CN111985092A - Intelligent automobile simulation test matrix generation method - Google Patents

Intelligent automobile simulation test matrix generation method Download PDF

Info

Publication number
CN111985092A
CN111985092A CN202010750334.6A CN202010750334A CN111985092A CN 111985092 A CN111985092 A CN 111985092A CN 202010750334 A CN202010750334 A CN 202010750334A CN 111985092 A CN111985092 A CN 111985092A
Authority
CN
China
Prior art keywords
simulation test
lstm
test matrix
cstm
matrix
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
Application number
CN202010750334.6A
Other languages
Chinese (zh)
Other versions
CN111985092B (en
Inventor
李伟
李爽
杨明
李鹏辉
陈龙
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Harbin Institute of Technology
Original Assignee
Harbin Institute of Technology
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Harbin Institute of Technology filed Critical Harbin Institute of Technology
Priority to CN202010750334.6A priority Critical patent/CN111985092B/en
Priority claimed from CN202010750334.6A external-priority patent/CN111985092B/en
Publication of CN111985092A publication Critical patent/CN111985092A/en
Application granted granted Critical
Publication of CN111985092B publication Critical patent/CN111985092B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/20—Design optimisation, verification or simulation
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
    • G06F17/10—Complex mathematical operations
    • G06F17/16—Matrix or vector computation, e.g. matrix-matrix or matrix-vector multiplication, matrix factorization
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/004—Artificial life, i.e. computing arrangements simulating life
    • G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Mathematical Physics (AREA)
  • General Engineering & Computer Science (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Biophysics (AREA)
  • Data Mining & Analysis (AREA)
  • Software Systems (AREA)
  • Evolutionary Computation (AREA)
  • Computing Systems (AREA)
  • Mathematical Analysis (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Computational Linguistics (AREA)
  • Biomedical Technology (AREA)
  • Artificial Intelligence (AREA)
  • Evolutionary Biology (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Computational Mathematics (AREA)
  • Molecular Biology (AREA)
  • Mathematical Optimization (AREA)
  • Pure & Applied Mathematics (AREA)
  • General Health & Medical Sciences (AREA)
  • Databases & Information Systems (AREA)
  • Algebra (AREA)
  • Geometry (AREA)
  • Computer Hardware Design (AREA)
  • Physiology (AREA)
  • Genetics & Genomics (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)
  • Traffic Control Systems (AREA)

Abstract

The invention discloses a method for generating an intelligent automobile simulation test matrix, and belongs to the technical field of intelligent automobile simulation experiments. The generation method comprises the following steps: firstly, extracting a simulation test matrix CSTM based on natural driving data; step two, generating a new simulation test matrix LSTM by using COLHS based on the simulation test matrix CSTMcor. The invention adopts an optimized Latin hypercube algorithm based on correlation control to generate an intelligent automobile simulation test matrix. The method can ensure that the simulation test matrix has good space coverage under any case number on one hand, and can also keep the correlation among variables in CSTM on the other hand. The COLHS used in the invention is a Cholesky decomposition method and a combined optimization methodA combined algorithm. The method can finish sampling at the non-positive timing of the correlation coefficient matrix, and can quickly and accurately control the correlation of the simulation test matrix at the positive timing of the correlation coefficient matrix.

Description

Intelligent automobile simulation test matrix generation method
Technical Field
The invention relates to a method for generating an intelligent automobile simulation test matrix, and belongs to the technical field of intelligent automobile simulation experiments.
Background
In order to link the intelligent automobile simulation experiment with a real traffic scene, many scholars research a simulation test matrix method based on natural driving data. The method for extracting the simulation test matrix by taking the movement of traffic participation as a Markov random process (the simulation test matrix extracted directly by depending on natural driving data is recorded as CSTM, each row of the CSTM corresponds to a test case, and each column corresponds to a variable) has been widely researched due to the fact that the method conforms to the randomness of traffic participants, and the reusability is high. However, in the early stages of development of smart-drive vehicle control algorithms, the direct use of CSTM was time consuming and not necessary. In addition, in some extreme cases, the extraction of natural driving data is difficult (such as extreme weather, extreme environment, extreme dangerous scene, etc.), and the case in CSTM is not enough. In order to solve the two problems, the method for extracting the simulation test matrix by means of the Markov random process and generating the simulation test matrix by means of optimized Latin hypercube sampling is provided. The invention can generate a simulation test matrix with any number of cases, good space coverage and very close to the correlation coefficient matrix of the CSTM according to the user requirements. The field of direct use of the method belongs to the field of intelligent automobile simulation testing.
Disclosure of Invention
The invention aims to provide a method for generating an intelligent automobile simulation test matrix to solve the problems in the existing simulation test.
A generation method of an intelligent automobile simulation test matrix comprises the following steps:
firstly, extracting a simulation test matrix CSTM based on natural driving data;
step two, generating a new simulation test matrix LSTM by using COLHS based on the simulation test matrix CSTMcor。
Further, in the step one, the method specifically comprises the following steps:
extracting a specified scene event from natural driving data;
step two, extracting simulation interested variables from specified scene events;
and step three, directly storing the static variables into the CSTM, regarding the dynamic variables as a Markov random process and fitting the Markov random process into a Markov random model, and then storing the model parameters into the CSTM.
Further, the scene events include a car following event, a lane changing event, a pre-crash event, a rider event, and a pedestrian event.
Further, in step one, the conditions of the specified scene event include weather conditions, lighting conditions, traffic participant behavior, driving environment conditions, and vehicle driving data.
Further, in step one and three, the static variables comprise the state of the vehicle, the state of the road, the state of the environment and the state of other traffic participants in the scene event.
Further, in the step one and three, the dynamic variables comprise the dynamic parameters of the self vehicle, the dynamic parameters of the self vehicle relative to other traffic participants and the dynamic parameters of the self other traffic participants.
Further, in the second step, the method specifically comprises the following steps:
step two, LHS sampling is carried out on CSTM, and a Latin hypercube simulation test matrix before correlation control is obtained and recorded as LSTM;
step two, extracting a CSTM correlation coefficient matrix, and recording the CSTM correlation coefficient matrix as Maarixcor；
Step two and step three, judging MatrixcorWhether is positive, and nsIf greater than m, if positive, and ns>m, executing the second step and the fourth step; otherwise, executing the step two;
step two, performing cholesky decomposition on the LSTM to obtain the LSTM 'subjected to correlation control, judging whether the correlation error of the LSTM' meets the requirement, and if not, performingStep two, step five; if so, LSTMcor＝LSTM′，LSTMcorThe simulation test matrix is obtained;
step two and five, the LSTM (or LSTM') is executed with a combined optimization method to obtain the final LSTMcor。
The main advantages of the invention are: the invention provides a method for generating an intelligent automobile simulation test matrix, which adopts an optimized Latin hypercube algorithm (COLHS) based on correlation control to generate the intelligent automobile simulation test matrix. The method can ensure that the simulation test matrix has good space coverage under any case number on one hand, and can also keep the correlation among variables in CSTM on the other hand. The COLHS used in the present invention is an algorithm that combines Cholesky decomposition and combinatorial optimization. The method can finish sampling at the non-positive timing of the correlation coefficient matrix, and can quickly and accurately control the correlation of the simulation test matrix at the positive timing of the correlation coefficient matrix.
Drawings
Figure 1 is a schematic view of LHS sampling;
FIG. 2 is a schematic flow chart of a method for generating an intelligent automobile simulation test matrix according to the present invention.
Detailed Description
The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, 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 invention.
The Latin hypercube sampling technique (LHS) is widely used in the field of simulation experiment design because it can make sample points uniform by layered sampling and comprehensively cover the distribution range of experimental variables. When experimental variables related to a simulation experiment are independent from each other, in order to make sample points uniformly fill the entire sample space, an Optimized Latin square sampling technique (hereinafter referred to as OLHS) based on a space filling criterion is mainly adopted at present. And when the experimental variables have correlation, correlation control needs to be carried out on the sampling result of the LHS.
The invention provides an intelligent automobile simulation test matrix generated by adopting an optimized Latin hypercube algorithm (COLHS) based on correlation control. The method can ensure that the simulation test matrix has good space coverage under any case number on one hand, and can also keep the correlation among variables in CSTM on the other hand. The COLHS used in the present invention is an algorithm that combines Cholesky decomposition and combinatorial optimization. The method can finish sampling at the non-positive timing of the correlation coefficient matrix, and can quickly and accurately control the correlation of the simulation test matrix at the positive timing of the correlation coefficient matrix.
A generation method of an intelligent automobile simulation test matrix comprises the following steps:
firstly, extracting a simulation test matrix CSTM based on natural driving data;
step two, generating a new simulation test matrix LSTM by using COLHS based on the simulation test matrix CSTMcor。
Further, in the step one, the method specifically comprises the following steps:
extracting a specified scene event from natural driving data;
step two, extracting variables (which can be divided into dynamic variables and static variables) of interest in simulation from specified scene events;
and step three, directly storing the static variables into the CSTM, regarding the dynamic variables as a Markov random process and fitting the Markov random process into a Markov random model, and then storing the model parameters into the CSTM.
Further, in step one, the specified scene event includes weather conditions (such as sunny days, rainy days, cloudy days, and the like), lighting conditions (such as lighting intensity, and the like), traffic participant behaviors (such as straight ahead driving, lane changing, steering, and the like), driving environment conditions (such as rural areas, cities, suburbs, and the like), and vehicle driving data (such as transverse/longitudinal/vertical speed, transverse/longitudinal/vertical acceleration, and the like).
Further, in step one and three, the static variables include the state of the own vehicle at the beginning of the event (such as the operation behavior of the driver, the initial speed, the distance of the own vehicle relative to other traffic participants, position information and the like), the state of the road (such as the number of lanes, the road level, the presence or absence of guardrails), the state of the environment (such as the weather condition, the illumination condition and the like) and the state of other traffic participants.
Further, in step one and three, the dynamic variables include the dynamic parameters of the own vehicle (such as the speed of the own vehicle, the acceleration of the own vehicle and the like in the whole scene event), the dynamic parameters of the own vehicle relative to other traffic participants (such as the transverse/longitudinal distance of the own vehicle relative to the front vehicle in the whole scene event and the like) and the dynamic parameters of the other traffic participants (such as the speed, the acceleration and the like of the front vehicle in the whole scene event).
Further, in the second step, the method specifically comprises the following steps:
step two, LHS sampling is carried out on CSTM, and a Latin hypercube simulation test matrix before correlation control is obtained and recorded as LSTM;
step two, extracting a CSTM correlation coefficient Matrix, and recording the CSTM correlation coefficient Matrix as Matrixcor；
Step two and step three, judging MaarixcorWhether is positive, and nsIf greater than m, if positive, and ns>m, executing the second step and the fourth step; otherwise, executing the step two;
step two and four, to MatrixcorExecuting cholesky decomposition to obtain LSTM 'subjected to correlation control, judging whether correlation errors of the LSTM' meet requirements or not, and if not, executing a fifth step; if so, LSTMcor＝LSTM′，LSTMcorThe simulation test matrix is obtained;
step two and five, to MaarixcorPerforming a combinatorial optimization method to obtain the final LSTMcor。
Specifically, S1, the CSTM is LHS sampled using Algorithm 1. Because the cumulative probability distribution function for each variable in the CSTM is unknown, a cumulative empirical distribution function is used instead. A sampling scheme is shown in figure 1.
Algorithm 1
Figure BDA0002609801450000051
S2, extracting a correlation coefficient Matrix (marked as Matrix) of the CSTMcor). The correlation coefficient matrix includes, but is not limited to, Spearman, Kendall, Pearson correlation coefficients.
S3, judging MatrixcorWhether is positive, and nsWhether greater than m. If positive, and ns>m, then S4 is connected, otherwise S5 is connected.
S4, executing a cholesky decomposition method, wherein the specific steps are shown as Algorithm 2
Algorithm 2
Figure BDA0002609801450000052
Figure BDA0002609801450000061
Judging whether the correlation error is required, if so, conforming to the LSTMcorThe required simulation test matrix is LSTM'. If not, S5 is followed.
S5, executing a combined optimization method to obtain the final LSTMcor. It should be noted that if the algorithm has executed S4, the LSTM described in this step is replaced with LSTM'.
The combinatorial optimization method is a common method for processing discrete problems by using an intelligent optimization algorithm, and a simulated annealing algorithm, a particle swarm algorithm, a genetic algorithm and the like are common. When applied to LHS dependency control, their main feature is that the dependency relationship between columns is "changed" by "swapping" the positions of several elements of each column in the LSTM matrix. The "change" is then made to gradually "approximate" the correlation matrix of LSTM to that of CSTM, according to the defined objective function.
Specifically, the exchange, the change and the approximation are realized by a defined disturbance operator, an acceptance criterion and the like in a simulated annealing algorithm; in the genetic algorithm, the method is realized by 'crossing', 'mutation' and 'selection' operators; the particle swarm optimization is realized through defined particle 'moving' and 'perturbation' operators.
In particular, if MatrixcorFor the positive definite matrix, an initial solution of the combinatorial optimization algorithm is generated by S4 (initial solution in simulated annealing algorithm, initial generation particles in particle swarm algorithm, initial population in genetic algorithm). If MatrixcorFor non-positive definite matrices, the initial solution for the combinatorial optimization Algorithm is randomly generated (3-6 steps in Algorithm 2 are called, with random ordering for each column in the LSTM).
One specific embodiment is set forth below:
the effectiveness of the invention is illustrated by taking the extraction of a car following scene simulation test case based on natural driving data as an example.
1. Extracting car following events
First, we extracted 2917 car-following events from 30 km of natural driving data provided by the research institute of automotive engineering in china. The extraction criteria were as follows:
(1)vr≥0m/s
(2)vl≥0m/s
(3)Rl(t)∈(0.1m，90m)
(4) no other vehicles between the front and rear vehicles cut into
(5) The front vehicle and the rear vehicle do not change the lane
(6) The car following length is more than or equal to 20s
vlAnd vrRespectively representing the speed of the front vehicle and the speed of the rear vehicle, RlThe relative distance between the front and rear cars.
2. Extraction of CSTM
And then establishing a front vehicle random model in a vehicle following scene according to the following formula.
Figure BDA0002609801450000071
rh～N(μr，σr 2)
Figure BDA0002609801450000072
Figure BDA0002609801450000073
Figure BDA0002609801450000074
Wherein H ═ H0，h1，h2]TIs a model parameter vector, rhTo follow a normal distribution of random numbers, the mean of the normal distribution is μrVariance is σr，alThe acceleration of the vehicle ahead is represented, and t is 0.04s, which is a sampling period. a islerrorIs rhWhen a is 0lAnd
Figure BDA0002609801450000075
the error between. H can be formed fromlMu. is calculated by robust regression modelrAnd σrCan be formed bylerrorCalculated using a Gaussian mixture model (Gaussian mixture model).
The CSTM in the following scene is defined as follows:
Figure BDA0002609801450000076
wherein v isrstart,vLstart,aLstart，RlstartRespectively, the initial velocity of the following vehicle, the initial velocity of the preceding vehicle, the initial acceleration of the preceding vehicle, and the relative distance between the preceding vehicle and the following vehicle at the initial time. T represents the number of iterations of equation 1, also for each car following eventThe sampling frequency of (2). n represents the number of following events.
3. Obtaining LSTMcor
To further illustrate the superiority of the present invention, we separately generated different n with the present inventionsThe lower LSTM. Wherein n issThe values are shown in table 1. The combinatorial optimization method used in this example is genetic algorithm-based combinatorial optimization (GA), the parameter settings are shown in table 2, and the correlation coefficient matrices used are all sperman correlation coefficients. In this example, the correlation coefficient matrix corresponding to the CSTM is a positive definite matrix.
Figure BDA0002609801450000077
Figure BDA0002609801450000081
TABLE 1 Small samples nsValue taking
Name of variable Value taking
pnum 10
mutation 0.1
itmax 1000
Table 2. the objective function of the GA parameter combination optimization method is shown in equation 2:
Figure BDA0002609801450000082
where F represents the fitness value of the objective function, and the optimization objective is to minimize F. i represents the ith row of the matrix and j represents the jth column of the matrix. W is a weight matrix that can be adjusted by the user of the present invention when the user for some reason has a focus on the relevance of the variables in the optimization process. A is the correlation coefficient matrix of LSTM (or LSTM'), and T is the correlation coefficient matrix of CSTM.
The operation is carried out for 10 times under each experiment number, a group of LSTMs is obtained by one operation, the mean value of the fitness value of the objective function is obtained, and the average calculation time length is shown in a table 3.
Figure BDA0002609801450000083
TABLE 3 fitness value and calculated time average
4.LSTMcorThe beneficial effects of (1).
(1) As shown in Table 3, LSTMcorThe correlation coefficient matrix of (a) is very close to CSTM. Therefore LSTMcorThe degree of correlation between variables in the CSTM is preserved to the greatest extent. This is of positive significance for simulation testing.
(2) This allows LSTM to be constructed due to the homogeneity of Latin hypercube sampling itselfcorThe distribution space to each variable can be well covered.
(3) Due to LSTMcorThe number of cases in (1) can be flexibly adjusted. Thus, at the beginning of the development of smart-drive automotive control algorithms, a small sample of LSTM may be usedcorAnd the CSTM is replaced, so that the coverage of the variable interval is ensured, and the simulation efficiency is improved. Additionally, in extreme cases where extraction of natural driving data is difficult (e.g., extreme weather, extreme environment, extreme dangerous scene, etc.), the LSTM may be usedcorTo expand the cases in CSTM.

Claims (7)

1. A method for generating an intelligent automobile simulation test matrix is characterized by comprising the following steps:
firstly, extracting a simulation test matrix CSTM based on natural driving data;
step two, generating a new simulation test matrix LSTM by using COLHS based on the simulation test matrix CSTMcor。
2. The method for generating the intelligent automobile simulation test matrix according to claim 1, wherein in the first step, the method specifically comprises the following steps:
extracting a specified scene event from natural driving data;
step two, extracting simulation interested variables from specified scene events;
and step three, directly storing the static variables into the CSTM, regarding the dynamic variables as a Markov random process and fitting the Markov random process into a Markov random model, and then storing the model parameters into the CSTM.
3. The method as claimed in claim 2, wherein the scene event includes a car following event, a lane changing event, a pre-collision event, a rider event and a pedestrian event.
4. The method as claimed in claim 2, wherein in step one, the conditions of the designated scene event include weather conditions, lighting conditions, traffic participant behavior, driving environment conditions and vehicle driving data.
5. The method for generating the intelligent automobile simulation test matrix according to claim 2, wherein in the first step three, the static variables comprise the motion state of the automobile, the road state, the environment state and the motion state of other traffic participants in the scene event.
6. The method as claimed in claim 2, wherein in step one or three, the dynamic variables include a dynamic parameter of the vehicle, a dynamic parameter of the vehicle relative to other traffic participants, and a dynamic parameter of the other traffic participants.
7. The method for generating the intelligent automobile simulation test matrix according to claim 1, wherein in the second step, the method specifically comprises the following steps:
step two, LHS sampling is carried out on CSTM, and a Latin hypercube simulation test matrix before correlation control is obtained and recorded as LSTM;
step two, extracting a CSTM correlation coefficient Matrix, and recording the CSTM correlation coefficient Matrix as Matrixcor；
Step two and step three, judging MatrixcorWhether is positive, and nsIf greater than m, if positive, and ns>m, executing the second step and the fourth step; otherwise, executing the step two;
step two, performing cholesky decomposition on the LSTM to obtain the LSTM 'subjected to correlation control, judging whether the correlation error of the LSTM' meets the requirement, and if not, performing step two; if so, LSTMcor＝LSTM′，LSTMcorThe simulation test matrix is obtained;
step two and five, the LSTM (or LSTM') is executed with a combined optimization method to obtain the final LSTMcor。
CN202010750334.6A 2020-07-30 Intelligent automobile simulation test matrix generation method Active CN111985092B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202010750334.6A CN111985092B (en) 2020-07-30 Intelligent automobile simulation test matrix generation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202010750334.6A CN111985092B (en) 2020-07-30 Intelligent automobile simulation test matrix generation method

Publications (2)

Publication Number Publication Date
CN111985092A true CN111985092A (en) 2020-11-24
CN111985092B CN111985092B (en) 2024-05-31

Family

ID=

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112346998A (en) * 2021-01-11 2021-02-09 北京赛目科技有限公司 Automatic driving simulation test method and device based on scene
CN113066282A (en) * 2021-02-26 2021-07-02 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) Vehicle-following coupling relation modeling method and system in mixed-line environment
CN113823096A (en) * 2021-11-25 2021-12-21 禾多科技（北京）有限公司 Random traffic flow barrier object arrangement strategy for simulation test

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104298540A (en) * 2014-10-23 2015-01-21 浙江大学 Underlaying model parameter correction method of microscopic traffic simulation software
CN105608267A (en) * 2015-12-21 2016-05-25 许昌学院 Multivariable global optimization algorithm
CN105740592A (en) * 2016-05-05 2016-07-06 中国人民解放军国防科学技术大学 Latin hypercube experiment design method based on sequential sampling
CN107038306A (en) * 2017-04-14 2017-08-11 哈尔滨工业大学 A kind of optimal Latin hypercube experimental design method based on self-adapted genetic algorithm
CN107436971A (en) * 2017-07-07 2017-12-05 东南大学 Suitable for the improvement Latin Hypercube Sampling method of non-positive definite form correlation control
CN110263381A (en) * 2019-05-27 2019-09-20 南京航空航天大学 A kind of automatic driving vehicle test emulation scene generating method

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104298540A (en) * 2014-10-23 2015-01-21 浙江大学 Underlaying model parameter correction method of microscopic traffic simulation software
CN105608267A (en) * 2015-12-21 2016-05-25 许昌学院 Multivariable global optimization algorithm
CN105740592A (en) * 2016-05-05 2016-07-06 中国人民解放军国防科学技术大学 Latin hypercube experiment design method based on sequential sampling
CN107038306A (en) * 2017-04-14 2017-08-11 哈尔滨工业大学 A kind of optimal Latin hypercube experimental design method based on self-adapted genetic algorithm
CN107436971A (en) * 2017-07-07 2017-12-05 东南大学 Suitable for the improvement Latin Hypercube Sampling method of non-positive definite form correlation control
CN110263381A (en) * 2019-05-27 2019-09-20 南京航空航天大学 A kind of automatic driving vehicle test emulation scene generating method

Cited By (4)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN112346998A (en) * 2021-01-11 2021-02-09 北京赛目科技有限公司 Automatic driving simulation test method and device based on scene
CN113066282A (en) * 2021-02-26 2021-07-02 北京航空航天大学合肥创新研究院(北京航空航天大学合肥研究生院) Vehicle-following coupling relation modeling method and system in mixed-line environment
CN113823096A (en) * 2021-11-25 2021-12-21 禾多科技（北京）有限公司 Random traffic flow barrier object arrangement strategy for simulation test
CN113823096B (en) * 2021-11-25 2022-02-08 禾多科技（北京）有限公司 Random traffic flow obstacle object arrangement method for simulation test

Similar Documents

Publication Publication Date Title
US20210197851A1 (en) Method for building virtual scenario library for autonomous vehicle
CN112036001B (en) Automatic driving test scene construction method, device, equipment and readable storage medium
Jia et al. Long short‐term memory and convolutional neural network for abnormal driving behaviour recognition
CN112487617A (en) Collision model-based risk prevention method, device, equipment and storage medium
Huang et al. Evaluation of automated vehicles in the frontal cut-in scenario—An enhanced approach using piecewise mixture models
CN113010967A (en) Intelligent automobile in-loop simulation test method based on mixed traffic flow model
CN111079800B (en) Acceleration method and acceleration system for intelligent driving virtual test
Huang et al. A versatile approach to evaluating and testing automated vehicles based on kernel methods
CN110807123A (en) Vehicle length calculation method, device and system, computer equipment and storage medium
CN109887279B (en) Traffic jam prediction method and system
CN115238958A (en) Dangerous event chain extraction method and system based on complex traffic scene
CN114493191A (en) Driving behavior modeling analysis method based on network appointment data
CN113094808A (en) Simulation data and artificial intelligence based automobile collision damage grade real-time prediction method
Jiang et al. Efficient and unbiased safety test for autonomous driving systems
Pandita et al. Preceding vehicle state prediction
CN113642114A (en) Modeling method for humanoid random car following driving behavior capable of making mistakes
CN111985092A (en) Intelligent automobile simulation test matrix generation method
CN111985092B (en) Intelligent automobile simulation test matrix generation method
CN110213741B (en) Method for detecting authenticity of vehicle sending information in real time based on width learning
CN115482662B (en) Method and system for predicting collision avoidance behavior of driver under dangerous working condition
CN109583632B (en) Electric vehicle charging behavior prediction method based on internet data
CN116946183A (en) Commercial vehicle driving behavior prediction method considering driving capability and vehicle equipment
CN115063976B (en) Vehicle conflict risk assessment and prediction method based on multichannel convolutional neural network
Xing et al. Optimizing longitudinal control model parameters of connected and automated vehicles using empirical trajectory data of human drivers in risky car-following scenarios
Weber et al. Data-driven bev modeling for realistic consumption calculation in traffic simulation

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