A driving style online evaluation method

Abstract

The application discloses a kind of driving style online evaluation methods, comprising the following steps: extracting the driving behavior data of several straight or turning road sections;Based on the driving behavior data of straight or turning road section is segmented into several road section primitives based on bayesian condensation type sequence segmentation algorithm;Based on the latent Dirichlet allocation model of variable coupling, the road section primitive corresponding to each road section is clustered, and is divided into different categories;Based on the kinetic energy size of different category road section primitives after clustering, different categories are valued;Based on the category valuation of the road section primitive of any road section, the road section primitive average intensity value Q of any road section, road section primitive transfer diversity value D, road section primitive transfer tendency value T are obtained;Based on driving style comprehensive evaluation function, the driving style evaluation value J of road section is obtained;Based on driving style evaluation value J, the driving style of road section is obtained;Solve the problem that a large amount of manpower and material resources are consumed when traditional driving style is evaluated and calculation is complicated in research process.

Description

Technical Field

[0001] This invention relates to the technical field of autonomous driving, and more particularly to the field of online evaluation methods for driving styles. Background Technology

[0002] Driving style analysis is one of the core technologies for improving traffic safety, enhancing vehicle fuel economy, and improving the intelligence level of autonomous vehicles. Accurate and efficient online driving style analysis has always been a research hotspot.

[0003] Existing research uses labeled driving behavior data to construct style classifiers, and then labels the driving behavior data with high quality. All labeled driving behavior data is then processed online by the style classifier to confirm driving style. However, labeling a large amount of driving behavior data with high quality will consume a lot of human and material resources, and the fact that all the driving behavior data is included in the calculation adds practical difficulties to driving style research.

[0004] Therefore, how to avoid labeling all real-time driving behavior data and avoid using all labeled driving behavior data in calculations to obtain driving style has become an urgent problem to be solved in this field. Summary of the Invention

[0005] To address the aforementioned technical problems, this invention provides an online driving style evaluation method. The method divides driving behavior parameters of a road segment into several road segment primitives, classifies and assigns values ​​to these primitives, and then uses these values ​​to calculate the average intensity, primitive transition diversity, and primitive transition tendency of the road segment primitives. This constructs a comprehensive driving style evaluation function to evaluate driving style. This method solves the problem of the high manpower and material resources required to label all driving behavior data and input all labeled data into the calculations to obtain driving style, as well as the cumbersome calculations during the research process.

[0006] The present invention provides an online evaluation method for driving style, comprising the following steps: extracting driving behavior data from several straight or turning road segments;

[0007] The driving behavior data of straight or turning road segments is divided into several road segment primitives based on the Bayesian condensed sequence segmentation algorithm.

[0008] The latent Dirichlet assignment model based on variable coupling clusters the segment primitives corresponding to each segment and divides them into different categories.

[0009] Values ​​are assigned to different categories of road segment primitives based on the kinetic energy of the different categories after clustering;

[0010] Based on the category assignment of road segment primitives for any road segment, obtain the average intensity value Q, the transfer diversity value D, and the transfer tendency value T of road segment primitives for any road segment.

[0011] The driving style evaluation value J of the road segment is obtained based on the comprehensive driving style evaluation function, which is:

[0012] J = w1·Q + w2·D + w3·T; w1, w2, and w3 are the weighting coefficients corresponding to the average intensity value Q of road segment basic elements, the transfer diversity value D of road segment basic elements, and the transfer tendency value T of road segment basic elements;

[0013] Based on the driving style judgment method, the driving style evaluation value J of the road segment is classified and labeled to obtain the driving style of the road segment.

[0014] Compared with the prior art, the present invention has the following beneficial effects: by classifying the driving behavior data of road segments into several road segment primitives, and then classifying the road segment primitives; then assigning values ​​to different categories of road segment primitives according to the kinetic energy of different categories of road segment primitives, thereby realizing the assignment of values ​​to different road segment primitives, and performing subsequent calculations through the assignment of values ​​to different road segment primitives, it avoids the problems of consuming a lot of manpower and resources and the cumbersome calculations in the research process of marking each driving parameter in the road segment and using the marked driving parameters for calculation;

[0015] Simultaneously, by assigning different values ​​to road segment primitives of different categories based on their kinetic energy levels, the average intensity value Q, the transfer diversity value D, and the transfer tendency value T of any road segment primitives are obtained. The driving style evaluation value J is then calculated using the comprehensive driving style evaluation function J = w1·Q + w2·D + w3·T. In other words, driving style is evaluated by assessing the changes in road segment primitives representing different kinetic energy states over time. This allows for the calculation of the driving style evaluation value J using assigned values ​​instead of the original driving behavior data, and accurate evaluation of driving style based on J.

[0016] Furthermore, the driving behavior data for the straight or turning sections includes speed, longitudinal acceleration, lateral acceleration, and jerk.

[0017] The beneficial effect of adopting the above-mentioned further scheme is that the segmented road segment primitives can reflect the vehicle's operating status.

[0018] Furthermore, the road segment primitives corresponding to each road segment are clustered into five categories;

[0019] The specific method for assigning values ​​to different categories is as follows:

[0020] Calculate the average kinetic energy of the road segment primitives included in each category, and use the average kinetic energy as the corresponding kinetic energy evaluation parameter for each category;

[0021] Then, the kinetic energy evaluation parameters corresponding to each category are arranged in ascending order and defined as follows: low intensity category, lower intensity category, medium intensity category, higher intensity category, and high intensity category;

[0022] Then, the low-intensity, lower-intensity, medium-intensity, higher-intensity, and high-intensity categories are assigned values ​​in an arithmetic sequence manner, where the assigned value is the intensity of each road segment element.

[0023] The advantages of adopting the above-mentioned further scheme are: it enables the assignment of values ​​to different types of road segment primitives, and the assignment of values ​​to road segment primitives can reflect the trend of changes in the vehicle driving status corresponding to the road segment primitives; at the same time, it is convenient, simple and easy to operate to use this assignment for calculation.

[0024] Furthermore, the method for obtaining the average intensity value Q of road segment primitives includes the following steps:

[0025] Arrange the road segment primitives of each road segment in chronological order and assign them numbers 1-N respectively;

[0026] The average strength value Q of the road segment is calculated using the following formula:

[0027]

[0028] N is the number of road segment primitives contained in the road segment, pri_i k Let be the strength of the k-th segment primitive.

[0029] The beneficial effect of adopting the above-mentioned further scheme is that by obtaining the average intensity value Q of the road segment basic elements, the macroscopic kinetic energy status of the vehicles in the evaluated road segment can be obtained, which is an important factor in evaluating the driving style of the segment.

[0030] Furthermore, the segment primitive transfer diversity value D is calculated using the following formula:

[0031] D = H + S; N = count(P) gh ≠0);

[0032]

[0033] The p ghLet p be the segment primitive transition probability, which represents the probability that a segment primitive will transition from one category g to another category h. (Specifically, the numerator is the number of times a segment primitive is of category g at a certain moment and of category h at the next moment; the denominator is the number of times a segment primitive is of category g at the previous moment.) gh ); g and h both represent one of the road segment primitive categories;

[0034] Based on the segment primitive transition probability p gh This yields the segment primitive transition probability matrix for each segment, where K is the number of primitive categories.

[0035]

[0036] Based on the road segment primitive transition probability matrix, calculate the information content H of the road segment primitive transition probability matrix:

[0037]

[0038] Calculate the number S of primitive transition forms contained in the road segment.

[0039] S = count(P) ij ≠0).

[0040] The beneficial effect of adopting the above-mentioned further scheme is that by obtaining the road segment primitive transfer diversity value D, the probability of the occurrence of the vehicle macroscopic kinetic energy state represented by a certain type of road segment primitive transforming into the vehicle macroscopic state represented by another type of road segment primitive can be evaluated, which is an important factor in evaluating the driving style of that segment.

[0041] Furthermore, the segment element transfer tendency T is calculated using the following formula:

[0042]

[0043] M represents the number of segment primitive transitions in a segment, trans m It is the strength difference trans of the road segment primitive at the m-th transfer. m =pri_h-pri_g,w m It is the weight of this transfer form;

[0044]

[0045] pri_g represents the strength of the road segment primitive before the transfer at the m-th transfer, and pri_h represents the strength of the road segment primitive after the transfer at the m-th transfer.

[0046] The beneficial effect of adopting the above-mentioned further scheme is that by obtaining the road segment basic element transfer tendency T, it is possible to evaluate the number of times the macroscopic kinetic energy state of the vehicle represented by a certain type of road segment basic element changes to the macroscopic kinetic energy state of the vehicle represented by another type of road segment basic element, which is an important factor in evaluating the driving style of that segment.

[0047] Furthermore, w1 = w 1o ·δ1; w2=w 2o ·δ2;w3=w 3o ·δ3;

[0048] w 1o Let w1 be the initial weight, and w 2o The initial weights corresponding to w2, w 3o The initial weights corresponding to w3;

[0049] δ1 is the adjustment coefficient corresponding to w1, δ2 is the adjustment coefficient corresponding to w2, and δ3 is the adjustment coefficient corresponding to w3.

[0050] The beneficial effect of adopting the above-mentioned further scheme is that: the weight coefficient is obtained by multiplying the corresponding initial weight and the corresponding adjustment coefficient. That is, the initial weight coefficient reflects the macro-proportion of the average strength value Q of road segment basic elements, the transfer diversity value D of road segment basic elements, and the transfer tendency value T of road segment basic elements in the evaluation formula. At the same time, the adjustment coefficient reflects that the initial weight coefficient is adjusted according to the specific average strength value Q of road segment basic elements, the transfer diversity value D of road segment basic elements, and the transfer tendency value T of road segment basic elements when they are different, which better reflects the reality.

[0051] Furthermore, the method for obtaining δ1, δ2, and δ3 includes the following steps:

[0052] Select M straight road segments or M turning road segments, and calculate the segment primitive transfer diversity value D. m1 The average strength value Q of the corresponding basic elements of each road segment at that time m1 This yields the data pair (Dm1, Qm1);

[0053] The calculation of the segment basic element transfer tendency value is T. m1 The average strength value Q of the corresponding basic elements of each road segment at that time m2 , obtain data pairs (T) m1 Q m2 );

[0054] The calculated segment primitive transfer diversity value is D. m1 The average value T of the corresponding road segment primitive at that time m2 , obtain data pairs (D m1 ,T m2 );

[0055] The segment element transfer diversity value of the M straight road segments or M turning road segments is not equal to 0, and the segment element transfer tendency value is not equal to 0.

[0056] Based on the least squares method, data from M road segments are used to analyze (D) m1 Q m1 ), (T m1 Q m2 ) and (D m1 ,T m2 ), Analysis of Q m1 With D m1 T m1 With Q m2 D m1 With T m2 The linear relationship between them yields the regression equation:

[0057] Q m1 =k1·D m1 +c, Q m2 =k2·T m1 +c, T m2 =k3·D m1 +c.

[0058] Construct a judgment matrix A based on k1, k2, and k3. Judgment matrix A is...

[0059] Multiply the judgment matrix A by its rows to obtain a new vector B.

[0060] Taking the nth root (n=3) of each component of vector B yields vector C.

[0061] Adjusting vector C to a form where the sum is 1 yields the adjustment coefficient δ. i i = 1, 2, 3;

[0062]

[0063] The beneficial effect of adopting the above-mentioned further scheme is that, through the above scheme, the adjustment coefficient δ can be obtained. i , i = 1, 2, 3.

[0064] Furthermore, after constructing the judgment matrix A, a consistency analysis is first performed on the judgment matrix A;

[0065] The specific steps of consistency analysis are as follows:

[0066] Calculate the consistency index (CI). Where λ max To determine the largest eigenvalue of matrix A, r is the order of the matrix, r = 3;

[0067] Calculate the consistency ratio (CR). Where RI = 0.52;

[0068] When CR < 0.1, the consistency of judgment matrix A is considered acceptable, and the judgment matrix is ​​reasonable; otherwise, the consistency of judgment matrix A is not acceptable, and then judgment matrix A is adjusted.

[0069] The specific adjustment method includes the following steps:

[0070] Find the minimum value of the regression coefficient in the judgment matrix, k i =min(k1,k2,k3);

[0071] Let k be the smallest regression coefficient in the judgment matrix. i The corresponding element is 1, that is, {k} i ,1 / k i} = 1; then perform consistency analysis again, and repeat this process until the matrix consistency is verified.

[0072] The beneficial effect of adopting the above-mentioned further scheme is that, through the above consistency analysis method, it is easier to achieve the obtained adjustment coefficient δ. i The values ​​of i = 1, 2, 3 are closer to reality.

[0073] Furthermore, the driving style determination method includes the following steps:

[0074] The driving style evaluation value J calculated for straight or turning road segments is subtracted from the threshold scores of different types of driving styles for the corresponding straight or turning road segments to obtain the driving style difference values ​​for different categories.

[0075] After taking the absolute value of the differences between different driving style categories, the driving style category corresponding to the driving style difference with the smallest absolute value is obtained as the driving style of the road segment.

[0076] The beneficial effect of adopting the above-mentioned further scheme is that the driving style evaluation value J calculated for each road segment can be judged by the above method, thereby confirming the driving style of each road segment. Attached Figure Description

[0077] Figure 1 This is a dispersion plot showing the relationship between the diversity of road segment element transfers and the average intensity of road segment elements when the road is straight.

[0078] Figure 2 This is a dispersion plot showing the relationship between the tendency of road segment elements to shift and the average intensity of road segment elements in straight road sections.

[0079] Figure 3 A scatter plot showing the relationship between the diversity of segment element transfers and the tendency of segment element transfers in straight road segments;

[0080] Figure 4 This is a schematic diagram illustrating the regression results between the diversity of road segment element transfer and the average intensity of road segment elements in a straight road segment;

[0081] Figure 5 This is a schematic diagram illustrating the regression results between the tendency of road segment element transfer and the average intensity of road segment elements in straight road sections;

[0082] Figure 6 A schematic diagram illustrating the regression results between the diversity of road segment element transfer and the tendency of road segment element transfer in straight road sections;

[0083] Figure 7 A scatter plot showing the relationship between the diversity of road segment element transfers and the average intensity of road segment elements in a turning section;

[0084] Figure 8 This is a dispersion plot showing the relationship between the tendency of road segment elements to shift and the average strength of road segment elements in a turning section.

[0085] Figure 9 A scatter plot showing the relationship between the diversity of segment primitive transfers and the tendency of segment primitive transfers in turning road segments;

[0086] Figure 10 This is a schematic diagram illustrating the regression results between the diversity of road segment element transfers and the average intensity of road segment elements in a turning road segment;

[0087] Figure 11 This is a schematic diagram illustrating the regression results between the tendency of road segment element transfer and the average intensity of road segment elements in a turning section.

[0088] Figure 12 This is a schematic diagram illustrating the regression results between the diversity of road segment primitive transfers and the tendency of road segment primitive transfers in a turning road segment;

[0089] Figure 13 This is a distribution map of driving styles on straight-ahead road sections;

[0090] Figure 14 This is a distribution map of driving styles on turning sections of road. Detailed Implementation

[0091] To better understand the technical solution of the present invention, the present invention will be further described below with reference to specific embodiments.

[0092] Example 1:

[0093] This embodiment provides an online driving style evaluation method, including the following steps: extracting driving behavior data from several straight or turning road segments; the driving behavior data from the straight or turning road segments includes speed, longitudinal acceleration, lateral acceleration, and jerk.

[0094] The driving behavior data of straight or turning road segments is divided into several road segment primitives based on the Bayesian condensed sequence segmentation algorithm.

[0095] The latent Dirichlet assignment model based on variable coupling clusters the segment primitives corresponding to each segment and divides them into different categories.

[0096] Based on the kinetic energy of the road segment primitives of different categories after clustering, values ​​are assigned to different categories. Specifically, the road segment primitives corresponding to each road segment are clustered into five categories. The specific method for assigning values ​​to different categories is as follows: calculate the average kinetic energy of the road segment primitives included in each category, and use the average kinetic energy as the corresponding kinetic energy evaluation parameter for each category; then, arrange the kinetic energy evaluation parameters corresponding to each category in ascending order and define them as: low intensity category, lower intensity category, medium intensity category, higher intensity category, and high intensity category; then, assign values ​​to the low intensity category, lower intensity category, medium intensity category, higher intensity category, and high intensity category in an arithmetic progression manner, whereby the assigned value is the intensity of each road segment primitive.

[0097] Based on the category assignment of road segment primitives for any road segment, obtain the average intensity value Q, the transfer diversity value D, and the transfer tendency value T of road segment primitives for any road segment.

[0098] The method for obtaining the average intensity value Q of road segment basic elements includes the following steps:

[0099] Arrange the road segment primitives of each road segment in chronological order and assign them numbers 1-N respectively;

[0100] The average strength value Q of the road segment is calculated using the following formula:

[0101]

[0102] N is the number of road segment primitives contained in the road segment, pri_i k Let be the strength of the k-th segment primitive.

[0103] The segment primitive transfer diversity value D is calculated using the following formula:

[0104] D = H + S; N = count(P) gh ≠0);

[0105]

[0106] The p ghLet p be the segment primitive transition probability, which represents the probability that a segment primitive will transition from one category g to another category h. Specifically, the numerator is the number of times a segment primitive is of category g at a certain moment and of category h at the next moment; the denominator is the number of times a segment primitive is of category g at the previous moment. The ratio of the numerator to the denominator is the segment primitive transition probability p. gh Both g and h represent one of the road segment basic element categories, namely, low intensity, relatively low intensity, medium intensity, relatively high intensity, and high intensity.

[0107] Based on the segment primitive transition probability p gh This yields the segment primitive transition probability matrix for each segment, where K is the number of primitive categories.

[0108]

[0109] Based on the road segment primitive transition probability matrix, calculate the information content H of the road segment primitive transition probability matrix:

[0110]

[0111] Calculate the number S of primitive transition forms contained in the road segment, S = count(P) ij ≠0).

[0112] The segment element transfer tendency T is calculated using the following formula:

[0113] M represents the number of segment primitive transitions in a segment, trans m It is the strength difference trans of the road segment primitive at the m-th transfer. m =pri_h-pri_g,w m It is the weight of this transfer form;

[0114]

[0115] pri_g represents the strength of the road segment primitive before the transfer at the m-th transfer, and pri_h represents the strength of the road segment primitive after the transfer at the m-th transfer.

[0116] The driving style evaluation value J of the road segment is obtained based on the comprehensive driving style evaluation function, which is:

[0117] J = w1·Q + w2·D + w3·T; w1, w2, and w3 are the weighting coefficients corresponding to the average intensity value Q of road segment basic elements, the transfer diversity value D of road segment basic elements, and the transfer tendency value T of road segment basic elements;

[0118] w1 = w 1o·δ1; w2=w 2o ·δ2;w3=w 3o ·δ3;

[0119] w 1o Let w1 be the initial weight, and w 2o The initial weights corresponding to w2, w 3o The initial weights corresponding to w3;

[0120] δ1 is the adjustment coefficient corresponding to w1, δ2 is the adjustment coefficient corresponding to w2, and δ3 is the adjustment coefficient corresponding to w3.

[0121] The method for obtaining δ1, δ2, and δ3 includes the following steps:

[0122] Select M straight road segments or M turning road segments, and calculate the segment primitive transfer diversity value D. m1 The average strength value Q of the corresponding basic elements of each road segment at that time m1 This yields the data pair (Dm1, Qm1);

[0123] The calculation of the segment basic element transfer tendency value is T. m1 The average strength value Q of the corresponding basic elements of each road segment at that time m2 , obtain data pairs (T) m1 Q m2 );

[0124] The calculated segment primitive transfer diversity value is D. m1 The average value T of the corresponding road segment primitive at that time m2 , obtain data pairs (D m1 ,T m2 );

[0125] The segment element transfer diversity value of the M straight road segments or M turning road segments is not equal to 0, and the segment element transfer tendency value is not equal to 0.

[0126] Based on the least squares method, data from M road segments are used to analyze (D) m1 Q m1 ), (T m1 Q m2 ) and (D m1 ,T m2 ), Analysis of Q m1 With D m1 T m1 With Q m2 D m1 With T m2 The linear relationship between them yields the regression equation:

[0127] Q m1 =k1·D m1+c, Q m2 =k2·T m1 +c, T m2 =k3·D m1 +c.

[0128] For M straight road segments, the dispersion plot of the relationship between the segment primitive transfer diversity and the average intensity of the segment primitives was obtained using the method described above. See details. Figure 1 The dispersion plot of the relationship between the segment element transfer tendency and the average segment element intensity for straight road sections is detailed in [link to diagram]. Figure 2 The scatter plot showing the relationship between the diversity and tendency of segment element transfer in straight-through road sections is detailed in [link to scatter plot]. Figure 3 ;

[0129] For the regression equations obtained using the above method for M straight road segments, and the corresponding regression results diagram showing the relationship between the diversity of road segment element transfer and the average intensity of road segment elements, please refer to [link to diagram]. Figure 4 The regression results between the tendency of road segment element transfer and the average intensity of road segment elements in straight-through road sections are illustrated in the diagram. Figure 5 The regression results between the diversity of road segment element transfers and the tendency of road segment element transfers in straight-through road sections are illustrated in the diagram. Figure 6 The specific regression equation is as follows:

[0130] Q m1 =-0.452·D m1 +0.694; Q m2 =0.048·T m1 +0.64; T m2 =0.27·D m1 +0.17;

[0131] For M turning road segments, the dispersion plot of the relationship between the segment element transfer diversity and the average segment element intensity was obtained using the method described above. See details. Figure 7 The dispersion plot of the relationship between the segment element transfer tendency and the average segment element strength of the turning section is shown in the attached diagram. Figure 8 The dispersion plot showing the relationship between the diversity and tendency of segment element transfer in turning sections is detailed in [link to diagram]. Figure 9 ;

[0132] For the regression equations obtained using the above method for M turning road segments, and the corresponding regression results between the diversity of road segment element transfer and the average intensity of road segment elements, please refer to the diagram. Figure 10 The regression results between the tendency of road segment element transfer and the average intensity of road segment elements in turning sections are illustrated in the diagram. Figure 11 The regression results between the diversity of road segment element transfer and the tendency of road segment element transfer in turning sections are illustrated in the diagram. Figure 12The specific regression equation is as follows:

[0133] The regression equation obtained for M turning road segments using the above method is as follows:

[0134] Q m1 =-0.232·D m1 +0.48; Q m2 =0.156·T m1 +0.50; T m2 =-0.009·D m1 +0.35;

[0135] Construct a judgment matrix A based on k1, k2, and k3. The judgment matrix is ​​as follows:

[0136] Multiply the judgment matrix A by its rows to obtain a new vector B.

[0137] Taking the nth root (n=3) of each component of vector B yields vector C.

[0138] Adjusting vector C to a form where the sum is 1 yields the adjustment coefficient δ. i i = 1, 2, 3;

[0139]

[0140] Based on the regression equations for the M straight road segments mentioned above, the adjustment coefficients δ1 = 0.39, δ2 = 0.15, and δ3 = 0.46 are obtained.

[0141] Based on the regression equations for the M turning road segments mentioned above, the adjustment coefficients δ1 = 0.72, δ2 = 0.15, and δ3 = 0.13 are obtained.

[0142] The initial weights of the straight road segments are obtained using the entropy method:

[0143] w 1o =0.11, w 2o =0.85, w 3o =0.04;

[0144] The initial weights of the straight road segments are obtained using the entropy method:

[0145] w 1o =0.14, w 2o =0.80, w 3o =0.06;

[0146] Based on the above data, the weighting coefficients w1, w2, and w3 of the average intensity value Q of the road segment basic element, the transfer diversity value D of the road segment basic element, and the transfer tendency value T of the road segment basic element are obtained.

[0147] Straight-through section: w1 = 0.23, w2 = 0.68, w3 = 0.09;

[0148] Turning sections: w1 = 0.44, w2 = 0.52, w3 = 0.04.

[0149] After constructing the judgment matrix A, a consistency analysis is performed on judgment matrix A; the specific steps of the consistency analysis are as follows:

[0150] Calculate the consistency index (CI). Where λ max To determine the largest eigenvalue of matrix A, r is the order of the matrix, r = 3;

[0151] Calculate the consistency ratio (CR). Where RI = 0.52;

[0152] When CR < 0.1, the consistency of judgment matrix A is considered acceptable, and the judgment matrix is ​​reasonable; otherwise, the consistency of judgment matrix A is not acceptable, and then judgment matrix A is adjusted.

[0153] The specific adjustment method includes the following steps:

[0154] Find the minimum value of the regression coefficient in the judgment matrix, k i =min(k1,k2,k3);

[0155] Let k be the smallest regression coefficient in the judgment matrix. i The corresponding element is 1, that is, {k} i ,1 / k i} = 1; then perform consistency analysis again, and repeat this process until the matrix consistency is verified.

[0156] The judgment matrix of the regression equation obtained by applying the above method to the M straight road segments is as follows:

[0157]

[0158] Straight row judgment matrix,

[0159] CR = 0.03 << 0.1

[0160] Consistency check passed

[0161] The judgment matrix of the regression equation obtained by applying the above method to the M turning road segments is as follows:

[0162]

[0163] Turning judgment matrix

[0164] CR = 0.01 << 0.1

[0165] Consistency check passed

[0166] The driving style evaluation value J of the road segment is classified and labeled based on the driving style judgment method to obtain the driving style of the road segment; the driving style judgment method includes the following steps:

[0167] The driving style evaluation value J calculated for straight or turning road segments is subtracted from the threshold scores of different types of driving styles for the corresponding straight or turning road segments to obtain the driving style difference values ​​for different categories.

[0168] After taking the absolute value of the differences between different driving style categories, the driving style category corresponding to the driving style difference with the smallest absolute value is obtained as the driving style of the road segment.

[0169] Based on the example of the regression equation for straight and turning road sections above, the comprehensive evaluation function of driving style is obtained;

[0170] Straight-through section: J = 0.23·Q + 0.68·D + 0.09·T;

[0171] Turning section: J = 0.44·Q + 0.52·D + 0.04·T.

[0172] Driving styles are categorized as: Cautious, Normal, and Aggressive;

[0173] The threshold scores for different driving styles on straight and turning sections are set as follows:

[0174]

[0175] Finally, we obtained the driving style distribution map for the M straight road segments. (See details below) Figure 13 For a detailed driving style distribution map of the M turning sections, please refer to [link / reference]. Figure 14 .

[0176] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to the technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the inventive concept. For example, the above-described features have similar functions to (but are not limited to) those disclosed in this application.

Claims

1. A method for online evaluation of driving style, characterized in that, Includes the following steps: Extract driving behavior data from several straight or turning road sections; The driving behavior data of straight or turning road segments is divided into several road segment primitives based on the Bayesian condensed sequence segmentation algorithm. The latent Dirichlet assignment model based on variable coupling clusters the segment primitives corresponding to each segment and divides them into different categories. Values ​​are assigned to different categories of road segment primitives based on the kinetic energy of the different categories after clustering; Based on the category assignment of road segment primitives for any road segment, obtain the average intensity value Q, the transfer diversity value D, and the transfer tendency value T of road segment primitives for any road segment. The driving style evaluation value J of the road segment is obtained based on the comprehensive driving style evaluation function, which is: ; , , These are the weighting coefficients corresponding to the average intensity value Q of road segment primitives, the diversity value D of road segment primitive transfer, and the tendency value T of road segment primitive transfer. Based on the driving style judgment method, the driving style evaluation value J of the road segment is classified and labeled to obtain the driving style of the road segment.

2. The online driving style evaluation method according to claim 1, characterized in that, The driving behavior data for straight or turning sections includes speed, longitudinal acceleration, lateral acceleration, and jerk.

3. The online driving style evaluation method according to claim 1, characterized in that, Each road segment's corresponding road segment primitives are clustered into five categories; The specific method for assigning values ​​to different categories is as follows: Calculate the average kinetic energy of the road segment primitives included in each category, and use the average kinetic energy as the corresponding kinetic energy evaluation parameter for each category; Then, the kinetic energy evaluation parameters corresponding to each category are arranged in ascending order and defined as follows: low intensity category, lower intensity category, medium intensity category, higher intensity category, and high intensity category; Then, the low-intensity, lower-intensity, medium-intensity, higher-intensity, and high-intensity categories are assigned values ​​in an arithmetic sequence manner, where the assigned value is the intensity of each road segment element.

4. The online driving style evaluation method according to claim 1, characterized in that, The method for obtaining the average intensity value Q of road segment basic elements includes the following steps: Arrange the road segment primitives of each road segment in chronological order and assign them numbers 1-N respectively; The average strength value Q of the road segment is calculated using the following formula: N is the number of road segment primitives contained in the road segment. Let be the strength of the k-th segment primitive.

Images