling module (represented by Scheduled Sampling in the figure) selects a real rear truckVehicle trajectory data SF(t-1) (ActualS is used in the figure)F(t-1)) and vehicle trajectory data of the following vehicle generated by the model at the time of t-1
(use in the figure)
Representation) are randomly selected. H (1) in the implementation module represents a hidden vector calculated by an LSTM model structure at the 1 st moment, and the input of h (1) is real vehicle track data S of a rear vehicle at the 1 st momentF(t)。
At the kth moment of model training, the planning sampling module respectively takes the probability as EKAnd 1-eKAnd taking the real and generated vehicle track data of the rear vehicle as the input track data of the vehicle-following model adopted at the k-th time. With the continuous training of the model, the embodiment of the invention can adjust the probability EKTo achieve the purpose of dynamically modifying the input trajectory data of the model. Therefore, by the planning sampling strategy, the embodiment of the invention realizes the transition from the model training to the generation of the rear vehicle track data based on the model, and avoids the error propagation problem caused by the model error in the actual vehicle track data generation process to the maximum extent.
In the initial stage of training of the model, since the network parameters in the model are not sufficiently trained, if the vehicle trajectory data generated by the model is directly input, the convergence rate of the model in the training process is reduced. Conversely, after the model is fully trained, the real vehicle trajectory data can be replaced more with the vehicle trajectory data generated by the model, reducing error propagation in the time dimension. Thus, during the training process, the probability ∈ with the real vehicle trajectory data as inputKShould be gradually reduced as the training progresses.
In the embodiment of the invention, the probability value adopted by the model in the training process of the vehicle trajectory data of the rear vehicle acquired at the time t is expressed by the following formula.
Wherein epoch is the maximum training period of the model, and the training period k determines the probability e through a decay function decapay (k)KWhen to fall to 0. Wherein, table one lists three typical attenuation functions applied, including: linear, exponential, and anti-Sigmoid decay functions.
In the embodiment of the invention, an overall training strategy is also adopted. The integral training strategy mainly solves the problem of error propagation on the space latitude of the model. Since errors are continuously accumulated and propagated on a spatial latitude, training a vehicle-following model by using a vehicle as a unit cannot effectively inhibit error propagation. In consideration of the front-rear position relation of the vehicles, the embodiment of the invention combines the vehicle-following models of a plurality of vehicles into a complete model which is used as a traffic state perception model, and trains the assembled model uniformly. Therefore, structural differences between the training of the traffic state perception model and the subsequent process self-checking of generating traffic perception state information based on the model are avoided.
When the traffic state perception model is specifically trained, the following algorithm 1 is adopted to train the model.
Wherein the content of the first and second substances,line 6 in Algorithm 1 describes the input selection process for the planning sample module in the model. The algorithm first belongs to the probability EKGenerating random binary array [ 2 ]1…,bT]And used to determine the source of vehicle trajectory data samples for use in training. In updating the network parameters in the model, the penalty function in algorithm 1 uses the MSE of the vehicle trajectory data of all vehicles involved in the model as a penalty value, i.e. a representation of line 9 in algorithm 1, where I represents the number of vehicles of all vehicles involved in the model, thereby ensuring that the vehicle-following models of all vehicles can be trained simultaneously.
Here, the partial derivative calculations in the vehicle-following model are distinguished from the LSTM model, since the vehicle-following model at each vehicle introduces a planned sampling strategy. The partial derivative of the model can be obtained by using the following formula:
wherein W represents a weight matrix in the model,
representing the output of the model at time t; h (t) represents the output of the hidden layer in the model at time t; ctVehicle trajectory data representing neurons in the model at time t is represented in vector form.
When the vehicle-following model in the model adopts an LSTM model, the equation is updated according to the state of the neurons of the LSTM model, and the multiplication term in the formula
Can be expressed as:
as can be seen from Algorithm 1, x is input due to the modelk＝[xk,L,xk,F]Is no longer a constant xk,LAnd xk,FHere, the two constants represent the front and rear vehicle trajectory data respectively, and are represented by vectors, so that the model can be obtained in the time dimension:
wherein, bk,FE {0,1} represents the switch variables that the planned sampling module of the model sets in the time dimension. For example, when bk,FWhen 1, the input x of the model is expressedk,FOutput of the model from the previous moment when bk,FWhen 1, the input x of the model is expressedk,FFrom the rear track data acquired at time k.
The model can be obtained at the spatial latitude:
wherein the content of the first and second substances,
representing the value of a memory unit in a car-following model of a front car at the moment k-1; bk,LE {0,1} represents the switching variables set by the planning sampling module in the model that control the propagation of the gradient in the spatial dimension. Therefore, in the process of back propagation of the model training, the gradient of the model not only follows the hidden variable hk-1Backward propagation, and input x along the model with a certain probabilitykTraveling in the direction of the vehicle up the moment and upstream.
It can be seen that the difference between the LSTM-based vehicle-following model and the traffic state perception model provided by the embodiment of the present invention is mainly reflected in the connection structure and the training method of the model. The LSTM model in the background art ignores temporal and spatial latitude error propagation issues, training only the following behavior for a single vehicle. On the basis of the vehicle-following model of each vehicle, the embodiment of the invention enables the model to effectively inhibit error propagation problem in the training process through a planning sampling strategy and an integral training mode.
Embodiments of the present application also provide a computer-readable storage medium storing instructions that, when executed by a processor, may perform the performing steps in a method for predicting traffic state awareness information as described above. In practical applications, the computer readable medium may be included in the apparatus/device/system described in the above embodiments, or may exist alone without being assembled into the apparatus/device/system. The computer readable storage medium carries one or more programs, and when the one or more programs are executed, the method for predicting traffic state perception information may be implemented as described in the embodiments. According to embodiments disclosed herein, the computer-readable storage medium may be a non-volatile computer-readable storage medium, which may include, for example and without limitation: a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing, without limiting the scope of the present disclosure. In the embodiments disclosed herein, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device.
The embodiment of the invention also provides electronic equipment, wherein the device for realizing the method in the embodiment of the application can be integrated.
Specifically, the method comprises the following steps:
the electronic device may include a processor of one or more processing cores, memory of one or more computer-readable storage media, and a computer program stored on the memory and executable on the processor. When the program of the memory is executed, a method of predicting traffic state perception information as described above may be implemented.
Specifically, in practical applications, the electronic device may further include a power supply, an input unit, an output unit, and other components. Those skilled in the art will appreciate that the configuration of the electronic device in the embodiments of the present invention is not intended to be limiting, and may include more or fewer components than those shown, or some components in combination, or a different arrangement of components. Wherein:
the processor is a control center of the electronic device, connects various parts of the whole electronic device by various interfaces and lines, and performs various functions of the server and processes data by running or executing software programs and/or modules stored in the memory and calling data stored in the memory, thereby performing overall monitoring of the electronic device.
The memory may be used to store software programs and modules, i.e., the computer-readable storage media described above. The processor executes various functional applications and data processing by executing software programs and modules stored in the memory. The memory may mainly include a storage program area and a storage data area, wherein the storage program area may store an operating system, an application program required for at least one function, and the like; the storage data area may store data created according to the use of the server, 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 volatile solid state storage device. Accordingly, the memory may also include a memory controller to provide the processor access to the memory.
The electronic equipment also comprises a power supply for supplying power to each component, and the power supply can be logically connected with the processor through the power management system, so that the functions of charging, discharging, power consumption management and the like can be managed through the power management system. The power supply may also include any component of one or more dc or ac power sources, recharging systems, power failure detection circuitry, power converters or inverters, power status indicators, and the like.
The electronic device may further include an input unit operable to receive input numeric or character information and generate keyboard, mouse, joystick, optical or trackball signal inputs related to user settings and function control.
The electronic device may further include an output unit that may be used to display information input by or provided to a user as well as various graphical user interfaces that may be made up of graphics, text, icons, video, and any combination thereof.
The flowchart and block diagrams in the figures of the present application illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments disclosed herein. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams or flowchart illustration, and combinations of blocks in the block diagrams or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
Those skilled in the art will appreciate that various combinations and/or combinations of features recited in the various embodiments and/or claims of the present disclosure can be made, even if such combinations or combinations are not explicitly recited in the present application. In particular, the features recited in the various embodiments and/or claims of the present application may be combined and/or coupled in various ways, all of which fall within the scope of the present disclosure, without departing from the spirit and teachings of the present application.
The principles and embodiments of the present invention are explained herein using specific examples, which are provided only to help understanding the method and the core idea of the present invention, and are not intended to limit the present application. It will be appreciated by those skilled in the art that changes may be made in this embodiment and its broader aspects and without departing from the principles, spirit and scope of the invention, and that all such modifications, equivalents, improvements and equivalents as may be included within the scope of the invention are intended to be protected by the claims.
Claims (9)
1. A method for predicting traffic-aware state information, the method comprising:
forming a traffic state perception model by vehicle-following models of a plurality of vehicles, wherein track data output by the vehicle-following model of the front vehicle is input into the vehicle-following model of the rear vehicle;
simultaneously training the vehicle following models of all vehicles in the traffic state perception model, wherein each vehicle is used as a rear vehicle, and front vehicle track data output at the current moment based on the vehicle following model of the front vehicle and the front vehicle track data collected at the current moment are sampled to obtain first sampling track data; sampling rear track data obtained by calculating a vehicle following model of a rear vehicle at the previous moment and rear track data acquired at the current moment to obtain second sampling track data; inputting the first sampling track data and the second sampling track data into a car following model of the rear car for training at the current moment, and adjusting a weight value in the car following model of the rear car according to the rear car track data output at the current moment until the set training period is finished;
when the traffic state perception model is applied, the acquired front vehicle track data and the acquired rear vehicle track data are input into a vehicle following model of a corresponding rear vehicle in the traffic state perception model, and the traffic state perception information of the rear vehicle is output and acquired.
2. The method of claim 1, wherein the vehicle-following model of the vehicle employs a long-term short-term memory artificial neural network structure.
3. The method of claim 1, wherein sampling the rear-vehicle trajectory data calculated by the vehicle-following model of the rear vehicle at the previous time with the rear-vehicle trajectory data collected at the current time to obtain second sampled trajectory data comprises:
the vehicle following model of the rear vehicle calculates and obtains rear vehicle track data at the previous moment according to the front vehicle track data acquired at the previous moment and the acquired rear vehicle track data;
and acquiring second sampling track data of the rear vehicle at the current moment according to the rear vehicle track data at the previous moment and the rear vehicle track data acquired at the current moment.
4. The method according to claim 3, wherein the obtaining of the second sampling trajectory data of the rear vehicle at the current time according to the rear vehicle trajectory data at the previous time and the rear vehicle trajectory data acquired at the current time comprises:
setting a first adjustment probability value aiming at the rear vehicle track data acquired at the current moment;
extracting the rear vehicle track data at the previous moment according to the ratio of the unoccupied first adjustment probability value; extracting rear vehicle track data acquired at the current moment according to the ratio of the first adjustment probability value, and combining the extracted vehicle track data to obtain a value serving as second sampling track data of the rear vehicle at the current moment;
the first adjustment probability value is obtained by adopting a set attenuation function to calculate, and gradually attenuates to be a set threshold value based on time in a training period of a car-following model of the rear car.
5. The method according to claim 1, wherein the sampling of the previous-vehicle track data output at the current time based on the vehicle-following model of the previous vehicle and the previous-vehicle track data collected at the current time to obtain the first sampled track data comprises:
calculating to obtain the front vehicle track data at the current moment by adopting a vehicle following model of the front vehicle;
acquiring track data of a front vehicle at the current moment;
setting a second adjustment probability value aiming at the previous vehicle track data at the current moment, and extracting and calculating the previous vehicle track data at the current moment according to the unoccupied rate of the second adjustment probability value; and extracting and acquiring the front vehicle track data of the current moment according to the ratio occupied by the second adjustment probability value, and combining the extracted front vehicle track data to obtain a value serving as the obtained first sampling track data.
6. The method according to claim 1, wherein the inputting the first and second sampling trajectory data into the car-following model of the rear car for training at the current time, and the adjusting the weight value in the car-following model of the rear car according to the rear-car trajectory data output at the current time until the set training period is over comprises:
adjusting the weight value of the car-following model of the rear car according to the loss function calculated by adopting the car-following model of the rear car during the training at the previous moment by adopting a gradient descent method, and performing the training at the current moment until the set training period is finished;
and the calculated loss function value is a value obtained by calculating the mean square error of the rear vehicle track data output by the vehicle following model of the rear vehicle at the previous moment.
7. A system for predicting traffic awareness status information, comprising: a setting unit of the traffic state perception model, a training unit of the traffic state perception model and an application unit of the traffic state perception model, wherein,
the device comprises a setting unit of a traffic state perception model, a processing unit and a control unit, wherein the setting unit is used for forming a traffic state perception model by vehicle-following models of a plurality of single vehicles, and track data output by the vehicle-following model of a front vehicle is input into the vehicle-following model of a rear vehicle;
the training unit of the traffic state perception model is used for simultaneously training the following models of the single vehicles in the traffic state perception model, wherein each vehicle is used as a rear vehicle, and front vehicle track data output at the current moment based on the vehicle following model of the front vehicle and the front vehicle track data collected at the current moment are sampled to obtain first sampling track data; sampling rear track data obtained by calculating a vehicle following model of a rear vehicle at the previous moment and rear track data acquired at the current moment to obtain second sampling track data; inputting the first sampling track data and the second sampling track data into a car following model of the rear car for training at the current moment, and adjusting a weight value in the car following model of the rear car according to the rear car track data output at the current moment until the set training period is finished;
and the application unit of the traffic state perception model is used for inputting the acquired front vehicle track data and the acquired rear vehicle track data into the vehicle following model of the corresponding rear vehicle in the traffic state perception model and outputting the acquired traffic state perception information of the rear vehicle.
8. A prediction apparatus for traffic awareness status information, comprising:
a memory; and a processor coupled to the memory, the processor configured to perform the method of predicting traffic awareness status information according to any one of claims 1 to 6 based on instructions stored in the memory.
9. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the method of predicting traffic awareness status information according to any one of claims 1 to 6.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210036552.2A CN114399107A (en) | 2022-01-12 | 2022-01-12 | Prediction method and system of traffic state perception information |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210036552.2A CN114399107A (en) | 2022-01-12 | 2022-01-12 | Prediction method and system of traffic state perception information |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114399107A true CN114399107A (en) | 2022-04-26 |
Family
ID=81231215
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210036552.2A Pending CN114399107A (en) | 2022-01-12 | 2022-01-12 | Prediction method and system of traffic state perception information |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114399107A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116992221A (en) * | 2023-07-31 | 2023-11-03 | 武汉天翌数据科技发展有限公司 | Fault detection method, device and equipment of operation and maintenance platform and storage medium |
-
2022
- 2022-01-12 CN CN202210036552.2A patent/CN114399107A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116992221A (en) * | 2023-07-31 | 2023-11-03 | 武汉天翌数据科技发展有限公司 | Fault detection method, device and equipment of operation and maintenance platform and storage medium |
| CN116992221B (en) * | 2023-07-31 | 2024-03-26 | 武汉天翌数据科技发展有限公司 | Fault detection method, device and equipment of operation and maintenance platform and storage medium |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lee et al. | Comparative analysis of energy management strategies for HEV: Dynamic programming and reinforcement learning | |
| Wegener et al. | Automated eco-driving in urban scenarios using deep reinforcement learning | |
| CN109791409B (en) | Motion control decision for autonomous vehicles | |
| Su et al. | A cooperative collision-avoidance control methodology for virtual coupling trains | |
| Jin et al. | A group-based traffic signal control with adaptive learning ability | |
| Wang et al. | Train trajectory optimization for improved on-time arrival under parametric uncertainty | |
| Gaikwad et al. | Vehicle velocity prediction and energy management strategy part 2: Integration of machine learning vehicle velocity prediction with optimal energy management to improve fuel economy | |
| CN108873905A (en) | Automatic driving vehicle control method, automatic driving vehicle and readable storage medium storing program for executing | |
| CN112400192A (en) | Method and system for multi-modal deep traffic signal control | |
| US11938969B2 (en) | Intelligent dynamic parking for autonomous vehicles | |
| CN115303290B (en) | System key level switching method and system of vehicle hybrid key level system | |
| CN114074680B (en) | Vehicle channel change behavior decision method and system based on deep reinforcement learning | |
| CN116311948B (en) | Vehicle path planning method based on traffic flow speed prediction and signal lamp state | |
| CN115330095A (en) | Mine car dispatching model training method, device, chip, terminal, equipment and medium | |
| CN115782880A (en) | Intelligent automobile lane change decision-making method and device, electronic equipment and storage medium | |
| CN115496201A (en) | Train accurate parking control method based on deep reinforcement learning | |
| CN114399107A (en) | Prediction method and system of traffic state perception information | |
| Hayat et al. | A holistic approach to the energy-efficient smoothing of traffic via autonomous vehicles | |
| CN116662815B (en) | Training method of time prediction model and related equipment | |
| CN114118547A (en) | Electric vehicle public charging station queuing waiting time estimation method and system | |
| CN115973179A (en) | Model training method, vehicle control method, device, electronic equipment and vehicle | |
| CN114030488B (en) | Method and device for realizing automatic driving decision, computer storage medium and terminal | |
| CN115454082A (en) | Vehicle obstacle avoidance method and system, computer readable storage medium and electronic device | |
| Bennajeh et al. | Anticipation model based on a modified fuzzy logic approach | |
| CN115597619A (en) | Vehicle path planning method and device, electronic equipment and automatic driving vehicle |
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 |