CN114239423A - Method for constructing prediction model of danger perception capability of driver on long and large continuous longitudinal slope section - Google Patents
Method for constructing prediction model of danger perception capability of driver on long and large continuous longitudinal slope section Download PDFInfo
- Publication number
- CN114239423A CN114239423A CN202210175408.7A CN202210175408A CN114239423A CN 114239423 A CN114239423 A CN 114239423A CN 202210175408 A CN202210175408 A CN 202210175408A CN 114239423 A CN114239423 A CN 114239423A
- Authority
- CN
- China
- Prior art keywords
- driver
- danger
- model
- training
- training set
- 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 79
- 238000000034 method Methods 0.000 title claims abstract description 21
- 238000012549 training Methods 0.000 claims abstract description 71
- 238000010276 construction Methods 0.000 claims abstract description 7
- 230000008859 change Effects 0.000 claims description 23
- 230000004434 saccadic eye movement Effects 0.000 claims description 8
- 238000007476 Maximum Likelihood Methods 0.000 claims description 6
- 230000001186 cumulative effect Effects 0.000 claims description 6
- 238000004364 calculation method Methods 0.000 claims description 5
- 150000001875 compounds Chemical class 0.000 claims description 3
- 238000001514 detection method Methods 0.000 abstract description 2
- 238000011897 real-time detection Methods 0.000 abstract description 2
- 230000006399 behavior Effects 0.000 description 19
- 238000004458 analytical method Methods 0.000 description 12
- 210000001747 pupil Anatomy 0.000 description 5
- 239000000243 solution Substances 0.000 description 5
- 238000010219 correlation analysis Methods 0.000 description 4
- 239000011159 matrix material Substances 0.000 description 4
- 230000008569 process Effects 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 3
- 238000011160 research Methods 0.000 description 3
- 230000007704 transition Effects 0.000 description 3
- 230000034994 death Effects 0.000 description 2
- 231100000517 death Toxicity 0.000 description 2
- 239000013598 vector Substances 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 206010039203 Road traffic accident Diseases 0.000 description 1
- 238000012512 characterization method Methods 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 230000001351 cycling effect Effects 0.000 description 1
- 238000002474 experimental method Methods 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- PCHJSUWPFVWCPO-UHFFFAOYSA-N gold Chemical compound [Au] PCHJSUWPFVWCPO-UHFFFAOYSA-N 0.000 description 1
- 239000010931 gold Substances 0.000 description 1
- 229910052737 gold Inorganic materials 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000002347 injection Methods 0.000 description 1
- 239000007924 injection Substances 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000003863 physical function Effects 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 238000012502 risk assessment Methods 0.000 description 1
- 238000013179 statistical model Methods 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
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
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
- G06F18/214—Generating training patterns; Bootstrap methods, e.g. bagging or boosting
-
- 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
- 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
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0639—Performance analysis of employees; Performance analysis of enterprise or organisation operations
- G06Q10/06393—Score-carding, benchmarking or key performance indicator [KPI] analysis
-
- 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)
- Business, Economics & Management (AREA)
- Human Resources & Organizations (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Economics (AREA)
- General Physics & Mathematics (AREA)
- General Business, Economics & Management (AREA)
- Tourism & Hospitality (AREA)
- Marketing (AREA)
- Entrepreneurship & Innovation (AREA)
- Evolutionary Computation (AREA)
- Development Economics (AREA)
- Data Mining & Analysis (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Game Theory and Decision Science (AREA)
- Artificial Intelligence (AREA)
- Quality & Reliability (AREA)
- Educational Administration (AREA)
- General Engineering & Computer Science (AREA)
- Operations Research (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Evolutionary Biology (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- Health & Medical Sciences (AREA)
- General Health & Medical Sciences (AREA)
- Primary Health Care (AREA)
- Traffic Control Systems (AREA)
Abstract
The invention belongs to the field of road safety detection. The invention provides a method for constructing a driver danger perception capability prediction model on a long and large continuous longitudinal slope section, which comprises the following steps of firstly, selecting training data, and constructing a safe training set and an unsafe training set; secondly, constructing an HMM model and initializing model parameters of the HMM model; then, training the HMM model by adopting a safety training set until the prediction performance reaches a preset value, and stopping training to obtain a safety HMM model; training the HMM model by adopting an unsafe training set, and stopping training until the prediction performance reaches a preset value to obtain an unsafe HMM model; and finishing the construction of the prediction model of the danger perception capability of the driver on the long and large continuous longitudinal slope section. The real-time detection of whether the driving state of the driver is in a safe state or not is realized.
Description
Technical Field
The invention belongs to the field of road safety detection, and particularly relates to a method for constructing a driver danger perception capability prediction model of a long and large continuous longitudinal slope section.
Background
Wilde's risk balance theory suggests that there is some balance between subjective risk and objective risk values for drivers, i.e., drivers adjust their driving behavior to maintain such balance. The driver's subjective perception of danger is therefore of great importance to driving safety. Objective risk and subjective risk are two levels of risk perception. Objective risk refers to various and specific obstacles encountered by the driver during driving, such as complicated climate conditions, poor road alignment, interference of other road participants, and the like. The gold festival states that objective risk is often expressed in terms of the distance of occurrence of the risk, the number of annual deaths, etc. The subjective risk refers to the subjective feeling value of the driver in the driving process and can be judged by the driver.
The danger perception capability may be understood as the driver's perception level of the degree of danger of the road. The driver makes reasonable driving behavior prejudgment by acquiring external driving information and further judging road conditions, and finally executes control of the vehicle.
Hidden Markov Models (HMMs) are concepts proposed based on the field of speech recognition, abbreviated HMMs. On one hand, when the HMM is applied in the acoustic field, each stable pronunciation unit is represented by an implicit sequence, and the change of sound is presented by state transition and residence; on the other hand, the HMM introduces a probability statistical model in the output probability calculation, and performs analysis on the speech parameters by using a probability density function, thereby determining the most reasonable state transition sequence, obtaining the maximum posterior probability, and identifying the maximum posterior probability.
The theoretical basis of the HMM model is Markov theory, a chain in the model corresponds to the state change trend, and the HMM state has no testability, but the state can be presented through a quantization index. The index changes according to the change of the state, so the state conditions at different moments can be known really through the index change.
How to construct a prediction model of the driver danger perception capability by using a hidden Markov model and realize the evaluation of uncertainty and invisibility of a driving state is a technical problem to be solved by the invention.
Disclosure of Invention
The invention aims to provide a method for constructing a prediction model of the danger perception capability of a driver on a long and large continuous longitudinal slope section aiming at the problems of uncertainty and invisibility of the driving state in the driving process, so as to realize the real-time detection of whether the driving state of the driver is in a safe state.
In order to achieve the purpose, the invention provides a method for constructing a driver danger perception capability prediction model on a long and large continuous longitudinal slope section, which comprises the following steps:
Specifically, in step 1, the calculation formula of the driver risk perception value is as follows:
in the formula (I), the compound is shown in the specification,
a driver risk perception value;
the subjective danger degree of the driver is divided into 10 grades, and the higher the score is, the higher the subjective danger perception degree of the driver is represented;
the objective risk degree of the road is 10 points, and the higher the score is, the higher the objective risk degree of the road section is; i is a road section number; j is the driver number.
Further, the objective road risk is determined by the cumulative frequency of road accident rates.
Further, in step 2, the HMM model is a left-right chain structure.
Further, in the step 3, after initializing the safety training set by means of a Baum-Welch algorithm, putting the safety training set into an HMM model for training, and obtaining a maximum likelihood solution of the safety HMM model through repeated iteration; initializing an unsafe training set by using a Baum-Welch algorithm, putting the unsafe training set into an HMM model for training, and obtaining a maximum likelihood solution of the unsafe HMM model after repeated iteration; and finishing the construction of the prediction model of the danger perception capability of the driver on the long and large continuous longitudinal slope section.
The invention has the advantages that the invention provides the driver subjective danger degree and the road passenger by innovationDetermining a driver danger perception value according to the ratio of the observation danger degree; and selecting the fixation time percentage, the saccade time percentage, the heart rate change rate and the vehicle speed as observation indexes for evaluating the danger perception capability, and determining two hidden states of safety and non-safety. Dividing the training set into a safe training set and an unsafe training set according to the driver danger perception value, training the HMM model by using the safe training set, training the HMM model by using the unsafe training set, and optimizing and perfecting the HMM model by using the training sample by means of the Baum-Welch algorithm to realize
And modifying the model parameters by using a multi-iteration optimization mode, thereby realizing the construction of the driver danger perception capability prediction model of the long and large continuous longitudinal slope section.
Drawings
Fig. 1 is a graph showing cumulative frequency of accident rate in example 1 of the present invention.
FIG. 2 is a graph of percentage of injection time versus risk perception in example 1 of the present invention.
FIG. 3 is a graph of percent scan time versus risk perception in example 1 of the present invention.
FIG. 4 is a graph of the rate of change of the center rate and the danger-sensing capability in example 1 of the present invention.
Fig. 5 is a graph of vehicle speed versus risk perception capability in embodiment 1 of the present invention.
Fig. 6 is a graph of the pupil diameter change rate and the risk perception capability in embodiment 1 of the present invention.
Detailed Description
The technical solution of the present invention will be described in detail with reference to the following examples.
Example 1
The embodiment provides a method for constructing a driver danger perception capability prediction model of a long and large continuous longitudinal slope section, which comprises the following steps of:
firstly, selecting a training sample:
selecting training data, and constructing a safe training set and an unsafe training set; the safety training set comprises a fixation time percentage, a saccade time percentage, a heart rate change rate and a vehicle speed which correspond to the condition that the danger perception value of the driver is more than or equal to 1; the unsafe training set comprises a fixation time percentage, a glance time percentage, a heart rate change rate and a vehicle speed corresponding to the condition that the danger perception value of the driver is smaller than 1.
The calculation formula of the driver danger perception value is as follows:
in the formula (I), the compound is shown in the specification,
a driver risk perception value;
the subjective danger degree of the driver is divided into 10 grades, and the higher the score is, the higher the subjective danger perception degree of the driver is represented;
the objective risk degree of the road is 10 points, and the higher the score is, the higher the objective risk degree of the road section is; i is a road section number; j is the driver number.
The driver danger perception capability comprises an objective danger value and a driver subjective danger perception degree. Due to the fact that the factors such as knowledge, driving experience, character and gender of the drivers are different, different drivers have different actual danger perceptions and judgments, namely different drivers can present different danger perceptions. The danger perceived by the driver is influenced by the objective danger and the perceived characteristics of the danger itself.
(1) For objective risk, the analysis is as follows:
objective risk is often expressed in terms of the distance at which the risk occurs, the number of annual deaths, etc. The objective risk of the long and large continuous longitudinal slope section can be used as a traffic accident risk analysis method for evaluating the objective risk of the long and large continuous longitudinal slope section. In the embodiment, the road section accident rate statistical table in the table 1 is used as a sample to realize the objective risk evaluation. According to table 1, a cumulative frequency curve of the accident rate is shown in fig. 1. It can be seen that the cumulative frequency percentage of the road section with the accident rate of 0 reaches 53.1%, the maximum value of the accident rate is 7.52 times per million kilometers, and the cumulative frequency percentage changes suddenly when the accident rate is 2.26 times per million kilometers, 4.11 times per million kilometers, or 5.52 times per million kilometers, so the objective risk levels and the risk values of different road sections of the long and large continuous longitudinal slope road section are divided according to the accident rate, and the result is shown in table 2, the objective risk of the road is realized by adopting a score of 10 cents, and the higher score represents the higher objective risk of the road section.
TABLE 1 statistical table of accident rates of road sections
TABLE 2 Objective Risk ratings Scale Table
According to the standard, the objective risk degree distribution condition of the road on the test road section is obtained, and is shown in table 3.
TABLE 3 Objective road Risk
(2) For subjective risk, the analysis was as follows:
according to the inquiry of the subjective danger perception degrees of drivers in different scenes in the experimental process, the subjective danger degrees of the drivers in different road sections in the experimental process are counted, and the road section danger degrees are quantitatively assigned by adopting 10 grades, wherein 0 grade represents that the danger perception degree is lowest, namely the driving danger is lowest, 10 grades represent that the danger perception degree is highest, and the driving danger is highest, so that the situation is close to the scene which can not be controlled by the drivers. The road section and objective risk statistics are shown in table 4.
TABLE 4 subjective driver Risk
Comparing table 3 with table 4, it is known that the subjective risk perception of the driver is not completely consistent with the objective risk, which is determined by the difference of the driver's own experience, character, current driving state, and other factors. The driver risk perception value is obtained from the quotient of the driver subjective risk and the road objective risk, as shown in table 5.
TABLE 5 driver Risk perception values
As can be seen from table 5, the risk perception capability of less than 1 indicates that the risk perceived by the driver is lower than the objective risk, because the driver is inexperienced or is an aggressive driver, and the driving behavior is dangerous behavior regardless of the driving environment with low risk. The danger perception capability is larger than 1, which indicates that the objective danger is in the control range of the driver, the driver can completely perceive the danger existing in the actual environment, and the type of driver has rich driving experience, belongs to conservative and cautious drivers and belongs to safe driving behaviors.
(3) The factors having influence on the driver's perception of danger value are analyzed as follows:
in the example, the relationship between the driving behavior physiological indexes and the danger perception capability of different drivers on different road sections is obtained by respectively carrying out maximum value, minimum value and average value on 5 driving behavior psychophysical indexes of 5 drivers in an experiment, including average staring time percentage, glancing time percentage, heart rate change rate and vehicle speed, and combining the danger perception capability result of the drivers.
Firstly, the relationship between the psychological index of the driver and the danger perception ability
And (3) combining and analyzing the physiological indexes, the driving behaviors and the danger perception capability of the driver to obtain the variation trends of different driving behavior parameters and the danger perception capability. As shown in fig. 2 to 6, it can be seen that the driving behavior parameters of the average fixation time percentage, the average saccade time percentage, the pupil diameter change rate, the heart rate, the driving speed and the risk perception capability show different degrees of correlation under different road sections, and the correlation cannot be clearly known for the pupil diameter change rate and the average saccade time percentage, so the correlation is analyzed by using statistical software, and the influence of the driving behavior parameters on the risk perception of the driver is determined by combining the trends of the graphs.
Correlation analysis of psychological index and danger perception ability
And carrying out correlation analysis on the driving behavior parameters and the driver danger perception capability, thereby more accurately obtaining the relationship characteristics of the driving behavior parameters and the driver danger perception capability. Common correlation analysis methods include a Pearson analysis method and a Spearman analysis method, and the difference between the Pearson analysis method and the Spearman analysis method is that two variables to be tested are classified variables or have one classified variable, and the Spearman analysis method is used; if both variables are continuous variables, Pearson analysis is used. Since the risk perception ability of this document belongs to the categorical variable, Spearman analysis was used and the results are shown in table 6.
TABLE 6 analysis of correlation between driving behavior parameters and Risk perception
It is known from table 6 that at the confidence level of 0.01, the risk perception capability has a significant correlation with the fixation time percentage, the saccade time percentage, the heart rate change rate maximum, the heart rate change rate average, the driving speed maximum, and the driving speed average. The maximum value of the pupil diameter change rate and the minimum value of the pupil diameter change rate have no correlation with the danger perception capability of the driver, so that the indexes are not listed in a danger perception prediction model.
The watching time percentage and the glancing time percentage reflect the level of attention of a driver to a traffic environment, the heart rate change rate reflects the level of physical functions of the driver, and the vehicle speed reflects the level of controlling a vehicle by the driver, so that the four indexes can express driving behavior characteristics from three levels of information perception, driver psychological load and vehicle control, and the correlation with the level of danger perception of the driver is high.
Thus, this example selects a training set of fixation time percentages, glance time percentages, heart rate change rates, and vehicle velocity configuration models.
Secondly, constructing an HMM model, and initializing model parameters of the HMM model:
(1) selecting HMM structures
The states of A are different, and corresponding HMM function structures are also different, and comprise a left-right chain type, a parallel branch type and the like. For the left-right chain structure HMM, the specific characterization is as follows: the state of the model is not fixed, and the model can also be transferred, but cannot be transferred in a striding mode or a reverse direction. Driving is not intermittent but continuous, so the state is not fixed, which is determined by the actual situation the driver is in, in particular in relation to the observation sequence, in short the next state is associated with the current situation. For this reason, the model tool is herein "left-right chain type".
(2) Designing HMM models
The main elements studied in this example are 4: one is the percentage of fixation time; second percentage of glance time; third, the rate of heart rate increase; fourthly, the vehicle speed is used for analyzing the danger perception capability by using the indexes, and the accuracy of the prediction capability of the danger perception capability is analyzed, wherein 2 results are obtained, and the accuracy is correct; secondly, errors occur, the corresponding sample library is safe and unsafe, and the risk perception capability is at the level of 1 or above under the safe condition; the risk perception capability in unsafe situations is an index below 1.
Since the safety of the driving behavior of the driver cannot be evaluated immediately, but the driving behavior index of the driver has testability, the safety of the driving behavior can be evaluated through the driving behavior index, and the parameters are described and calculated by means of an HMM model based on the evaluation: λ = [ N, M, A, B, π]The method is simplified as follows: λ = [ A, B, π =]The specific meanings of the parameters are as follows: n is the number of states, N is the number of hidden states, and this study corresponds toThe sensing capabilities of (1) include 2, one is security; second is unsafe, so N = 2; m is the number of observation events for each state, and there are 4 specific influencing factors for this study, i.e. the number of observation states is 4, i.e. M = 4; a is a state transition matrix, which refers to a matrix corresponding to safe and unsafe conversion ratios, and the research selects an averaging method, and the corresponding expression is as follows:
(ii) a B is an observation event probability distribution matrix, which refers to an observation value probability matrix obtained correspondingly by the combination of 4 influence factors in the research; pi is the initial state vector, the research selects a mean value method, and the corresponding expression is
。
Thirdly, training a model:
training the HMM model by adopting a safety training set until the prediction performance reaches a preset value, and stopping training to obtain a safe HMM model; training the HMM model by adopting an unsafe training set, and stopping training until the prediction performance reaches a preset value to obtain an unsafe HMM model; and finishing the construction of the prediction model of the danger perception capability of the driver on the long and large continuous longitudinal slope section.
Training a safety model by adopting a safety training set: correspondingly assigning values of N =2 and M =4, initializing a safety training set by means of a Baum-Welch algorithm, putting the safety training set into an HMM model for training, and obtaining a maximum likelihood solution of the safety HMM model through repeated iteration, wherein the maximum likelihood solution is as follows:
for an observation sequence of HMMs, the column vectors are used to represent:
in this connection, it is possible to use,
mean fixation time percentage, saccade time, referred to in turnPercentage, heart rate change rate, and vehicle speed.
With the aid of the Baum-Welch algorithm, the initial parameters λ = [ A, B, π]And reestimating the observation sequence O (t) to obtain likelihood probability
(ii) a In which the new parameters
Derived from a correlation analysis
The former is more advantageous than λ. Over repeated iterations
The improvement is continuously carried out,
no further cycling is performed. That is when
Model of this phase, which is cut off, i.e. convergence begins, with continued improvement
The model is the model sought.
An unsafe training set is used to train an unsafe model in the same manner. And finally obtaining a driver danger perception capability prediction model of the long and large continuous longitudinal slope section.
Fourthly, detecting the model
The log-likelihood values for the safe and unsafe conditions were calculated as shown in table 7.
And analyzing whether the initial input value is matched with the model parameter value or not according to A, B and pi matching values, wherein the larger the index is, the better the matching degree is, and the 0 means the maximum matching degree. Here, the probability value is directly proportional to the matching degree, and when the index value is large, it indicates that the observation sequence and the model are fit to each other.
TABLE 7 prediction model log-likelihood value of danger perception capability of growing continuous longitudinal slope
The parameters were further analyzed according to the above criteria as follows:
for result verification identification, the accuracy can be expressed by the following formula:
in the formula: TRN and TON are the number of samples that are predicted accurately and the total number of samples, respectively, and the calculation results in this example are shown in table 8.
TABLE 8 prediction results of driver's danger feeling ability value on long and large continuous longitudinal slopes
As can be seen from Table 8, the prediction condition of the risk perception capability in the safe state is better, and the prediction accuracy in both states is more than 75% from the overall analysis, so that the prediction model of the risk perception capability of the driver in the long and large continuous longitudinal slope section constructed according to the invention is reliable.
Claims (5)
1. The method for constructing the driver danger perception capability prediction model of the long and large continuous longitudinal slope section is characterized by comprising the following steps of:
step 1, selecting training data, and constructing a safe training set and an unsafe training set; the safety training set comprises a fixation time percentage, a saccade time percentage, a heart rate change rate and a vehicle speed which correspond to the condition that the danger perception value of the driver is more than or equal to 1; the unsafe training set comprises a fixation time percentage, a glance time percentage, a heart rate change rate and a vehicle speed which correspond to the condition that the danger perception value of the driver is smaller than 1;
step 2, constructing an HMM model and initializing model parameters of the HMM model;
step 3, training the HMM model by adopting a safety training set, and stopping training until the prediction performance reaches a preset value to obtain a safety HMM model; training the HMM model by adopting an unsafe training set, and stopping training until the prediction performance reaches a preset value to obtain an unsafe HMM model; and finishing the construction of the prediction model of the danger perception capability of the driver on the long and large continuous longitudinal slope section.
2. The method for constructing the prediction model of the danger awareness capability of the driver on the long and continuous longitudinal slope section according to claim 1, wherein in the step 1, the calculation formula of the danger awareness value of the driver is as follows:
in the formula (I), the compound is shown in the specification,
a driver risk perception value;
the subjective danger degree of the driver is divided into 10 grades, and the higher the score is, the higher the subjective danger perception degree of the driver is represented;
the objective risk degree of the road is 10 points, and the higher the score is, the higher the objective risk degree of the road section is; i is a road section number; j is the driver number.
3. The method for constructing the prediction model of the danger awareness ability of the driver on the long and continuous longitudinal slope section according to claim 2, wherein the objective road danger degree is determined by the cumulative frequency of road accident rates.
4. The method for constructing the prediction model of danger awareness capability of drivers on long and continuous longitudinal slope sections according to claim 3, wherein in the step 2, the HMM model is of a left-right chain type structure.
5. The method for constructing the driver danger awareness prediction model of the long and large continuous longitudinal slope section according to claim 3, wherein in the step 3, after initializing a safety training set by means of a Baum-Welch algorithm, putting the safety training set into an HMM model for training, and obtaining a maximum likelihood solution of the safety HMM model through repeated iteration; initializing an unsafe training set by using a Baum-Welch algorithm, putting the unsafe training set into an HMM model for training, and obtaining a maximum likelihood solution of the unsafe HMM model after repeated iteration; and finishing the construction of the prediction model of the danger perception capability of the driver on the long and large continuous longitudinal slope section.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210175408.7A CN114239423A (en) | 2022-02-25 | 2022-02-25 | Method for constructing prediction model of danger perception capability of driver on long and large continuous longitudinal slope section |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210175408.7A CN114239423A (en) | 2022-02-25 | 2022-02-25 | Method for constructing prediction model of danger perception capability of driver on long and large continuous longitudinal slope section |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN114239423A true CN114239423A (en) | 2022-03-25 |
Family
ID=80748347
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210175408.7A Pending CN114239423A (en) | 2022-02-25 | 2022-02-25 | Method for constructing prediction model of danger perception capability of driver on long and large continuous longitudinal slope section |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN114239423A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115081756A (en) * | 2022-08-19 | 2022-09-20 | 四川省公路规划勘察设计研究院有限公司 | Road section brake drum temperature prediction and road section risk assessment method based on long and large longitudinal slope road section |
| CN115089182A (en) * | 2022-05-23 | 2022-09-23 | 长安大学 | Method for evaluating risk perception capability of multidimensional driver |
| CN115158274A (en) * | 2022-08-31 | 2022-10-11 | 四川省公路规划勘察设计研究院有限公司 | Long and large longitudinal slope dangerous road section identification method based on truck braking and heavy braking characteristics |
| CN115331449A (en) * | 2022-10-17 | 2022-11-11 | 四川省公路规划勘察设计研究院有限公司 | Method and device for identifying accident prone area of long and large continuous longitudinal slope section and electronic equipment |
| CN115320626A (en) * | 2022-10-11 | 2022-11-11 | 四川省公路规划勘察设计研究院有限公司 | Danger perception capability prediction method and device based on human-vehicle state and electronic equipment |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002260146A (en) * | 2001-03-02 | 2002-09-13 | Toyota Central Res & Dev Lab Inc | Driver risk recognition characteristic storage method, driver risk computing system, driving ability diagnosis and evaluation system, and preventive safety system for vehicle |
| CN106740864A (en) * | 2017-01-12 | 2017-05-31 | 北京交通大学 | A kind of driving behavior is intended to judge and Forecasting Methodology |
| CN113077129A (en) * | 2021-03-19 | 2021-07-06 | 东南大学 | Driver danger perception capability evaluation method suitable for high-altitude area |
| CN113449790A (en) * | 2021-06-25 | 2021-09-28 | 贵州省都匀公路管理局 | Mountain trunk highway high-risk road section identification method based on SVM |
-
2022
- 2022-02-25 CN CN202210175408.7A patent/CN114239423A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| JP2002260146A (en) * | 2001-03-02 | 2002-09-13 | Toyota Central Res & Dev Lab Inc | Driver risk recognition characteristic storage method, driver risk computing system, driving ability diagnosis and evaluation system, and preventive safety system for vehicle |
| CN106740864A (en) * | 2017-01-12 | 2017-05-31 | 北京交通大学 | A kind of driving behavior is intended to judge and Forecasting Methodology |
| CN113077129A (en) * | 2021-03-19 | 2021-07-06 | 东南大学 | Driver danger perception capability evaluation method suitable for high-altitude area |
| CN113449790A (en) * | 2021-06-25 | 2021-09-28 | 贵州省都匀公路管理局 | Mountain trunk highway high-risk road section identification method based on SVM |
Non-Patent Citations (3)
| Title |
|---|
| 刘平 等: "汶马高速公路长大下坡路段设计对策及管控措施研究", 《路基工程》 * |
| 张兴雨: "换道行为风险感知预测方法", 《中国优秀博硕士学位论文全文数据库(硕士)工程科技Ⅱ辑》 * |
| 艾倩楠: "基于驾驶员指标的驾驶员风险感知水平评价", 《中国安全科学学报》 * |
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN115089182A (en) * | 2022-05-23 | 2022-09-23 | 长安大学 | Method for evaluating risk perception capability of multidimensional driver |
| CN115081756A (en) * | 2022-08-19 | 2022-09-20 | 四川省公路规划勘察设计研究院有限公司 | Road section brake drum temperature prediction and road section risk assessment method based on long and large longitudinal slope road section |
| CN115158274A (en) * | 2022-08-31 | 2022-10-11 | 四川省公路规划勘察设计研究院有限公司 | Long and large longitudinal slope dangerous road section identification method based on truck braking and heavy braking characteristics |
| CN115158274B (en) * | 2022-08-31 | 2022-11-29 | 四川省公路规划勘察设计研究院有限公司 | Long and large longitudinal slope dangerous road section identification method based on truck braking and heavy braking characteristics |
| CN115320626A (en) * | 2022-10-11 | 2022-11-11 | 四川省公路规划勘察设计研究院有限公司 | Danger perception capability prediction method and device based on human-vehicle state and electronic equipment |
| CN115331449A (en) * | 2022-10-17 | 2022-11-11 | 四川省公路规划勘察设计研究院有限公司 | Method and device for identifying accident prone area of long and large continuous longitudinal slope section and electronic equipment |
| CN115331449B (en) * | 2022-10-17 | 2023-02-07 | 四川省公路规划勘察设计研究院有限公司 | Method and device for identifying accident-prone area of long and large continuous longitudinal slope section and electronic equipment |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114239423A (en) | Method for constructing prediction model of danger perception capability of driver on long and large continuous longitudinal slope section | |
| CN109733390B (en) | Self-adaptive lane change early warning method based on driver characteristics | |
| CN109597317B (en) | Self-learning-based vehicle automatic driving method and system and electronic equipment | |
| US7283056B2 (en) | Method and computer program for identification of inattentiveness by the driver of a vehicle | |
| EP1997705B1 (en) | Drive behavior estimating device, drive supporting device, vehicle evaluating system, driver model making device, and drive behavior judging device | |
| DE102016217538B4 (en) | Vehicle attention control system | |
| CN109766930A (en) | A kind of method for predicting residual useful life of the mine machinery equipment based on DCNN model | |
| US20090040054A1 (en) | Real-time driving danger level prediction | |
| CN108876165B (en) | Driver safety monitoring learning system | |
| US11040720B2 (en) | Sleepiness level prediction device and sleepiness level prediction method | |
| JP2000074791A (en) | Method and apparatus for analysis of traveling characteristic of automotive vehicle, and method and apparatus exerting influence to traveling characteristic | |
| Heine et al. | Algorithm for driver intention detection with fuzzy logic and edit distance | |
| Jin et al. | Study on the impact degrees of several driving behaviors when driving while performing secondary tasks | |
| CN111027618B (en) | Automobile dynamic property and economic expectation quantification method | |
| CN112863245B (en) | Vehicle track change real-time prediction method based on deep neural network | |
| CN113761715A (en) | Method for establishing personalized vehicle following model based on Gaussian mixture and hidden Markov | |
| CN112396118A (en) | GM-HMM-based driver acceleration intention modeling method | |
| CN114139460A (en) | Method and system for constructing and diagnosing driving comfort evaluation model of continuous downhill road section | |
| CN113460074B (en) | Automatic driving manual takeover request time adjustment method and system | |
| CN114103988B (en) | Safety monitoring device, vehicle comprising same, and corresponding method, device and medium | |
| CN114148349A (en) | Vehicle personalized following control method based on generation countermeasure simulation learning | |
| Hayashi et al. | Prediction of stopping maneuver considering driver's state | |
| CN118035691B (en) | Quantitative evaluation method, equipment and storage medium for intelligent driving perception quality | |
| JP7509656B2 (en) | Apparatus and program for determining proper driving condition | |
| CN115447616B (en) | Method and device for generating objective index of vehicle driving |
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 | ||
| RJ01 | Rejection of invention patent application after publication |
Application publication date: 20220325 |
|
| RJ01 | Rejection of invention patent application after publication |