CN116631221B - Monte Carlo simulation-based on-road vehicle running risk quantitative calculation method - Google Patents

Abstract

The invention discloses a method for quantitatively calculating the running risk of an on-road vehicle based on Monte Carlo simulation, which comprises the following steps: step 1) defining an emergency event which happens according to probability according to specific working conditions; step 2): providing reasonable assumption and describing the motion process of the vehicle in the collision process, establishing a kinematic equation and deducing the collision judgment condition; step 3) counting the distribution of variables involved in the motion process of the host vehicle and the traffic participant under each working condition from unmanned aerial vehicle or driving simulation data; step 4): monte Carlo simulation was used to calculate the incidence of collision events. The risk measurement index based on the risk perception of the driver and the perception characteristic of the human ear to the sound is defined in the calculation method, and compared with the existing risk measurement index, the index provided by the invention is more in line with the visual perception of the driver to the risk.

Description

A quantitative calculation method for vehicle operation risks in transit based on Monte Carlo simulation

Technical field

The invention belongs to the field of transportation engineering, and in particular relates to a quantitative calculation method for vehicle operation risks in transit based on Monte Carlo simulation.

Background technique

The "National Road Traffic Safety Science and Technology Action Plan" issued by the Ministry of Science and Technology, the Ministry of Public Security and the Ministry of Transport in 2008 encourages the research and application of road traffic safety technology to achieve a year-by-year decrease in the number of deaths in road traffic accidents and to further reduce major road traffic accidents. , the death rate per 10,000 vehicles is close to the level of moderately developed countries. However, since 2011, the number of road traffic fatalities has begun to show stable fluctuations, while the rate of decline in the death rate per 10,000 vehicles has begun to slow down, and there is still a gap with moderately developed countries. These phenomena indicate that the protective capabilities of existing prevention and control measures have reached their limits, and new methods need to be explored to further improve the level of safety protection. As an important part of the national strategy of building a transportation power, traffic safety has been raised to an unprecedented level.

The shift in road traffic safety research from accident prevention to risk prevention and control is an important direction currently being explored and has gradually become a research hotspot. The definition of risk in "Cihai" is: the possibility that people encounter natural disasters, accidents and other unforeseen events that can lead to personal casualties, property damage and other economic losses in production, construction and daily life. The International Organization for Standardization (ISO) defines risk as “the impact of uncertainty on objectives.” Therefore, road traffic operation risk can be attributed to the possibility of a driver encountering a traffic accident in daily driving activities, which is affected by random variations in road conditions, traffic participant behavior, vehicle conditions, and meteorological environment.

Relevant research on road traffic operation risks can be roughly divided into two aspects: qualitative and quantitative. The former focuses on risk source analysis, accident formation mechanisms, etc., while the latter focuses on accident rates (including severity) and risk assessment. Obviously, quantitative research is more helpful for us to analyze the dominant factors in the formation of operational risks and propose effective prevention and control strategies. However, in terms of quantitative analysis, each indicator is not in the same dimension, and the comparability between each other is poor, making it difficult to form a consensus. One of the important reasons is the lack of basic theories and methods that can quantitatively analyze and characterize the process from random variation of traffic behavior to the formation of operational risks.

Contents of the invention

In response to the above problems, the present invention proposes a quantitative calculation method for vehicle risks in transit based on probability generalization and collision detection. Based on the risk formation mechanism, the road traffic environment and the uncertainty of the behavioral subject's judgment and control are taken into account. Through the driver's behavior, Characteristics are used to predict unique behavioral changes and potential collision probabilities. Based on the similarity between the driver's risk perception and the human ear's perception of sound, a risk measurement index that is consistent with people's intuitive perception is defined in the form of decibels.

The first aspect of the present invention proposes a quantitative calculation method for vehicle risks in transit, which is characterized by obtaining the probability distribution of speed, relative distance and other indicators under each working condition based on driving behavior data, and using Monte Carlo based on probability generalization and collision detection. The simulation calculates the occurrence rate of collision events. The method includes the following steps:

Step 1) Define emergencies that occur with probability according to specific working conditions;

Step 2) Put forward reasonable assumptions and describe the movement process of the vehicle during the conflict, establish kinematic equations and derive the collision determination conditions;

Step 3) Statistics of the distribution of variables involved in the movement of the vehicle and traffic participants under each working condition from UAV or driving simulation data;

Step 4) Use Monte Carlo simulation to calculate the occurrence rate of collision events.

The second aspect of the present invention proposes a risk measurement index based on the driver's risk perception and the human ear's perception characteristics of sound. It is characterized in that the risk measurement index is defined based on the human ear's perception characteristics of sound, which is more in line with the driver's perception of risk. Intuitive perception, the calculation formula of risk measurement index is defined as:

In the formula:

R _individual ——single sample risk perception index, unit dB;

P _individual - individual collision probability calculated using the initial state of a single sample;

P _base ——Basic risk, the average collision probability calculated using the sampling values of the initial state distribution of the full sample under typical working conditions.

The physical meaning of R _individual is: based on the initial state of the current sample, the probability of collision is the basic risk. times.

The advantages of the present invention are as follows:

1. Based on the probabilistic characteristics of risk, the driver's risk perception is quantified as an assessment of the possibility of a collision with the vehicle, and a calculation method for vehicle risk quantification in transit is proposed based on probability generalization and collision detection.

2. Based on the driver's risk perception and the sound perception characteristics of the human ear, a measurement index that is consistent with people's intuitive perception is defined.

Description of the drawings

Figure 1 is an overall flow chart of the quantitative calculation method of vehicle operation risk on the road based on Monte Carlo simulation provided by the present invention;

Figure 2 is a schematic diagram of the lane changing condition of multiple lanes in the same direction in the embodiment of the present invention;

Figure 3 is a distribution diagram of deceleration time of obvious braking samples in the embodiment of the present invention;

Figure 4 is a probability mass function diagram of the distance between the front of the vehicle at the starting point of the lane change and the rear of the vehicle in front of the current lane in an embodiment of the present invention;

Figure 5 is a probability mass function diagram of the emergency braking deceleration of the leading vehicle in the embodiment of the present invention;

Figure 6 is a probability mass function diagram of the deceleration time of the leading vehicle in the embodiment of the present invention.

Detailed ways

In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described here are only used to explain the present invention and do not limit the scope of the present invention. The specific implementation is as follows:

During the lane-changing behavior of a multi-lane highway in the same direction, the collision risk mainly comes from the collision risk between the vehicle and the vehicle in front of the current lane and the collision risk between the vehicle and the vehicle behind the target lane, as shown in Figure 2. The following uses the risk quantification model calculation of the vehicle and the vehicle in front of the current lane during the lane-changing behavior of a multi-lane highway in the same direction as an example to illustrate:

1) Definition of emergencies

The hypothesis of an emergency between this vehicle and the vehicle in front of the current lane is: before and after changing lanes, the vehicle in front suddenly brakes, causing the vehicle to be unable to avoid a collision.

2) Put forward reasonable assumptions and describe the movement process of the vehicle during the conflict, establish kinematic equations and derive the collision determination conditions

When the vehicle acceleration a<-1.5m·s ^-2 is regarded as a sudden braking, the deceleration process of the motor vehicle is simplified as a uniform deceleration motion. The deceleration can be the average value of a during the deceleration process. The simplified calculation method is:

_{amean≈0.5amax _}

In the formula,

a _mean - the average deceleration of the vehicle during deceleration, unit m·s ^-2 ;

a _max ——constant, which is the maximum value of deceleration during this braking process, unit m·s ^-2 .

According to the principle of kinematics, the distance traveled by the vehicle in front during the process of decelerating to the lowest point of speed can be calculated by the following formula:

S(t)＝v _f,0 t _dec +0.25a _max t _dec ²

In the formula,

v _f,0 - the initial speed of the vehicle in front before decelerating, unit m·s ^-1 ;

t _dec - deceleration time of the vehicle in front, unit s.

The meanings of other symbols are the same as before.

When the vehicle in front brakes suddenly, the vehicle's deceleration process can be divided into two stages. The first stage is the driver's reaction stage, and the vehicle's motion state will not change much during this stage; the second stage is the braking stage. During the driving phase, the driver of the vehicle will apply the brakes to reduce the speed of the vehicle. In the "Highway Route Design Specification", the driver's designed reaction time in the parking sight distance formula is 2.5 seconds. When driving on the highway, the driver is often very vigilant, and the actual reaction time will be lower than this value. This invention uses Scaner to simulate The device tests the driver's reaction time, and the measured reaction time is t _r =0.6s.

Calculate and count the deceleration duration of obvious braking samples in the UAV data. The results are shown in Figure 3. Design a computer program starting from the lane ID change point, and search forward and backward respectively for the extreme point of the trajectory coordinates or the minimum curvature point as the starting and end point of the lane change. When the program cannot search for the start and end points, use dynamic chart tools to manually determine the start and end points. Since the vehicle's body crosses the lane edge and enters the target lane, the risk of collision with the vehicle in front of it dissipates immediately. In the statistical drone data set, the time from the starting point of the lane change to the moment when the rear wheel presses the line on the opposite side of the vehicle's lane change accounts for the total lane change time. proportion, the sample mean is 0.65. Assuming that the minimum deceleration value and acceleration change form of the vehicle are consistent with the vehicle in front, the motion equation of the vehicle is:

S _B,f =v _B,f (0.65t _h )+0.25a _max (0.65t _h -t _r ) ²

In the formula,

S _B,f - is the distance traveled by the vehicle from the time when the vehicle in front decelerates to when it enters the opposite lane, unit m;

v _B,f - the initial speed of the vehicle before deceleration, that is, the speed at the starting point of the lane change, unit m·s ^-1 ;

t _h ——The total lane changing time of the vehicle, unit s.

The meanings of other symbols are the same as before.

The deceleration time is roughly equivalent to the length of t _r +0.65t _h . Therefore, it can be assumed that when the vehicle body crosses the line, the speed of the vehicle in front just reaches the lowest point. Then the equation of motion of the vehicle in front correspondingly becomes:

S _f =v _f (0.65t _h )+0.25a _max (0.65t _h ) ²

In the formula,

S _f - is the distance traveled by the vehicle in front from the time it decelerates until the vehicle enters the opposite lane, unit m;

v _f ——The speed of the vehicle ahead when the vehicle is at the starting point of lane change, unit m·s ^-1 .

The meanings of other symbols are the same as before.

Finally, the conditions for determining a collision between two cars can be obtained:

D _f +S _f -S _B,f ≤0

3) Statistics of the distribution of variables involved in the movement of the vehicle and traffic participants under various working conditions from UAV or driving simulation data

①.Initial state variables of the vehicle

The initial state variable of the vehicle is only the distance D _f between the vehicle and the rear of the preceding vehicle at the starting point of the lane change. Statistics of the lane change starting point D _f of the vehicle group behind the target lane in the UAV data. Taking 10m as the step size, calculate the probability mass of each segment, and the results are shown in Figure 4. When calculating the basic risk, D _f is sampled from this function. Assuming that the samples in each paragraph are equally likely to be drawn, a uniform distribution with a step size of 0.5 times needs to be added to the sampled value.

D _f,sample =D _f,draw +ε _Df

ε _Df ~Uniform(-5,5)

In the formula,

D _f,sample - the final sample value of D _f , unit m;

D _f,draw ——D _f value extracted from the probability mass function, unit m;

ε _Df - a random item obeying a uniform distribution, unit m;

Uniform(-5,5) - a uniform distribution with a lower bound of -5 and an upper bound of 5.

②.Initial state variables of the interactive vehicle

The initial state variables of the interactive vehicle include the maximum deceleration a _max and deceleration time t _dec of the preceding vehicle in this lane. When calculating basic risk, the above two variables should be sampled from the following 2 probability mass functions.

A. Probability mass function of maximum deceleration a _max

Select the acceleration of the entire sample trajectory of the UAV data and extract the maximum deceleration value a _max of each trajectory. Taking 1m·s ^-2 as the step size, the probability of a _max in each segment is calculated, and its distribution is shown in Table 1 and Figure 5.

Table 1 Values of the probability mass function of the vehicle in front of the vehicle braking suddenly

paragraph Probability Paragraph center value/(m·s ^-2 ) 1 0.987176 -0.5 2 0.010779 -2 3 0.001617 -3 4 0.000344 -4 5 6.50E-05 -5 6 9.29E-06 -6 7 9.29E-06 -7

Since the center of the paragraph is discrete, in order to increase the randomness of the sample, assuming equal probability sampling within each paragraph, and adding a uniform distribution of 0.5 times the step size to each segment, the sample of a _max is determined according to the following formula:

a _{max, sample} = a _{max, draw} +ε _am

ε _am ~Uniform(-0.5,0.5)

In the formula,

a _{max, sample} ——a _max final sample value, unit m·s ^-2 ;

a _{max, draw} ——a _min value extracted from the probability mass function, unit m·s ^-2 ;

ε _am ——Random item obeying uniform distribution, unit m·s ^-2 ;

Uniform(-0.5,0.5) - a uniform distribution with a lower bound of -0.5 and an upper bound of 0.5.

B. Probability mass function of deceleration time t _dec

Count the deceleration duration of each of the above deceleration sample trajectories. Statistical analysis confirms that there is no obvious linear relationship between t _dec and a _max , so they can be sampled independently. The deceleration time sample is divided into segments with a step size of 0.5s, and the probability mass of each segment is calculated. The results are shown in Figure 6. When sampling, assuming that samples from each statistical paragraph are equally likely to be drawn, add a uniform distribution of 0.5 times the step size to the sampled values, that is:

t _{dec, sample} = t _{dec, draw} +ε _td

ε _td ~Uniform(-0.25,0.25)

t _{dec, sample} ——t _dec final sample value, unit s;

t _{dec, draw} - t _dec value extracted from the probability mass function, unit s;

ε _td ——Random item obeying uniform distribution, unit s;

Uniform(-0.25,0.25) - a uniform distribution with a lower bound of -0.25 and an upper bound of 0.25.

③. Variables related to the initial state

Variables related to the initial state include the speed difference v _f -v _B,f between the vehicle and the preceding vehicle at the lane change starting point and the lane change time t _h .

A. Calculation method of the speed difference v _f -v _B,f between the vehicle in front and the vehicle in front

The front and rear vehicle speed difference (v _f -v _b ) has a linear relationship with D _f at the starting point of the lane change. Let v _d,f = v _f -v _B,f , establish a regression model with v _d,f as the dependent variable and D _f as the independent variable, and obtain the regression equation as shown in Table 2.

Table 2 Linear regression model of the speed difference between the vehicle in front and the vehicle in front

coefficient standard deviation t-statistic p value Constant term 0.746 0.229 3.265 0.001 The distance between this vehicle and the rear of the vehicle in front 0.072 0.004 19.740 <0.001

The standard deviation of the residual is 2.37. In order to restore the true distribution of v _{d, f} as much as possible, the residual term needs to be considered during sampling. Then the sampling value of v _{d, f} can be calculated with the following formula:

v _d,f,sample =0.072D _f,sample +0.746+ε _vdf

ε _vdf ~Normal(0,2.37)

In the formula,

v _d,f,sample - the final sampling value of _{v d,f} , unit m·s ^-1 ;

ε _vdf ——normal distribution residual term;

Normal(0,2.37) - Normal distribution with expectation 0 and standard deviation 2.37.

The meanings of other symbols are the same as before.

B. Calculation method of lane changing time t _h

Using D _f as the independent variable and the lane changing time t _h as the dependent variable, a simple regression analysis was conducted. The results are shown in Table 3.

Table 3 Linear regression equation of lane changing time

coefficient standard deviation t-statistic p value Constant term 4.985 0.679 7.224 <0.001 The distance between this vehicle and the rear of the vehicle in front 0.0185 0.008 2.39 0.021

The standard deviation of the residuals is 2.26. Similarly, in order to restore the true distribution of lane changing times as much as possible, the residuals need to be put into the equation for sampling. Then the sampling value of t _h can be calculated using the following formula:

t _h,sample =0.0185D _f,sample +4.985+ε _th

ε _th ~Normal(0,2.26)

t _h,sample - the final sampling value of t _h , unit s;

ε _th ——Normally distributed residual term;

Normal(0,2.26) - Normal distribution with expectation 0 and standard deviation 2.26.

The meanings of other symbols are the same as before.

4) Use Monte Carlo simulation to calculate the occurrence rate of collision events

①.Basic risks under typical working conditions

Monte Carlo simulation was used to compile a simulation program to calculate the basic risk under specific working conditions. The sampling times of the simulation were all 10 ⁸ times, and the basic risk value of the collision between the vehicle and the vehicle in front of the current lane was 1.5×10 ^-7 . The basic risk refers to the average level of collision accidents between vehicles and other traffic participants in typical working conditions under emergency conditions. The value represents the ratio of the number of samples that meet the collision conditions to the total number of samples in the simulation in 10 ⁸ sampling simulations.

②.Single sample risk based on driving behavior spectrum

In multi-lane lane changing conditions, the variables and value methods involved in the single-sample risk calculation of the vehicle before the collision are shown in the following table:

Table 4 Variables and value methods for single-sample risk calculation of multi-lane lane changing behavior

Taking the driving sample where the initial state D _f is 67m, the speed of the vehicle is 37.4m·s ^-1 , the speed of the vehicle ahead in the current lane is 22.2m·s ^-1 v _d,f = 15.2m·s ^-1 as an example, the number of sampling simulations Taking 10 ⁸ times, the collision probability is 1.06×10 ^-4 , and the basic risk of the vehicle in front of the collision is 1.5×10 ^-7 . Substituted into the risk measurement index formula, the single sample risk value is calculated to be 2.85dB.

The specific embodiments described above are only used to explain the present invention so that those skilled in the technical field can understand and apply the present invention, and are not intended to limit the present invention. Any modifications, improvements, etc. made by those skilled in the art to the above embodiments based on the technical essence of the present invention within the spirit and principles of the present invention shall be within the protection scope of the present invention.

Claims (3)

1. A quantitative calculation method for vehicle operation risks on the road based on Monte Carlo simulation, which is characterized by: including the following steps: Step 1) Define an emergency event that occurs with probability according to specific working conditions; Step 2) Put forward reasonable assumptions and describe the movement process of the vehicle during the conflict, establish kinematic equations and derive the collision determination conditions; Step 3) Statistics of the distribution of variables involved in the movement of the vehicle and traffic participants under various working conditions from UAV or driving simulation data; wherein, the variable distribution is used to determine the sampling of each variable in the model space Step 4) Use Monte Carlo simulation to calculate the occurrence rate of collision events; Among them, the calculation of the incidence rate draws on the human ear's perception characteristics of sound to define the risk measurement index, which is more in line with the driver's intuitive perception of risk. The calculation formula is: In the formula, R _individual is the risk perception index of a single sample, in dB; P _individual is the individual collision probability calculated using the initial state of a single sample; P _base is the basic risk, calculated using the sampling values of the initial state distribution of the full sample under typical working conditions. Average collision probability; the physical meaning of R _individual is: according to the initial state of the current sample, the probability of collision is the basic risk times. 2. A method for quantifying operation risks of vehicles in transit based on Monte Carlo simulation according to claim 1, characterized in that: the specific working conditions described in step 1) refer to the working conditions under specific traffic conflicts on specific road sections. condition. 3. A method for quantifying the operation risk of vehicles in transit based on Monte Carlo simulation according to claim 1, characterized in that: the kinematic equation described in step 2) is based on the vehicle avoiding other traffic participants in emergencies. The kinematic equation is constructed based on the motion characteristics of each traffic participant during the process.