Analysis and evaluation method for static driving adaptability of old drivers

Abstract

The invention relates to an analysis and evaluation method for static driving adaptability of an old driver, and the method comprises the following steps: obtaining an evaluation index of a target driver, and carrying out the normalization and assimilation processing of the evaluation index, and obtaining index data; inputting the index data into a driving adaptability evaluation model, and obtaining a comprehensive evaluation score index output by the driving adaptability evaluation model; and determining an evaluation grade corresponding to the comprehensive evaluation score index in the target preset score interval. According to the method, multiple driving adaptability indexes are comprehensively considered, the driving adaptability of the old driver can be systematically analyzed and evaluated, the driving adaptability level of the old driver can be comprehensively known, potential driving adaptability problems can be found as early as possible, the risk of driving accidents is reduced, and safe driving of the old driver is promoted. The driving safety and comfort of old drivers are improved, scientific basis and reference are provided for related decisions, and management departments and individuals are helped to formulate reasonable driving training and support measures.

Description

Technical Field

[0001] The present application relates to the technical field of driving evaluation and analysis, and in particular to a method for analyzing and evaluating the static driving adaptability of elderly drivers. Background Art

[0002] Driving suitability testing is designed to assess whether an individual has the ability to continue driving safely, and usually includes an assessment of the driver's visual ability, cognitive function, operational skills, and other basic driving-related abilities. Through a comprehensive assessment, the driver's reaction ability and risk handling ability in the face of complex traffic environments are determined.

[0003] At present, different countries have adopted different strategies and measures in the driving adaptability testing of elderly drivers, including establishing standardized assessment procedures, promoting self-assessment tools, and providing targeted retraining courses, etc., to help elderly drivers maintain their driving ability and reduce safety hazards caused by potential cognitive or physiological defects. However, there is a lack of an evaluation method that combines multiple test indicators for the adaptability analysis of elderly drivers. Because a single driving adaptability test indicator only reflects the quality of a driver's driving adaptability in a certain aspect, there may be large differences in the single test indicators of different drivers. A more comprehensive method for evaluating the driving adaptability of middle-aged and elderly drivers is needed, which combines the influence of multiple indicators and comprehensively considers the influence and degree of each indicator on driving adaptability. Summary of the invention

[0004] In order to solve or partially solve the problems existing in the related technologies, the present application provides an analysis and evaluation method for the static driving adaptability of elderly drivers, provides ideas and solutions for the analysis and evaluation of the driving adaptability of the elderly, and helps to realize the driving adaptability evaluation of elderly drivers.

[0005] The first aspect of the present application provides a method for analyzing and evaluating the static driving adaptability of elderly drivers, comprising the following steps:

[0006] Obtaining evaluation indicators of the target driver, normalizing and homogenizing the evaluation indicators to obtain indicator data, the evaluation indicators including the target driver's night vision, dynamic vision, deep vision, complex reaction and precise operation indicators;

[0007] Inputting the index data into a driving adaptability evaluation model to obtain a comprehensive evaluation score index output by the driving adaptability evaluation model, wherein the driving adaptability evaluation model includes a factor analysis method based on a double base point method for driving adaptability evaluation and a weighted geometric mean model used as a synthesis model;

[0008] An evaluation level corresponding to the comprehensive evaluation score index in a target preset score range is determined, wherein the comprehensive evaluation standard for driving adaptability in the preset score range is obtained based on an improved K-means clustering algorithm.

[0009] Among them, the expression of the driving adaptability evaluation model is:

[0010]

[0011] In the formula, F i represents the i-th common factor, X1 represents the standardized value of night vision, X2 represents the standardized value of dynamic vision, X3 represents the standardized value of deep vision, X4 represents the standardized value of complex reaction, and X5 represents the standardized value of precise operation.

[0012] Among them, the indicator data obtained by normalizing and convergence processing the evaluation indicators include:

[0013] Based on the minimum-maximum normalization method, the evaluation indicators are normalized to obtain normalized data;

[0014] Obtain normalized data of night vision, complex reaction and precise operation indicators as indicator data;

[0015] And, the normalized data of deep vision and dynamic vision are converged and used as index data.

[0016] Among them, based on the minimum-maximum standardization method and based on the improved logistic value function, the expression of the improved logistic value function is:

[0017]

[0018] In the formula, y(x) is the normalized criterion score, and θ j are their mean and standard deviation respectively, and θ js are the mean and standard deviation of the initial values, κ j is a determined parameter.

[0019] Among them, the convergence of deep vision and dynamic vision includes:

[0020] Acquire normalized data of deep vision and normalized data of dynamic vision;

[0021] The deep vision normalized data was subtracted from 1 to obtain the deep vision index data of deep vision;

[0022] And, the dynamic vision normalized data is subtracted from 1 to obtain the dynamic vision index data of dynamic vision.

[0023] Among them, the factor analysis method improved based on the double base point method includes the following steps:

[0024] Calculate the correlation coefficients between indicators and construct the correlation coefficient matrix;

[0025] Solve the eigenvalues ​​and eigenvectors of the correlation coefficient matrix, and select the principal components with eigenvalues ​​greater than 1 as common factors. The number of highest loading variables is reduced by calculating the loading matrix and rotating the loading matrix.

[0026] The common factor score of each indicator was calculated based on the Thomson factor regression method;

[0027] Calculate the comprehensive factor scores and make them consistent;

[0028] According to the optimal vector and the worst vector, the degree of closeness of each evaluation object to the optimal factor solution is calculated.

[0029] Among them, the evaluation levels include good level, general level and attention level. The score interval of the good level is the first interval, the score interval of the general level is the second interval, and the score interval of the attention level is the third interval.

[0030] The technical solution provided by this application may have the following beneficial effects:

[0031] The present application provides an analysis and evaluation method for the static driving adaptability of elderly drivers. The method comprehensively considers multiple driving adaptability indicators. By establishing an evaluation system and an evaluation model, it can systematically analyze and evaluate the driving adaptability of elderly drivers, comprehensively understand the driving adaptability level of elderly drivers, discover potential driving adaptability problems early, reduce the risk of driving accidents, and promote safe driving of elderly drivers. The method provides a systematic and comprehensive approach that can assist in evaluating the driving adaptability of elderly drivers, help improve the driving safety and comfort of elderly drivers, provide a scientific basis and reference for relevant decision-making, and help management departments and individuals formulate reasonable driving training and support measures.

[0032] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0033] The above and other objects, features and advantages of the present application will become more apparent through a more detailed description of exemplary embodiments of the present application in conjunction with the accompanying drawings, wherein the same reference numerals generally represent the same components in the exemplary embodiments of the present application.

[0034] Figure 1 It is a schematic diagram of the flow chart of the analysis and evaluation method shown in the embodiment of the present application;

[0035] Figure 2 is a correlation coefficient diagram of various driving adaptability indicators of the analysis and evaluation method shown in the embodiment of the present application;

[0036] Figure 3 is a normal PP diagram of the driving adaptability comprehensive evaluation score index of the analysis and evaluation method shown in the embodiment of the present application;

[0037] Figure 4 This is the relationship between the driving adaptability evaluation scores of the three types of models and the number of accidents in the analysis and evaluation method shown in the embodiment of the present application. DETAILED DESCRIPTION

[0038] The embodiments of the present application will be described in more detail below with reference to the accompanying drawings. Although the embodiments of the present application are shown in the accompanying drawings, it should be understood that the present application can be implemented in various forms and should not be limited by the embodiments described herein. On the contrary, these embodiments are provided to make the present application more thorough and complete, and to fully convey the scope of the present application to those skilled in the art.

[0039] It should be understood that although the terms "first", "second", "third", etc. may be used in this application to describe various information, this information should not be limited to these terms. These terms are only used to distinguish the same type of information from each other. For example, without departing from the scope of this application, the first information may also be referred to as the second information, and similarly, the second information may also be referred to as the first information. Thus, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of this application, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined.

[0040] In the description of the present application, it should be understood that the terms "length", "width", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", etc., indicating the orientation or position relationship, are based on the orientation or position relationship shown in the drawings, and are only for the convenience of describing the present application and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present application.

[0041] Unless otherwise clearly specified and limited, the terms "installed", "connected", "connected", "fixed" and the like should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium, it can be the internal connection of two elements or the interaction relationship between two elements. For ordinary technicians in this field, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0042] The technical solution of the embodiments of the present application is described in detail below with reference to the accompanying drawings.

[0043] like Figure 1 The method for analyzing and evaluating the static driving adaptability of an elderly driver shown in the figure comprises the following steps:

[0044] S1. Obtain evaluation indicators of the target driver, normalize and homogenize the evaluation indicators to obtain indicator data.

[0045] The evaluation indicators include the target driver's night vision, dynamic vision, deep vision, complex reaction and precise operation indicators. In the field of road traffic safety, the statistical analysis of driving adaptability indicators plays an important role in evaluating and improving the driver's driving ability. Night vision, dynamic vision, deep vision, complex reaction ability and precise operation ability indicators are the key elements of driving adaptability, which are directly related to the driver's driving performance and safety risks in different environments and situations.

[0046] Night vision, or dark-adapted vision, refers to the ability of the human eye to discern details of objects in dark conditions. Night vision may be affected by a variety of reasons, including natural environmental factors, eye diseases, or refractive problems

[199] . With age, night vision usually declines, which may be caused by factors such as the degeneration of retinal photoreceptor cells and slower pupil response. According to national standards, the standard for night vision of motor vehicle drivers is less than or equal to 5 seconds. Dark adaptation time is used as an indicator to evaluate night vision, and night vision is tested using a night vision tester.

[0047] Dynamic vision, also known as moving body vision, refers to the vision tested when one or both of the person and the sight mark are in motion. It involves the eye's ability to capture moving targets, decompose images, and perceive details. Dynamic vision mainly reflects the driver's sensitivity to external objects while driving. It is tested using a dynamic vision tester, and the standard for night vision of motor vehicle drivers is set to be greater than or equal to 0.2s. Drivers with lower dynamic vision test values ​​also have lower sensitivity to external objects while driving.

[0048] Deep vision, also known as stereoscopic vision or depth perception, refers to people's ability to perceive the relative distance of objects in three-dimensional space, which is very important for judging the distance and speed between vehicles. With age, deep vision may decline, which may cause problems in driving, walking or other activities that require precise distance judgment. The national standard stipulates that the normal value of deep vision for motor vehicle drivers is -25mm to +25mm for beginners and -22mm to 22mm for in-service drivers. Deep vision is tested by using a deep vision tester, and the measurement is repeated 5 times in a row, and the average of the judgment error is taken as the evaluation result. If the measured value of the driver being tested is large, it means that his perception of spatial distance is relatively poor, and it is easy to make inaccurate judgments of position and distance during driving.

[0049] Complex reaction ability, that is, the ability to select reaction, involves the ability of an individual to quickly and accurately identify each stimulus when receiving multiple stimuli, and to make different responses according to each stimulus. It requires the individual to have the ability to quickly process information, make decisions and execute actions. In the fields of psychology and physiology, complex reaction ability is usually evaluated through the selection reaction time experiment. The experiment presents two or more stimuli and requires the subject to make different responses to each stimulus and record the time required. When the driver under test starts the test, he presses the left hand button immediately when he sees the yellow pattern, presses the right hand button immediately when he sees the green pattern, and steps on the right pedal immediately when he sees the red pattern. When he hears a beep in the headset, he does not operate no matter what color pattern he sees until the test is completed. Practice 3 times and test 10 times. The national standard stipulates that the normal value of complex reaction of motor vehicle drivers is less than or equal to 8 times for first-time drivers and less than or equal to 5 times for in-service drivers. If the driver under test makes many wrong choices in reaction, it means that the driver's reaction accuracy is poor when facing complex situations.

[0050] The precision operation ability test is an important parameter for evaluating whether the testee can accurately and quickly perform tasks in complex or emergency situations. This indicator usually involves multiple dimensions, including but not limited to reaction time, action accuracy, coordination, and decision-making ability in specific situations. The testee clicks the start button to control the ball on the screen to move through the narrow channel set by the system. If the ball falls, the test ends. The higher the level the tester passes, the system will set a time limit. Within the specified time, the ball must be moved to the end point. Similarly, if the red dot falls off the channel, the test ends. The national standard stipulates the normal value range of the operating function of motor vehicle drivers, which is less than or equal to 130 times for beginners and less than or equal to 150 times for in-service drivers. If the number of operating errors of the tested driver is large, in areas with large traffic such as urban areas, such drivers are prone to uneven distribution of attention to the left and right, rapid attention transfer, uncoordinated movements, etc., which are prone to traffic accidents.

[0051] Driving adaptability index is directly related to the driver's performance in complex traffic environments. There is also correlation between driving adaptability indexes. Through correlation analysis of driving adaptability indexes including night vision, dynamic vision, deep vision, complex reaction ability and precise operation ability, the interaction between different indicators and their comprehensive impact on driving safety can be revealed.

[0052] A single driving adaptability test index only reflects the quality of a driver's driving adaptability in a certain aspect, and there may be large differences in the single test index of different drivers. In order to more comprehensively evaluate the driving adaptability of middle-aged and elderly drivers, the influence of indicators such as night vision, dynamic vision, deep vision, complex reactions and precise operation should be combined, and the influence and degree of each indicator on driving adaptability should be comprehensively considered.

[0053] 1) Determination of the value range of a single indicator

[0054] The national standard GB18463-2001 Physical Conditions and Assessment Requirements for Motor Vehicle Drivers stipulates that the night vision standard for motor vehicle drivers is <=5s.

[0055] The national standard GB18463-2001 Physical Conditions and Assessment Requirements for Motor Vehicle Drivers stipulates that the normal value of a motor vehicle driver's dynamic vision is >= 0.2s.

[0056] The national standard GB18463-2001 Physical Conditions and Assessment Requirements for Motor Vehicle Drivers stipulates the normal values ​​of deep vision for motor vehicle drivers, which are -25mm to +25mm for initial test drivers and -22mm to +22mm for in-service drivers.

[0057] The national standard GB18463-2001 Physical Conditions and Assessment Requirements for Motor Vehicle Drivers stipulates the normal value of complex reactions for motor vehicle drivers: <=8 times for initial test drivers and <=5 times for in-service drivers.

[0058] The national standard GB18463-2001 Physical Conditions and Assessment Requirements for Motor Vehicle Drivers stipulates the normal range of motor vehicle drivers' operating functions: for first-time drivers, it is <=130 times; for in-service drivers, it is <=110 times.

[0059] The national standard GB18463-2001 Physical Conditions and Assessment Requirements for Motor Vehicle Drivers stipulates the normal value ranges for various indicators of motor vehicle drivers. However, due to the introduction of retirement policies and the adjustment of the upper age limit for applying for a driver's license, the number of elderly drivers has shown an upward trend, and the physical and cognitive functions of such drivers may decline to varying degrees with age. The applicability of previous indicator standards for this driving group remains to be verified. Therefore, based on the test data of this driving adaptability test, various indicators are analyzed to determine the normal value ranges of unilateral and bilateral quantiles for middle-aged and elderly drivers, in order to obtain the value ranges of night vision, dynamic vision, deep vision, complex reactions and precise operations, as shown in the following table.

[0060] Table 1 Value ranges of various driving adaptability indicators

[0061] Tab.1 The value range of each driving adaptability indicators

[0062]

[0063] According to the percentile values ​​in the above table, combined with the characteristics of the middle-aged and elderly driver population, the normal value range of each indicator is determined. Night vision reflects the night vision function of the driver being tested. The faster the reaction time of night vision, the better. This article selects the 80% percentile as the normal value range, and the normal value range of night vision is ≤9s.

[0064] Dynamic vision can reflect the driver's ability to distinguish objects in motion. Middle-aged and elderly drivers may have difficulty in identifying moving objects. In this paper, the 75% percentile on one side is selected as the normal value range for dynamic vision, and the normal value range of dynamic vision is ≥0.3s. Deep vision reflects the driver's ability to distinguish longitudinal depth. In this paper, the 90% percentile is selected as the normal value range for deep vision, and the normal value range of deep vision is -42mm~42mm. The number of wrong reactions detected by complex reactions can reflect the driver's ability to respond to changes in the external environment. The fewer the number of wrong reactions, the shorter the reaction time, and the higher the safety guarantee for driving. In this paper, the 95% percentile on one side of the number of complex reaction errors is selected as the normal value range, and the normal value range of complex reactions is ≤6 times. Precise operation can reflect the driver's coordination ability and operation accuracy. The stronger the operation ability, the more beneficial it is to safe driving. In this paper, the 80% percentile is selected as the normal value range, and the normal value range of precise operation is 124 times.

[0065] In order to reasonably construct a driving adaptability evaluation system, the driving suitability test indicators formulated in the "Automobile Driver Suitability Test and Evaluation"

[214] are referred to, namely the night vision, dynamic vision, deep vision, complex reaction and precise operation indicators tested in this experiment. At the same time, data processing is carried out according to the "Automobile Driver Suitability Test and Evaluation"

[215] to construct a driver adaptability evaluation system. The proposed evaluation system includes night vision (Y1), dynamic vision (Y2), deep vision (Y3), complex reaction (Y4) and precise operation (Y5). Except for the dynamic vision indicator, the smaller the value of the remaining indicators, the better. The mean of each indicator is taken as the sub-indicator, which is represented by the mean value of night vision (X1), the mean value of dynamic vision (X2), the mean value of depth perception judgment error (X3), the number of wrong reactions (X4) and the number of wrong operations (X5).

[0066] The evaluation index is based on the improved logistic linear function. In order to effectively control the size of each normalized criterion score, normalization should consider controlling the standardized score within a limited range, and the mean of the standardized score of each indicator should be a constant. Therefore, an improved logistic value function is introduced, as shown in formula (1), which is used to normalize the standard score. Let X ij =x ij (≥0) indicates the criterion score of the i-th (1≤i≤n) indicator in the j-th standard (1≤j≤m), achieving the goal of i j to y i Dimensionless mapping of j.

[0067]

[0068] In the formula, y(x) is the normalized criterion score, and θ j are their mean and standard deviation respectively, and θ js are the mean and standard deviation of the initial values, κ j is the determined parameter. The selected index has two attributes, namely, the bigger the better (LB) and the smaller the better (SB).

[0069] Assuming the weight vector is known, take κ for the initial weight vector j =1, The calculation formula is as follows:

[0070]

[0071] In the formula, w j is the indicator weight. Fix μ0, according to Eq. Calculate and obtain the parameter κ j When x ij approximate When , it is approximately linear, and when it is not equal to 0 or 1, it satisfies the transformation of equal size in formula (1). At this time, κ j The mean of is approximately equal to 1.

[0072] In order to reasonably construct a driving adaptability evaluation model, the test data must be preprocessed first. The units of the various driving adaptability indicators tested are not unified, and the values ​​vary greatly. In order to perform reasonable calculations, standardization is required. In addition, there may be certain correlations between night vision, dynamic vision, deep vision, complex reactions and precise operation indicators. In order to avoid the influence of some indicators not being in line with the actual situation, it is necessary to conduct an independent analysis of the single indicators. First, the minimum-maximum standardization method is used to normalize the data. In particular, considering that the smaller the values ​​of night vision, dynamic vision, complex reactions and precise operation indicators are, the better, while the smaller the absolute value of deep vision is, the better, and the larger the dynamic vision indicator value is, the better. In order to ensure that the trend changes of the above five indicator values ​​are consistent with the trend of driving adaptability, the absolute value of deep vision is also required as the indicator value, and the dynamic vision indicator value is converged, that is, after completing the standardization of the dynamic vision indicator value, the value obtained by subtracting the value from 1 is used as the corresponding indicator value. Some of the processed data are shown in the following table. After the normalization and convergence of the indicators are completed, there is no difference in the order of magnitude of the indicator values, so there is no need to unify the order of magnitude.

[0073] Table 2 Partial data after processing of various driving adaptability indicators

[0074] T ab.2 Partial data of processed driving adaptability indicators

[0075]

[0076]

[0077] S2. Input the index data into the driving adaptability evaluation model to obtain the comprehensive evaluation score index output by the driving adaptability evaluation model.

[0078] The driving adaptability evaluation model includes a factor analysis method based on a double base point method for driving adaptability evaluation and a weighted geometric mean model used as a synthesis model.

[0079] Factor analysis is a multivariate statistical analysis method that identifies potential factors by studying the correlation between variables for data dimension reduction and structural detection. The TOPSIS double base point method is a multi-criteria decision analysis method used to evaluate and select the optimal solution. It is based on the concept of ideal solution and calculates the distance of each solution to the positive ideal solution and the negative ideal solution for sorting and selection. This paper combines the two methods and uses the factor analysis method improved based on the double base point method to evaluate driving adaptability. The specific steps are as follows:

[0080] a. Suppose there are n indicators, each indicator has m observation values, and the original data is converted into a matrix based on data standardization.

[0081] b. Calculate the correlation coefficients between indicators, construct the correlation coefficient matrix R, and perform principal component analysis.

[0082] c. Solve the eigenvalue and eigenvector of the correlation coefficient matrix R. The eigenvalue is greater than 1.

[0083] d. Solve the loading matrix of each common factor C = [C] mxp. In order to reflect the meaning of the common factor variables, rotate the loading matrix to minimize the number of variables with the highest loading.

[0084] e. After the factor variables are determined, the Thompsons factor regression method is used to calculate the common factor score of each indicator.

[0085] f. Calculate the comprehensive factor score y using formula (3).

[0086]

[0087] In the formula, δ i (i=1,2,…,p) is the variance contribution rate, F i Score the common factors.

[0088] g. Establish an indicator system, assuming that the research object set is R ti (i=1,2,…,n), the index set is I tj (i=1,2,…,n), t is the test point.

[0089] h. for R ti Perform factor analysis and obtain the object set R ti Factor F ti and the factor score matrix (Y ti ) n×l .

[0090] i. Using factor scores as evaluation indicators, a matrix with l indicators, n evaluation objects and Y is formed. ti The new indicator system of the data is then consistent with formula (4).

[0091]

[0092] Where U ti To unify the matrix, Y ti is the factor score set.

[0093] j. Find the maximum value U of each column in the matrix max and the minimum value U min , the optimal vector U is calculated by formula (5) + and the worst vector U - .

[0094]

[0095] k. Calculate the i-th evaluation object and the optimal solution through formula (6) and the worst solution degree of proximity.

[0096]

[0097] 1. Calculate the closeness Ci between the ith evaluation object and the optimal factor solution through formula (7). The larger the Ci, the better the effect of the ith evaluation object.

[0098]

[0099] Furthermore, in order to avoid exaggerating or reducing the influence of some driving adaptability indicators, the correlation of each indicator was analyzed. The normality test found that the data of each indicator did not obey the normal distribution, so a non-parametric test method was required. Considering the sample size, the Kendall rank correlation coefficient was selected to conduct a correlation analysis of the five indicators of night vision, dynamic vision, deep vision, complex reaction and precise operation. The results are as follows: Figure 2 shown.

[0100] Depend on Figure 2 It can be seen that most of the correlation coefficients between the driving adaptability indicators are >0.1, indicating that there is a certain correlation between the indicators. Therefore, it is feasible to use factor analysis in this paper. The results of the KMO test and Bartlett sphericity test in the correlation analysis are shown in Table 3. The KMO test result is 0.606>0.6, which meets the prerequisite requirements of factor analysis, and the data passes the Bartlett sphericity test (p=0.002<0.05), indicating that the research data is suitable for factor analysis.

[0101] Table 3 KMO and Bartlett's sphericity test results

[0102] Tab.3 Results of KMO and Bartlett sphericity test

[0103]

[0104] New independent indicators are extracted by factor analysis, that is, the new indicators are obtained by the variance contribution rate of the new indicators generated after independent testing, and then the variance contribution rate is rotated by the maximum variance method. The results are shown in Table 4. The factor extraction and the amount of information extracted by the factors are analyzed, and it is found that the factor analysis extracts a total of 4 factors. The variance explanation rates of these 4 factors after rotation are 27.323%, 22.148%, 20.432% and 20.037%, respectively, and the cumulative variance explanation rate after rotation is 89.940%.

[0105] Table 4 Variance contribution rate

[0106] Tab.4 Variance contribution rate

[0107]

[0108] According to the results in Table 4, the variance cumulative contribution rate of the first four independent factors reached 89.940%, exceeding 80%. By applying the relevant knowledge of statistics and the significance of the variance cumulative contribution rate, it can be seen that the first four independent factors can fully reflect the driving adaptability level of the driver. Therefore, this paper extracts the first four factors as new indicators. The maximum variance rotation method (varimax) is used to further rotate the new factors in order to clarify the corresponding relationship between the factors and the research items. The factor loading matrix is ​​shown in Table 5.

[0109] Table 5 Factor loading matrix of new factors

[0110] Tab.5 The factor load matrix of the new factor

[0111]

[0112] Note: The percentage of information that can be extracted from a question is usually 0.4.

[0113] Table 5 shows the information extraction of each factor for the research items, as well as the corresponding relationship between the factors and the research items. It can be found that the commonality values ​​corresponding to all research items are higher than 0.4, which means that there is a strong correlation between the research items and the factors, and the factors can effectively extract information. After ensuring that the factors can extract most of the information of the research items, the corresponding relationship between the factors and the research items is analyzed. The absolute value of the factor loading coefficient is greater than 0.4, which means that the item has a corresponding relationship with the factor.

[0114] If the purpose of factor analysis is to condense information, the "component score coefficient matrix" table is ignored. If factor analysis is used for weight calculation, the "component score coefficient matrix" is needed to establish the relationship equation between factors and research items, that is, to establish a relationship expression based on standardized data, as shown in formula (8). Each common factor is the coefficient of the linear expression of the initial indicator variable. After the analysis is brought in, new indicators that are independent of each other are obtained.

[0115]

[0116] In the formula, F i represents the i-th common factor, X1 represents the standardized value of night vision, X2 represents the standardized value of dynamic vision, X3 represents the standardized value of deep vision, X4 represents the standardized value of complex reaction, and X5 represents the standardized value of precise operation.

[0117] For evaluation problems, the commonly used composite models are weighted sum model, weighted product model and weighted geometric mean model. The first two models have a large compensation, while the weighted geometric mean model first takes the logarithm of each indicator value, then calculates the arithmetic mean, and then converts it into an exponential form to obtain the geometric mean, and finally multiplies it with the weight and sums it to obtain the weighted geometric mean. This data processing method makes the weighted geometric mean model less affected by extreme values ​​and more robust. Therefore, this paper uses the weighted geometric mean model as the composite model, and the calculation is as shown in formula (9).

[0118]

[0119] Combined with the weighted geometric mean model, the variance contribution rate of the common factor is used as the weight, and the sum of the variance contribution rate is used as the basis for weight normalization. Considering the calculation based on the percentage system, the comprehensive score is expanded by 100 times. In addition, the current values ​​of all indicators are as small as possible. In order to meet the conventional scoring standards, the current score is subtracted from 100. The final driver's driving adaptability evaluation model is shown in formula (10).

[0120]

[0121] Verify the comprehensive scoring index based on the PP diagram. By comparing the cumulative distribution function of continuous data with the cumulative distribution function of the specified theoretical distribution, we can evaluate whether the data conforms to a certain theoretical distribution. The PP diagram is used to analyze whether the driving adaptability comprehensive evaluation score index DAE conforms to the normal distribution. The actual data cumulative ratio is used as the X-axis, and the corresponding normal distribution cumulative ratio is used as the Y-axis to make a scatter plot. The normal PP diagram is as follows: Figure 3 As shown in , if the scatter plot is approximately a diagonal straight line, it means that the data is normally distributed, otherwise it means that the data is non-normal. Figure 3 It can be seen that the driving adaptability comprehensive evaluation score index DAE basically conforms to the normal distribution on the whole.

[0122] In order to evaluate the effectiveness of the driving adaptability evaluation model DAE constructed based on the improved logistic linear function combined with Topsis, factor analysis and linear weighted model, the evaluation results of the model DAE1 and the other two models are compared and analyzed. That is, the driving adaptability evaluation model DAE1 constructed by the Topsis method and the driving adaptability evaluation model DAE2 constructed based on the maximum and minimum standardization combined with Topsis, factor analysis and linear weighted model. The model formula is shown in formula (11). The correlation scatter plot of the comprehensive score of driving adaptability and the number of accidents in the above three methods is drawn, as shown in Figure 4 As shown, the higher the score and the fewer accidents, the better the driving adaptability.

[0123]

[0124] Depend on Figure 4 It can be seen that the scores of each driving adaptability evaluation model generally follow the rule that the higher the score, the fewer accidents, and the better the driving adaptability. Some points with more accidents are distributed at higher scores or points with fewer accidents are scattered at lower scores, which may be due to differences in the states of the test drivers during the test.

[0125] Furthermore, the correlation coefficient between the comprehensive score of driving adaptability and the number of accidents was calculated. The larger the correlation coefficient, the more effective the model evaluation result. First, the Shapiro-Wilk test was used to test the normality of the data. The data all obeyed the normal distribution, so the Pearson correlation coefficient was used for data analysis. The results are shown in Table 6.

[0126] Table 6 ANOVA results of different driving adaptability evaluation scores and accident numbers

[0127] Tab.6 Results of variance analysis of different driving adaptabilityevaluation scores and accident frequency

[0128]

[0129] Note: ***, **, * represent 1%, 5%, 10% significance levels respectively

[0130] According to Table 6, the scores of the three types of driving adaptability evaluation models and the number of accidents are significant at the 1% level, and the correlation coefficient of the model DAE constructed in this study is significantly higher than that of the comparison models DAE1 and DAE2, indicating that the model constructed in this paper has a high correlation with the number of accidents, and the new model is better than the comparison model.

[0131] In order to evaluate the effectiveness of the constructed model DAE for driving adaptability evaluation, the samples were divided into four groups, namely, the accident-free group, the one-time accident group, the two-time accident group, and the three-time or more accident group. The one-way ANOVA was used to test the differences between the groups. First, the variance homogeneity test was performed. The results are shown in Table 7. The driving adaptability evaluation score was the largest in the accident-free group and the smallest in the accident-prone group. Moreover, the evaluation score gradually decreased with the increase in the number of accidents. In addition, the significant P value of DAE was 0.011**, and the data did not meet the variance homogeneity. The significant P value of DAE1 was 0.236, and the data met the variance homogeneity. The significant P value of DAE2 was 0.152, and the data met the variance homogeneity. Since DAE did not meet the variance homogeneity, Welch's variance test was used. The P value of the variance analysis result was 0.000***≤0.05, so the statistical results were significant, indicating that there were significant differences in DAE with different numbers of accidents. DAE1 and DAE2 meet the variance homogeneity, and the single sample variance test is used. The P values ​​of the variance analysis results are 0.015**≤0.05 and 0.011**≤0.05 respectively. Therefore, the statistical results are significant, indicating that there are significant differences in the number of accidents in DAE1 and DEA2. It can be seen that the driving adaptability evaluation scores and the number of accidents of the three models are significantly different at the 1% level, but the differences between the accident groups of the new model DAE are significantly higher than those of the two comparison models.

[0132] Table 7 ANOVA results of different driving adaptability evaluation scores and accident numbers

[0133] Tab.7 Results of variance analysis of different driving adaptabilityevaluation scores and accident frequency

[0134]

[0135]

[0136] Note: ***, **, * represent 1%, 5%, 10% significance levels respectively

[0137] By comparing the effects of different methods, the optimal strategy was selected, and the effect quantification analysis method was used to compare the difference amplitude. The results are shown in Table 8. The results of the effect quantification analysis show that based on DAE, η 2 =0.288, indicating that 28.8% of the data differences come from differences between different groups, and Cohen's f = 0.637, reflecting that the degree of difference in the effect quantification of the data is a large degree of difference. Based on DAE1, η 2 =0.201, indicating that 20.1% of the data differences come from differences between different groups, and Cohen's f = 0.501, reflecting that the degree of difference in the effect quantification of the data is a large degree of difference. Based on DAE2, η 2 =0.212, indicating that 21.2% of the data differences come from differences between different groups, Cohen's f = 0.519, reflecting that the degree of difference in the effect quantification of the data is a large degree of difference. It can be seen that the inter-group difference of DAE is the largest (2182.152), and its η 2 The highest value (η 2 =0.288), indicating that DAE has the highest degree of explanation for the variability of the dependent variable, compared with DAE1 (η 2 =0.201) and DAE2(η 2 =0.212) has a stronger effect. This shows that the evaluation results of the new model constructed in this paper have a high correlation with the number of accidents, and it is significantly better than the other two types of comparative models in distinguishing the differences between the accident groups.

[0138] Table 8 Effect quantification analysis results

[0139] Tab.8 Effect quantification analysis results

[0140]

[0141] In summary, the improved logistic nonlinear value function used to construct the driving adaptability evaluation model controls the difference in means to a certain extent compared with the linear normalized data processing method. The new model adopts a method combining factor analysis and Topsis. Factor analysis uses dimensionality reduction technology to transform multiple related variables into a few irrelevant potential factors, thereby eliminating the multicollinearity problem in the original data, more objectively determining the weight of each indicator, and further combining the Topsis method to take into account the characteristics of multi-attribute decision-making. The weighted geometric mean model with less compensation than the linear weighted model is further adopted to achieve a more effective driving adaptability evaluation.

[0142] S3. Determine the evaluation level corresponding to the comprehensive evaluation score indicator in the target preset score range.

[0143] The comprehensive evaluation criteria for driving adaptability in the preset score range are obtained based on the improved K-means clustering algorithm.

[0144] The comprehensive evaluation score indicators of driving adaptability are clustered to obtain the comprehensive evaluation criteria of driving adaptability for middle-aged and elderly drivers. The K-means clustering algorithm is a classic unsupervised clustering algorithm that minimizes the variance of data points within a cluster through an iterative process, thereby maximizing the differences between different clusters. However, the K-means clustering algorithm requires the number of clusters K to be specified in advance, and is sensitive to the selection of the initial centroid. Different initial values ​​may lead to different clustering results. The initial cluster center points of the traditional K-means clustering algorithm are selected randomly. Improper selection of the initial center points may cause the algorithm to fall into a local optimal solution. This paper adopts an improved Kmeans++ clustering algorithm to optimize the selection method of the initial center points to improve the convergence speed and accuracy of the clustering algorithm.

[0145] The improved Kmeans++ clustering algorithm and the original K-means clustering algorithm were used to cluster the comprehensive evaluation score indicators, and the clustering effects of the two methods were compared. Based on the better clustering effect, the comprehensive evaluation standard of driving adaptability of middle-aged and elderly drivers was further obtained. The results are shown in Table 9. It can be found that the clustering effect of Kmeans++ is significantly better than that of K-means in terms of silhouette coefficient, DBI index and CH index, so the clustering results of the former are selected for further analysis.

[0146] Table 9 Comparison of improved K-means clustering results and original K-means clustering results

[0147] Tab.9 Comparison between Improved K-means Clustering Results andOriginal K-means Clustering Results

[0148] Clustering Algorithms Silhouette coefficient DBI CH K-means clustering 0.544 0.555 136.575 K-means++ clustering 0.564 0.500 186.628

[0149] The original data were analyzed according to the comprehensive scoring index, and the test drivers were classified according to the comprehensive evaluation index. The results are shown in Table 10. According to the clustering results of Kmeans++, the test drivers were divided into three categories: good, general, and need to pay attention. It can be seen that the score range of good driving adaptability is [80.23, 95.90], the score range of general driving adaptability is [70.16, 80.22], and the score range of need to pay attention driving adaptability is [44.43, 67.77]. Among them, the proportion of test drivers whose driving adaptability needs to be paid attention is relatively large, and they may need to pay further attention to whether the current state is suitable for continuing to drive.

[0150] Table 10 Driving adaptability score ranges of test drivers

[0151] Tab.10 Driving adaptability scores of drivers

[0152]

[0153] Finally, it should be noted that, in this article, relationships such as first and second, etc. are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms include, include or any other variations are intended to cover non-exclusive inclusion, so that a process, method, article or device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, article or device.

[0154] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0155] The embodiments of the present application have been described above, and the above description is exemplary, not exhaustive, and is not limited to the disclosed embodiments. Many modifications and changes will be apparent to those of ordinary skill in the art without departing from the scope and spirit of the described embodiments. The selection of terms used herein is intended to best explain the principles of the embodiments, practical applications, or improvements to the technology in the market, or to enable other persons of ordinary skill in the art to understand the embodiments disclosed herein.

Claims

1. A method for analyzing and evaluating the static driving adaptability of elderly drivers, characterized in that: The following steps are involved: Acquire evaluation indicators of the target driver, and normalize and converge the evaluation indicators to obtain indicator data, wherein the evaluation indicators include night vision, dynamic vision, deep vision, complex reaction and precise operation indicators of the target driver; Inputting the index data into a driving adaptability evaluation model to obtain a comprehensive evaluation score index output by the driving adaptability evaluation model, wherein the driving adaptability evaluation model includes a factor analysis method based on a double base point method for driving adaptability evaluation and a weighted geometric mean model used as a synthesis model; Determine the evaluation level corresponding to the comprehensive evaluation score indicator in the target preset score range, wherein the driving adaptability comprehensive evaluation standard in the preset score range is obtained based on an improved K-means clustering algorithm.

2. The analysis and evaluation method for static driving adaptability of elderly drivers according to claim 1 is characterized in that: The expression of the driving adaptability evaluation model is: In the formula, F i represents the i-th common factor, X1 represents the standardized value of night vision, X2 represents the standardized value of dynamic vision, X3 represents the standardized value of deep vision, X4 represents the standardized value of complex reaction, and X5 represents the standardized value of precise operation.

3. The analysis and evaluation method for static driving adaptability of elderly drivers according to claim 1 is characterized in that: The normalization and convergence processing of the evaluation index to obtain the index data includes: Based on the minimum-maximum standardization method, the evaluation indicators are respectively normalized to obtain normalized data; Acquiring normalized data of the night vision, the complex reaction and the precise operation index as the index data; And, the normalized data of the deep vision and the dynamic vision are converged and processed as index data.

4. The analysis and evaluation method for static driving adaptability of elderly drivers according to claim 3 is characterized in that: The minimum-maximum standardization method is based on an improved logistic value function, and the expression of the improved logistic value function is: In the formula, y(x) is the normalized criterion score, and θ j are their mean and standard deviation respectively, and θ js are the mean and standard deviation of the initial values, κ j is a determined parameter.

5. The analysis and evaluation method for static driving adaptability of elderly drivers according to claim 3 is characterized in that: The convergence processing of the deep vision and the dynamic vision comprises: Acquire normalized data of deep vision and normalized data of dynamic vision; Subtracting the deep vision normalized data from 1 to obtain the deep vision index data of the deep vision; And, the dynamic vision normalization data is subtracted from 1 to obtain the dynamic vision index data of the dynamic vision.

6. The analysis and evaluation method for static driving adaptability of elderly drivers according to claim 1 is characterized in that: The factor analysis method improved based on the double base point method comprises the following steps: Calculate the correlation coefficients between indicators and construct the correlation coefficient matrix; Solving the eigenvalues ​​and eigenvectors of the correlation coefficient matrix, selecting the principal components with eigenvalues ​​greater than 1 as common factors, wherein the number of highest loading variables is reduced by calculating the loading matrix and rotating the loading matrix; The common factor score of each indicator was calculated based on the Thomson factor regression method; Calculating comprehensive factor scores and performing consistency processing on the comprehensive factor scores; According to the optimal vector and the worst vector, the degree of closeness of each evaluation object to the optimal factor solution is calculated.

7. The method for analyzing and evaluating the static driving adaptability of elderly drivers according to claim 1, characterized in that: The evaluation levels include a good level, a general level, and a level requiring attention. The score interval of the good level is a first interval, the score interval of the general level is a second interval, and the score interval of the level requiring attention is a third interval.

Images