CN113449790A - Mountain trunk highway high-risk road section identification method based on SVM - Google Patents

Abstract

The invention discloses a mountainous area trunk highway high-risk road section identification method based on SVM, which comprises the following steps: s1, constructing an SVM model based on road alignment parameters; s2, acquiring road alignment parameters of the mountain trunk road, and inputting the road alignment parameters into an SVM (support vector machine) model based on the road alignment parameters to obtain an identification result of the high-risk road section of the mountain trunk road; s3, constructing an SVM model based on psychological parameters of a driver; and S4, acquiring the psychological parameters of the driver when the driver drives on the mountain trunk road, and inputting the psychological parameters into an SVM (support vector machine) model based on the psychological parameters of the driver to obtain the identification result of the high-risk road section of the mountain trunk road. The mountainous area trunk road high-risk road section identification method based on the SVM can effectively identify the high-risk road sections existing in the mountainous area trunk road, and is high in reliability and accuracy.

Description

Identification method of high-risk section of mountain trunk highway based on SVM

technical field

The invention relates to the field of mountain trunk highways, in particular to a method for identifying high-risk sections of mountain trunk highways based on SVM.

Background technique

Due to the special geological conditions in mountainous areas, the traffic safety problems of trunk highways in mountainous areas are becoming more and more serious. The alignment conditions of trunk highways in mountainous areas are poor, the operation of mixed traffic flow is more complicated, and the roads are mostly water and cliffs. It brings great challenges. Compared with other high-grade roads, mountain trunk roads are more prone to traffic accidents.

At present, the identification of high-risk sections of trunk highways in mountainous areas in China mainly focuses on the analysis of accident data and the risk assessment of transportation infrastructure, etc., and does not fully consider the alignment conditions of trunk highways in mountainous areas and the psychological conditions of drivers to effectively identify High-risk sections of mountain arterial highways.

SUMMARY OF THE INVENTION

In view of this, the object of the present invention is to overcome the defects in the prior art, and provide a method for identifying high-risk sections of mountain trunk highways based on SVM, which can effectively identify the high-risk sections existing in mountain trunk highways, with strong reliability and high accuracy.

The method for identifying high-risk sections of trunk highways in mountainous areas based on SVM of the present invention comprises the following steps:

S1. Build an SVM model based on road alignment parameters;

S2. collect the road alignment parameters of trunk highways in mountainous areas, and input the road alignment parameters into the SVM model based on road alignment parameters, obtain the identification result of high-risk sections of trunk highways in mountainous areas;

S3. Construct an SVM model based on the driver's physiological parameters;

S4. collect the physiological parameters of the driver when driving on the mountain trunk road, and input the physiological parameters into the SVM model based on the driver's physiological parameters to obtain the identification result of the high-risk section of the mountain trunk highway.

Further, the step S1 specifically includes:

S11. collect the road alignment parameters of the trunk road in the test mountain area, and use the road alignment parameters as characteristic parameters; the road alignment parameters include road curve radius, road curve-chord ratio and road longitudinal slope;

S12. Establish a feature vector Data _r based on road alignment parameters according to the feature parameters; the Data _r ={x ₁ (i), x ₂ (i)... x _m (i)}; wherein, x _j (i) is the m-th feature parameter on the i-th road segment;

S13. Determine the label quantity based on road alignment parameters;

S14. Use the feature vector and the label amount as samples, use α% samples as training samples, and use β% samples as test samples;

S15. Determine a kernel function, and generate an SVM model based on road alignment parameters.

Further, the label quantity Label _r based on the road alignment parameters is determined according to the following formula:

Label _r = {y ₁ (i), y ₂ (i)};

Among them, y ₁ (i) and y ₂ (i) are both the safety conditions of the i-th road segment.

Further, in step S15, the kernel function adopts the Sigmoid kernel function.

Further, the step S3 specifically includes:

S31. collect the physiological and psychological parameters of the driver when driving on the trunk road in the test mountainous area, and use the described physiological and psychological parameters as characteristic parameters; the described physiological and psychological parameters include heart rate growth rate, SDNN and breathing frequency;

S32. according to the characteristic parameter, establish the characteristic vector Data _p based on the driver's physiological parameter; Described Data _p ={a ₁ (i), a ₂ (i)...a _k (i)}; Wherein, a _j ( i) is the k-th characteristic parameter when the driver is driving on the i-th road segment;

S33. Determine the label quantity based on the driver's physiological parameters;

S34. Use the feature vector and the label amount as a sample, use a sample of λ% as a training sample, and use a sample of μ% as a test sample;

S35. Determine the kernel function, and generate an SVM model based on the driver's physiological parameters.

Further, the label quantity Label _p based on the driver's physiological parameters is determined according to the following formula:

Label _p = {b ₁ (i), b ₂ (i)};

Among them, b ₁ (i) and b ₂ (i) are both the safety conditions of the driver when driving on the i-th road segment.

Further, in step S35, the kernel function adopts the RBF kernel function.

则为安全，若驾驶员的风险感知能力小于阈值

则为高危；Further, the safety conditions include safety and high risk; if the driver's risk perception ability is greater than the threshold

is safe, if the driver's risk perception ability is less than the threshold

high risk;

The driver's risk perception ability is:

Among them, U _ij is the driver's risk perception ability; F _ij is the driver's subjective risk degree; f _ij is the objective risk degree of the road; i is the road section number; j is the driver's number.

Further, the subjective risk degree F _ij of the driver is represented by the operating speed gradient; the operating speed gradient is:

Among them, ΔI _v is the running speed gradient; ΔV is the difference between the _running speed at the starting point and the end running speed of the road segment unit; L is the length of the road segment unit.

The beneficial effects of the present invention are as follows: a method for identifying high-risk sections of trunk highways in mountainous areas based on SVM disclosed in the present invention constructs an SVM model based on road alignment parameters by collecting the road alignment parameters of trunk trunk highways in mountainous areas, and collects drivers driving on trunk roads in mountainous areas. The SVM model based on the physiological and psychological parameters of the driver is constructed based on the physiological and psychological parameters of the road, and the SVM model corresponding to the parameters is used according to the collected different parameters to effectively identify whether the section of the mountain trunk highway is safe or high-risk. The identification method of the present invention is reliable. Strong and accurate.

Description of drawings

Below in conjunction with accompanying drawing and embodiment, the present invention is further described:

FIG. 1 is a schematic flow chart of the method of the present invention.

Detailed ways

The present invention is further described below in conjunction with the accompanying drawings of the description, as shown in the figure:

The method for identifying high-risk sections of trunk highways in mountainous areas based on SVM of the present invention comprises the following steps:

S1. Build an SVM model based on road alignment parameters;

S2. collect the road alignment parameters of trunk highways in mountainous areas, and input the road alignment parameters into the SVM model based on road alignment parameters, obtain the identification result of high-risk sections of trunk highways in mountainous areas;

S3. Construct an SVM model based on the driver's physiological parameters;

S4. collect the physiological parameters of the driver when driving on the mountain trunk road, and input the physiological parameters into the SVM model based on the driver's physiological parameters to obtain the identification result of the high-risk section of the mountain trunk highway. Wherein, the identification result of the high-risk section of the mountain trunk highway includes the safety of the mountain trunk highway section and the high risk of the mountain trunk highway section.

The SVM is the English abbreviation of Support Vector Machine, and the Chinese name is Support Vector Machine, also known as Support Vector Network. The SVM is a supervised learning model and associated learning algorithm that analyzes data in classification and regression analysis.

The invention constructs an SVM model based on road alignment parameters and drivers' physiological parameters, thereby realizing the identification of high-risk sections of trunk highways in mountainous areas, and providing technical support for the identification of high-risk sections with sharp curves and steep slopes in the design stage and operation management stage of trunk highways in mountainous areas.

In this embodiment, the step S1 specifically includes:

S11. Collect the road alignment parameters of the trunk road in the test mountainous area, and use the road alignment parameters as characteristic parameters; the road alignment parameters include the road curve radius, the road curve-chord ratio and the road longitudinal slope; Wherein, the sharp curve and steep slope of the mountain trunk road The road section mainly involves four scenarios: straight line, single curve, reverse circular curve in the same direction, and multi-curve combination. Considering that the line-of-sight section and radius ratio cannot be quantified, these two indicators are excluded, and the road curve radius and road curve string are finally selected. ratio and road longitudinal slope as road alignment parameters;

S12. Establish a feature vector Data _r based on road alignment parameters according to the feature parameters; the Data _r ={x ₁ (i), x ₂ (i)... x _m (i)}; wherein, x _j (i) is the m-th feature parameter on the i-th road segment;

S13. Determine the label quantity based on road alignment parameters;

S14. Take the feature vector and the label amount as samples, take α% samples as training samples, and take β% samples as test samples; the value of α is 70, and the value of β is 30;

S15. Determine a kernel function, and generate an SVM model based on road alignment parameters. Among them, MATLAB software is used to generate the SVM model based on road alignment parameters.

In this embodiment, the label quantity Label _r based on the road alignment parameter is determined according to the following formula:

Label _r = {y ₁ (i), y ₂ (i)};

Among them, y ₁ (i) and y ₂ (i) are both the safety conditions of the i-th road segment. Wherein, the y ₁ (i) and y ₂ (i) are represented by numbers, such as 1 or 2; 1 means safe, and 2 means high risk.

In this embodiment, in step S15, described kernel function adopts Sigmoid kernel function; Compared with RBF kernel function and polynomial kernel function, described Sigmoid kernel function is higher to the identification accuracy of high-risk road sections under different characteristic parameter combinations.

In this embodiment, the step S3 specifically includes:

S31. collect the physiological parameters of the driver when driving on the trunk road in the test mountainous area, and use the physiological parameters as characteristic parameters; the physiological parameters include heart rate growth rate, SDNN and breathing frequency; wherein, the SDNN refers to The overall standard deviation of the normal sinus R-R interval in the whole process; the changes of the driver's physiological indicators can effectively represent the driver's response to the external environmental risk; when the objective road risk increases, the driver's heart rate increases and breathing increases, in order to avoid accidents occurs, the driver responds with deceleration braking or steering avoidance behavior. Therefore, many indicators such as heart rate, heart rate variability, respiration amplitude, and respiration frequency can be selected as the characteristic parameters of the input layer of the SVM model, but too many characteristic parameters are selected. It will increase the training difficulty of the SVM model and reduce the identification accuracy of the SVM model. Based on the idea of dimension reduction by principal component analysis, the heart rate growth rate, SDNN and respiratory frequency are finally selected as the characteristic parameters of the input layer of the SVM model;

S32. according to the characteristic parameter, establish the characteristic vector Data _p based on the driver's physiological parameter; Described Data _p ={a ₁ (i), a ₂ (i)...a _k (i)}; Wherein, a _j ( i) is the k-th characteristic parameter when the driver is driving on the i-th road segment;

S33. Determine the label quantity based on the driver's physiological parameters;

S34. Take the feature vector and the label amount as a sample, take a sample of λ% as a training sample, and take a sample of μ% as a test sample; the value of λ is 70, and the value of μ is 30;

S35. Determine the kernel function, and generate an SVM model based on the driver's physiological parameters. Among them, the MATLAB software is used to generate the SVM model based on the physiological parameters of the driver.

In the present embodiment, the label amount Label _p based on the driver's physiological parameter is determined according to the following formula:

Label _p = {b ₁ (i), b ₂ (i)};

Among them, b ₁ (i) and b ₂ (i) are both the safety conditions of the driver when driving on the i-th road segment. Wherein, the b ₁ (i) and b ₂ (i) are represented by numbers, such as 1 or 2; 1 means safe, and 2 means high risk.

In this embodiment, in step S35, described kernel function adopts RBF kernel function; Compared with Sigmoid kernel function and polynomial kernel function, described RBF kernel function is higher to the identification accuracy of high-risk road sections under different characteristic parameter combinations.

则为安全，若驾驶员的风险感知能力小于阈值

则为高危；其 中，所述阈值

取值为1；In this embodiment, the safety conditions include safety and high risk; if the driver's risk perception ability is greater than the threshold

is safe, if the driver's risk perception ability is less than the threshold

high risk; where the threshold

The value is 1;

The driver's risk perception ability is:

Among them, U _ij is the driver's risk perception ability; F _ij is the driver's subjective risk degree, using a 10-point scale, the higher the score, the higher the driver's subjective risk perception value, the more dangerous; f _ij is the road The objective risk degree is quantified according to the speed gradient, and the objective risk degree of the road is normalized to the (0,10) score interval. The higher the score is, the higher the objective risk value of the road section is; i is the road section number; j is the Driver number.

If U _ij is greater than 1, it means that the driver's subjective risk is greater than the objective risk of the road, indicating that the driver is fully aware of the objective risks existing in the road section during driving, and the road section is a safe road section;

If U _ij is less than 1, it means that the driver’s subjective risk degree is less than the objective risk degree of the road, indicating that the driver does not fully recognize the risks existing in the road section, and there is the possibility of inducing traffic accidents due to ignoring the risks. The road section is a high risk section.

In this embodiment, the subjective risk F _ij of the driver is represented by the operating speed gradient; the operating speed gradient is:

Among them, ΔI _v is the running speed gradient; ΔV is the difference between the _running speed at the starting point and the end running speed of the road segment unit; L is the length of the road segment unit.

Taking the running speed gradient as the speed change in a unit section to measure the safety of the road section, it avoids the defect of a single index of running speed coordination evaluation, and helps to find out the road sections with unfavorable traffic safety more accurately. According to the evaluation experience of the coordination of the running speed of the adjacent road sections, when the running speed of the adjacent road section is acceleration, it generally has little impact on safety; and when the running speed of the adjacent road section is deceleration and the deceleration range is large within a short distance, it is generally considered that Excessive deceleration will affect safety.

Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be Modifications or equivalent replacements without departing from the spirit and scope of the technical solutions of the present invention should be included in the scope of the claims of the present invention.

Claims (9)

1. A mountainous area trunk road high-risk road section identification method based on SVM is characterized by comprising the following steps: the method comprises the following steps:

s1, constructing an SVM model based on road alignment parameters;

s2, acquiring road alignment parameters of the mountain trunk road, and inputting the road alignment parameters into an SVM (support vector machine) model based on the road alignment parameters to obtain an identification result of the high-risk road section of the mountain trunk road;

s3, constructing an SVM model based on psychological parameters of a driver;

and S4, acquiring the psychological parameters of the driver when the driver drives on the mountain trunk road, and inputting the psychological parameters into an SVM (support vector machine) model based on the psychological parameters of the driver to obtain the identification result of the high-risk road section of the mountain trunk road.

2. The SVM-based method for identifying the high-risk road section of the mountain trunk road according to claim 1, wherein: the step S1 specifically includes:

s11, acquiring road alignment parameters of a trunk road in a test mountain area, and taking the road alignment parameters as characteristic parameters; the road linear parameters comprise road curve radius, road curve chord ratio and road longitudinal slope;

s12, establishing characteristic vector Data based on road alignment parameters according to the characteristic parameters_r(ii) a The Data_r＝{x₁(i),x₂(i)…x_m(i) }; wherein x is_j(i) The m characteristic parameter on the ith road section is obtained;

s13, determining the label quantity based on the road alignment parameters;

s14, taking the feature vectors and the label quantity as samples, taking alpha% samples as training samples, and taking beta% samples as test samples;

and S15, determining a kernel function, and generating an SVM model based on the road alignment parameters.

3. The SVM-based method for identifying the high-risk road section of the mountain trunk road as claimed in claim 2, wherein: determining a Label quantity Label based on road alignment parameters according to the following formula_r：

Label_r＝{y₁(i),y₂(i)}；

Wherein, y₁(i) And y₂(i) All are safety conditions of the ith road section.

4. The SVM-based method for identifying the high-risk road section of the mountain trunk road as claimed in claim 2, wherein: in step S15, the kernel function is a Sigmoid kernel function.

5. The SVM-based method for identifying the high-risk road section of the mountain trunk road according to claim 1, wherein: the step S3 specifically includes:

s31, acquiring psychological parameters generated when a driver drives on a highway of a trunk in a test mountain area, and taking the psychological parameters generated as characteristic parameters; the psychogenic parameters comprise heart rate increase rate, SDNN and breathing frequency;

s32, establishing a feature vector Data based on psychological parameters of a driver according to the feature parameters_p(ii) a The Data_p＝{a₁(i),a₂(i)…a_k(i) }; wherein, a_j(i) The kth characteristic parameter when the driver drives on the ith road section is obtained;

s33, determining the label quantity based on the psychological parameters of the driver;

s34, taking the feature vector and the label quantity as samples, taking a lambda% sample as a training sample, and taking a mu% sample as a test sample;

and S35, determining a kernel function, and generating the SVM model based on the psychological parameters of the driver.

6. According to claimThe SVM-based mountain trunk highway high-risk road section identification method is characterized in that: determining Label quantity Label based on psychological parameters of driver according to the following formula_p：

Label_p＝{b₁(i),b₂(i)}；

Wherein, b₁(i) And b₂(i) Are all safe conditions when the driver is driving on the ith road segment.

7. The SVM-based method for identifying the high-risk road section of the mountain trunk road as claimed in claim 5, wherein: in step S35, the kernel function is an RBF kernel function.

8. The SVM-based method for identifying high-risk road segments of mountain trunk roads according to claim 3 or 6, wherein: the safety conditions include safety and high risk; if the risk perception capability of the driver is larger than the threshold value

Then for safety, if the risk perception capability of the driver is less than a threshold value

It is a high risk;

the risk perception capability of the driver is as follows:

wherein, U_ijRisk perception for the driver; f_ijSubjective risk for the driver; f. of_ijIs the objective risk of the road; i is a road section number; j is the driver number.

9. The SVM-based method for identifying the high-risk road section of the mountain trunk road as claimed in claim 8, wherein: the subjective risk degree F of the driver_ijThe running speed gradient is adopted for representation; the running speed gradient is as follows:

wherein, Delta I_vIs the running speed gradient; delta V_FortuneThe difference value of the starting point running speed and the end point running speed of the road section unit is obtained; l-section unit length.

Images