Driving training method, device and readable storage medium based on artificial intelligence

CN118711268BActive Publication Date: 2024-12-03WUXI SAIBO YINGKE TECH
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202411179004.0
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-08-27
Publication Date
2024-12-03
Estimated Expiration
2044-08-27

Smart Images

  • Figure CN118711268B_ABST
    Figure CN118711268B_ABST
Patent Text Reader

Abstract

The invention discloses a driving training method, device and readable storage medium based on artificial intelligence, which relate to the technical field of driving school training and are used to solve the problem that the existing punch-in timekeeping system has the problem of cheating in punching in and cannot truly reflect the effective class hours of students. The invention compares the real-time position coordinates of the on-board timing terminal and the corresponding training car to facilitate accurate positioning and avoid the phenomenon of cheating in punching in. The face video of the main driver is analyzed to ensure that the student is trained by the same person to avoid the problem that the punch-in time of the student is inconsistent with the actual training time. The driving data of the main driver's seat is analyzed to obtain the effective class hours, so as to register for the test according to the effective class hours of the students and improve the pass rate of the test. The training strategy is formulated according to the total training value of each test item of the students, and corresponding prompt labels are generated to facilitate the coach to understand the learning progress of the students.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of driving school training, and in particular to a driving training method, device and readable storage medium based on artificial intelligence. Background Art

[0002] Driving school clocking is an important part of the training process. It ensures that students can study according to the prescribed hours and improves the efficiency and quality of training. Driving schools are usually equipped with clocking machines. Students need to sign in and out of the driving school within the specified time. The clocking machine will automatically record the students' study hours. When clocking in, students need to insert their student cards into the clocking machine and perform fingerprint sign-in or password verification according to the screen prompts.

[0003] However, the existing clock-in timekeeping system has the problem of cheating, and some students may circumvent the study hours requirement by cheating (such as punching in for others, forging study hours records, etc.). This not only violates the training regulations, but also may have a negative impact on the students' driving skills and safety awareness; at the same time, the clock-in timekeeping system cannot truly reflect the students' effective study hours. Summary of the invention

[0004] The purpose of the present invention can be achieved through the following technical solutions:

[0005] In a first aspect, the present invention provides a driving training method based on artificial intelligence, comprising the following steps:

[0006] The trainee's fingerprint and face authentication is performed through the vehicle timing terminal. After successful authentication, the authentication information is collected. At the same time, the vehicle timing terminal and the corresponding training car are located in real time to obtain positioning information, and the moving speed of the corresponding training car is obtained. When the moving speed of the training car is greater than the set speed threshold, the timing starts, and when the moving speed is equal to zero, the timing stops. At the same time, the training car information of the corresponding training car in this time period is collected;

[0007] The real-time position coordinates of the vehicle timing terminal and the corresponding training car are compared. If the position coordinates of the two coincide, the face video of the main driver is analyzed to obtain the single real-time duration. If the single real-time duration is less than or equal to the set duration threshold, the timing is invalid, and a corresponding timing invalidation instruction is generated and sent to the corresponding student's smart terminal; if the single real-time duration is greater than the set duration threshold, the driving data of the main driver's seat and the training data within the preset range of the training car are analyzed to obtain the effective class hours and driving training value respectively;

[0008] Conduct driving training value analysis on students to obtain the total training value of each test item of the students, and formulate training strategies based on the total training value of each test item of the students.

[0009] As a preferred embodiment of the present invention, the specific process of analyzing the facial video of the main driver is: dividing the facial video of the main driver into several frames of pictures, sorting the pictures in chronological order, comparing the facial features in the picture with the facial pictures in the authentication information and determining whether they are the same person. If they are the same person, the picture is marked as a same-person picture. If they are not the same person, the picture is marked as a different-person picture. Then, starting from the moment of the first same-person picture, the statistics are counted backward in sequence. If there is no different-person picture, the duration of the facial video is marked as a single real-time duration. If a different-person picture appears, the time difference between the previous same-person picture and the first same-person picture is calculated to obtain the divided time length. Then, the first same-person picture after the different-person picture is taken as the starting moment. Then, the same statistics are performed until several frames of pictures are counted, and all the divided time lengths are summed to obtain the single real-time duration.

[0010] As a preferred embodiment of the present invention, the specific process of analyzing the driving data of the main driving seat is:

[0011] The number of clutch-pressing moments is counted to obtain the total clutch-pressing times, and the displacement of each clutch-pressing is compared with a preset displacement threshold. If the displacement of the clutch-pressing is less than the displacement threshold, the displacement difference between the two is calculated to obtain a bias threshold, and several bias threshold coefficients are set. Each bias threshold coefficient corresponds to a value range, and the bias threshold is matched with all value ranges to obtain the corresponding bias threshold coefficient. The bias threshold and the bias threshold coefficient matched to the corresponding bias threshold coefficient are analyzed to obtain the total bias system value.

[0012] Analyze the number of flameouts and flameout moments: if the number of flameouts is greater than the set number threshold, sort the flameout moments in chronological order, calculate the time difference between two adjacent flameout moments to get the flameout interval duration, sort the flameout interval duration values ​​in chronological order, substitute the sorted values ​​into the line graph to get the duration points corresponding to the flameout interval duration values, connect two adjacent duration points to get the flameout interval line, calculate the slope of the flameout interval line, if the slope is negative, mark the slope as the first slope, and set all the first Sum the slopes and take the absolute values ​​to get the total value of the line drop; draw the duration threshold line in the line graph, identify the closed area below the duration threshold line and enclosed by the duration threshold line, mark it as the low threshold area, calculate the area of ​​all low threshold areas and sum them up, and mark the summed value as the low threshold total value; substitute it into model 1 to output the extinction base value; sum up all the single real time durations, partial system total values ​​and extinction base values ​​corresponding to the students respectively to get the total real time duration, partial system base value and extinction base total value, analyze the total real time duration, partial system base value and extinction base total value to get the effective class hours.

[0013] As a preferred embodiment of the present invention, the specific process of analyzing the training data within the preset range of the training vehicle is as follows:

[0014] The training data within a preset range of the training car is identified and divided into project data for each test item. The project data for each test item is parsed to obtain project training information, wherein the project training information includes the body position of the training car, project boundaries and quantity corresponding to each practice; the project training information is analyzed to obtain a single training value and the number of training failures for each test item.

[0015] As a preferred embodiment of the present invention, the model is , where XF is the flameout base value, XJ1 is the value of the total line drop value, DY2 is the value of the low threshold total value, PJ3 is the average value of all flameout intervals, Pmin is the value of the minimum flameout interval, XH5 is the value of the number of flameouts, cy is the set number threshold, η1, η2, η3, η4, η5 and η6 are all preset weight factors.

[0016] As a preferred embodiment of the present invention, the specific process of analyzing the driving training value of the trainee is as follows:

[0017] Sort the single training values ​​of each test item in chronological order, determine the size of two adjacent single training values, if the previous single training value is smaller than the next single training value, calculate the difference between the two and mark it as the training improvement value, sum up all the training improvement values ​​to get the total training improvement value, sum up all the single training values ​​and take the average to get the average training value, mark the largest single training value as the maximum training value, then convert the total training improvement value, average training value, maximum training value and number of training failures into straight line lengths according to a certain ratio, and construct two perpendicular lines with their center points intersecting with the straight line lengths corresponding to the total training improvement value and the average training value. Segments, connect the endpoints of the two line segments in sequence to obtain a rhombus, randomly select a point on any side of the rhombus, take the point as the starting point, draw a height whose length is equal to the length of the straight line corresponding to the maximum training value and is perpendicular to the rhombus, construct a quadrangular prism with the rhombus and the height, select the top surface of the quadrangular prism, draw a preset circle at the center of its top surface, select the center of the preset circle, take the center of the circle as the starting point to draw a height equal to the length of the straight line corresponding to the number of training failures inside the quadrangular prism, construct a cylinder with the preset circle and the height, dig out the entity corresponding to the cylinder on the quadrangular prism, calculate the entity volume of the remaining quadrangular prism and mark the value of its entity volume as the total training value of each test item.

[0018] As a preferred embodiment of the present invention, the specific process of formulating a training strategy according to the total training value of each test item of the trainee is as follows:

[0019] Compare the total training value of the test item with the preset training threshold value corresponding to the test item:

[0020] If the total training value is greater than or equal to the training threshold, a prompt label "No strategy formulation required, please continue training for the next test item" is generated and sent to the corresponding student and coach's smart terminal;

[0021] If the total training value is less than the training threshold, the difference between the two is calculated to obtain the training threshold difference, and the training threshold difference is compared with the difference range, which includes difference range 1 and difference range 2;

[0022] If the training threshold difference belongs to the difference range 1, the training threshold difference is multiplied by the duration conversion coefficient to obtain the set duration of the test item, and at the same time, "training the test item according to the set duration" is generated and sent together with the set duration to the corresponding student and coach's smart terminal;

[0023] If the training threshold difference falls within difference range 2, a prompt label of "Unqualified, re-learn the training for this test item" is generated.

[0024] In a second aspect, the present invention provides a driving training device based on artificial intelligence, comprising:

[0025] The vehicle-mounted timing terminal is used to perform fingerprint and face authentication on the trainees. After successful authentication, the authentication information is collected. At the same time, the vehicle-mounted timing terminal and the corresponding training car are positioned in real time to obtain positioning information, and the moving speed of the corresponding training car is obtained. When the moving speed of the training car is greater than the set speed threshold, the timing starts, and when the moving speed is equal to zero, the timing stops. At the same time, the training car information of the corresponding training car in the time period is collected;

[0026] An information analysis module is used to analyze the authentication information, positioning information and training vehicle information to obtain the effective time;

[0027] The driving training analysis module is used to analyze the training data within the preset range of the training car corresponding to the student to obtain the student's driving training value, and then analyze the student's driving training value to obtain the total training value of each test item of the student;

[0028] The training strategy module is used to formulate training strategies based on the total training value of each test item of the trainees.

[0029] In a third aspect, the present invention provides a computer-readable storage medium, which enables the terminal to execute the above method when instructions in the storage medium are executed by a processor of a terminal.

[0030] Compared with the prior art, the present invention has the following beneficial effects:

[0031] 1. The present invention compares the real-time position coordinates of the vehicle-mounted timing terminal and the corresponding training vehicle, which facilitates accurate positioning and avoids the phenomenon of clocking in and cheating.

[0032] 2. The present invention analyzes the facial video of the main driver to ensure that the trainee is trained by the same person, avoiding the problem of inconsistency between the trainee's clock-in time and the actual training time.

[0033] 3. The present invention analyzes the driving data of the main driver's seat to obtain effective class hours, so as to facilitate registration for the test according to the effective class hours of the students and improve the pass rate of the test.

[0034] 4. The present invention formulates training strategies based on the total training value of each test item of the trainee and generates corresponding prompt labels so that the coach can understand the learning progress of the trainee. BRIEF DESCRIPTION OF THE DRAWINGS

[0035] In order to facilitate understanding by those skilled in the art, the present invention is further described below with reference to the accompanying drawings.

[0036] Figure 1 is a block diagram of a device module of the present invention;

[0037] Figure 2 This is a functional block diagram of the vehicle-mounted timing terminal of the present invention;

[0038] Figure 3 It is a line graph schematic diagram of the present invention;

[0039] Figure 4 It is a schematic diagram of a quadrangular prism of the present invention. DETAILED DESCRIPTION

[0040] The technical solution of the present invention will be clearly and completely described below in conjunction with the embodiments. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. Example 1

[0041] See also Figure 1 - Figure 2As shown, an artificial intelligence-based driving training device is used for actual vehicle training of driving school students for learning subjects two and three, including an on-board timing terminal and an artificial intelligence cloud platform that is communicatively connected to the on-board timing terminal; a sign-in authentication unit, a positioning unit, a timing unit and a collection and analysis unit are arranged in the on-board timing terminal; the sign-in authentication unit is used to perform fingerprint and face authentication on the student, and after successful authentication, authentication information is obtained, and an authentication success instruction is generated and sent to the positioning unit and the timing unit. After receiving the authentication success instruction, the positioning unit performs real-time positioning of the on-board timing terminal and the corresponding training car to obtain positioning information; after receiving the authentication success instruction, the timing unit obtains the moving speed of the corresponding training car, and starts timing when the moving speed of the training car is greater than the set speed threshold, and stops timing when the moving speed is equal to zero, and at the same time collects the training car information of the corresponding training car within the time period, and stops collecting the training car information of the training car of the student when the student signs out through the sign-in authentication unit.

[0042] By positioning the on-board timing terminal and the training car to ensure that their positions are consistent, cheating in clocking in can be prevented. This can effectively avoid the situation where the trainee's clocking in position is inconsistent with the training car position, resulting in inaccurate timing.

[0043] The collection and analysis unit is used to summarize the authentication information, positioning information and training car information and send it to the artificial intelligence platform; among them, the authentication information includes the fingerprint, face picture and time of successful authentication during authentication; the positioning information includes the real-time position coordinates of the on-board timing terminal and the corresponding training car; the training car information includes the license plate number of the training car, the face video of the main driver's seat, the driving data of the main driver's seat and the training data within the preset range of the training car; the driving data includes clutch information and the number and time of shutdown; the clutch information includes the time of clutch depression and the corresponding clutch displacement, and the training data within the preset range of the training car includes the body position of the training car in the preset area of ​​each test item and the corresponding item sideline position.

[0044] The artificial intelligence platform includes a database, an information analysis module and an execution processing module.

[0045] The database is used to store authentication information, positioning information and training vehicle information.

[0046] The information analysis module is used to analyze the authentication information, positioning information and training vehicle information to obtain the effective time and driving training value. The specific analysis process is as follows:

[0047] Compare the real-time position coordinates of the on-board timing terminal and the corresponding training car. If the position coordinates of the two coincide, analyze the facial video of the main driver;

[0048] The specific process of analyzing the main driver's face video is as follows: the main driver's face video is divided into several frames of pictures, the pictures are sorted in chronological order, the facial features in the picture are compared with the facial pictures in the authentication information to determine whether they are the same person. If they are the same person, the picture is marked as a same-person picture. If they are not the same person, the picture is marked as a different-person picture. Then, the time of the first same-person picture is taken as the starting time, and then the count is made backward in sequence. If there is no different-person picture, the duration of the face video is marked as a single real-time duration. If a different-person picture appears, the time between the previous same-person picture and the first same-person picture is calculated. The time difference is obtained by the time difference, and the first picture of the same person after the picture of the different person is taken as the starting time, and then the same statistics are performed until several frames of pictures are counted, and all the time differences are summed up to get the single real time. If the single real time is greater than the set time threshold, the driving data of the main driver's seat and the training data within the preset range of the training car are analyzed; if the single real time is less than or equal to the set time threshold, the timing is invalid, and the corresponding timing invalidation instruction is generated and sent to the corresponding student's smart terminal; by analyzing the facial video of the main driver, it is ensured that the student is trained as the same person, avoiding the problem of inconsistency between the student's punch-in time and the actual training time.

[0049] Analyze the driving data of the main driver's seat: count the number of clutch-pressing moments to get the total number of clutch-pressing times, compare the displacement of each clutch-pressing with the preset displacement threshold, and if the displacement of the clutch-pressing is less than the displacement threshold, calculate the displacement difference between the two to get the bias threshold Δyj, j=1, 2, ..., n, n represents the total number of bias thresholds, and the value is a positive integer; set a number of bias threshold coefficients PYi, i=1, 2, ..., 10; each bias threshold coefficient PYi corresponds to a value range (ai-1, ai], a0=0; and 0<a1<a2<...<a10; that is, PY1=(0, a1], PY10=(a9, a10]; if Δyj∈(ai-1, ai], match the corresponding bias threshold coefficient PYi, and substitute the bias threshold and the matched bias threshold coefficient into the model The total partial coefficient value LC is obtained; the number of flameouts and flameout moments are analyzed: if the number of flameouts is greater than the set number threshold, the flameout moments are sorted in chronological order, the time difference between two adjacent flameout moments is calculated to obtain the flameout interval duration, the flameout interval duration values ​​are sorted in chronological order, and the sorted values ​​are substituted into the line graph to obtain the duration points corresponding to the flameout interval duration values, and the flameout interval line is obtained by connecting two adjacent duration points, such as Figure 3As shown, calculate the slope of the blackout line. If the slope is negative, mark the slope as the first slope, sum all the first slopes and take the absolute value to get the total value of the line drop; draw the duration threshold line in the line chart, identify the closed area below the blackout line and the duration threshold line, mark it as the low threshold area, calculate the area of ​​all low threshold areas and sum them, and mark the summed value as the total low threshold value; substitute into the model Output the flameout base value XF; where XJ1 is the value of the total line drop value, DY2 is the value of the total low threshold value, PJ3 is the average value of all flameout intervals, Pmin is the value of the minimum flameout interval, XH5 is the value of the flameout times, cy is the set times threshold, η1, η2, η3, η4, η5 and η6 are all preset weight factors, and their sizes are custom settings;

[0050] Sum up all the single real-time durations, partial system total values ​​and extinction base values ​​corresponding to the students to obtain the total real-time duration, partial system base value and extinction base total value and mark them as £T, £LC and £XF respectively, and set the preset weights corresponding to the total real-time duration, partial system base value and extinction base total value as fa1, fa2 and fa3;

[0051] Substitute the preset class hour analysis model YX=(£T×fa1-£LC×fa2-£XF×fa3+JZ×fa4) / ts to obtain the effective class hour YX, where JZ is the student's driving training value, fa4 is the preset weight of the driving training value, and ts is the class hour conversion coefficient; the effective class hour is sent to the database for storage, and the student's effective class hour is sent to the corresponding vehicle management office's driving test appointment platform.

[0052] By analyzing the driving data of the main driver's seat, the effective class hours can be obtained, so that students can register for the exam according to their effective class hours, thereby improving the pass rate of the exam. This avoids the existing clock-in timer not being able to reflect the actual training time of students.

[0053] The driving training analysis module is used to analyze the training data within the preset range of the training car corresponding to the trainee to obtain the student's driving training value. The specific analysis process is as follows:

[0054] The training data within the preset range of the training vehicle is identified and divided into project data for each test item, where the test items include reverse parking, parallel parking, fixed-point parking and starting on a slope, right-angle turning, curve driving and straight driving, etc.;

[0055] The project data of each test project is parsed to obtain project training information, wherein the project training information includes the body position of the training vehicle, project boundaries and quantity corresponding to each exercise.

[0056] Parse the number of project edges:

[0057] When the number of project edge lines is equal to one, the corresponding body edge line is obtained according to the body position of the training car, and the positional relationship between the body edge line and the project edge line is determined. If the body edge line exceeds the project edge line or overlaps or intersects, the single training value is zero, and the number of unqualified training times increases by one. If the body edge line does not exceed the project edge line, determine whether the two are parallel. If they are parallel, calculate the distance between the two to obtain the training distance, and compare the training distance with the preset distance threshold. If the training distance is less than or equal to the preset distance threshold, the training is marked as excellent training, and the preset The fixed value GD1 is marked as the training value. If the training distance is greater than the preset distance threshold, the distance difference between the two is calculated, and the distance difference is multiplied by the corresponding weight coefficient to obtain the distance deviation value. The result obtained by subtracting the distance deviation value from the preset fixed value GD1 is marked as the single training value. If they are not parallel, the vertical distances between the two ends of the body edge line and the project edge line are calculated to obtain the end distance one and the end distance two, respectively. The values ​​of the end distance one and the end distance two are marked as DJ1 and DJ2, and the result obtained by substituting them into the model GD1-|DJ2-DJ1|×μ1 is marked as the single training value, where μ1 is the set weight coefficient.

[0058] When the number of project edge lines is equal to two, obtain the corresponding two body edge lines according to the body position of the training car, and determine whether the two body edge lines have an intersection with the project edge line. If there is an intersection, the single training value is zero, and the number of unqualified training times increases by one; if there is no intersection and the body position of the training car is within the range of the two project edge lines, calculate the average distance between the two body edge lines and the corresponding project edge lines to obtain the average distance one and the average distance two, mark the values ​​of the average distance one and the average distance two as PJ1 and PJ2, substitute them into the model GD2 / (|PJ1-PJ2|×μ2+1) to obtain the result marked as the single training value, where GD2 is a preset constant and μ2 is a set weight coefficient;

[0059] When the number of project edge lines is equal to three, the corresponding three body edge lines are obtained according to the body position of the training car, and it is determined whether the three body edge lines have intersections with the project edge lines. If there is an intersection, the single training value is zero, and the number of unqualified training times increases by one; if there is no intersection and the body position of the training car is within the range of the corresponding three project edge lines, the average distances between the three body edge lines and the corresponding project edge lines are calculated to obtain average distance three, average distance four and average distance five, which are marked as PJ3, PJ4 and PJ5 respectively, and the result obtained by substituting into the model GD3 / (|PJ3-PY3|×μ3+|PJ4-PY4|×μ4+|PJ5-PY5|×μ5+1) is marked as the single training value, where GD3 is a preset constant, PY3, PY4 and PY5 are standard reference distances corresponding to the three body edge lines, and μ3, μ4 and μ5 are all set weight coefficients;

[0060] When the number of project edge lines is equal to four, the corresponding four body edge lines are obtained according to the body position of the training car, and it is determined whether the four body edge lines have any intersection with the corresponding ones, and the number of body edge lines with intersections is counted to obtain the number of pressed lines. If the number of pressed lines is equal to four, the single training value is zero, and the number of unqualified training times increases by one; if the number of pressed lines is less than four, the vertical distances between the two ends of the body edge line without intersection and the corresponding project edge line are calculated to obtain end distance three and end distance four and marked as DJ3 and DJ4, and substituted into the model GD4 / [|DJ3-PY6|+|DJ4-PY6|×μ6+1]-|DJ3-DJ4|×μ7 to obtain the results corresponding to the body edge lines without intersections, where GD4 is a preset constant, PY6 is the standard reference distance corresponding to the body edge lines without intersections, μ6 and μ7 are preset weight coefficients, and the results corresponding to all body edge lines without intersections are summed to obtain a single training value.

[0061] The single training value and the number of unqualified training times for each test item are sent to the database for storage; at the same time, the driving training value of the trainees is analyzed. The specific analysis process is as follows:

[0062] Sort the single training values ​​of each test item in chronological order, determine the size of two adjacent single training values, if the previous single training value is smaller than the next single training value, calculate the difference between the two and mark it as the training improvement value, sum up all the training improvement values ​​to get the total training improvement value, sum up all the single training values ​​and take the average to get the average training value, mark the largest single training value as the maximum training value, then convert the total training improvement value, average training value, maximum training value and number of training failures into straight line lengths according to a certain ratio, and use the total training improvement value and average training value as the corresponding values. The length of the straight line is used to construct two perpendicular line segments whose center points intersect. The endpoints of the two line segments are connected in sequence to obtain a rhombus. A point on any side of the rhombus is randomly selected, and the point is used as the starting point to construct a height that is equal to the length of the straight line corresponding to the maximum training value and perpendicular to the rhombus. A quadrangular prism is constructed with the rhombus and the height. The top surface of the quadrangular prism is selected, and a preset circle is drawn at the center of the top surface. The center of the preset circle is selected, and a height that is equal to the length of the straight line corresponding to the number of training failures is constructed inside the quadrangular prism with the center of the preset circle as the starting point. A cylinder is constructed with the preset circle and the height, and the entity corresponding to the cylinder is excavated on the quadrangular prism, such as Figure 4 As shown, the solid volume of the remaining quadrangular prism is calculated and the numerical value of its solid volume is marked as the total training value of each test item. The total training values ​​of all the trainee's trained test items are summed up to obtain the trainee's driving training value and sent to the database for storage; at the same time, the total training value of each test item of the trainee is sent to the training strategy module.

[0063] The training strategy module is used to formulate training strategies based on the total training value of each test item of the trainees. The specific process is as follows:

[0064] Compare the total training value of the test item with the preset training threshold value corresponding to the test item:

[0065] If the total training value is greater than or equal to the training threshold, a prompt label "No strategy formulation required, please continue training for the next test item" is generated and sent to the corresponding student and coach's smart terminal;

[0066] If the total training value is less than the training threshold, the difference between the two is calculated to obtain the training threshold difference, and the training threshold difference is compared with the difference range, which includes difference range 1 and difference range 2;

[0067] If the training threshold difference belongs to difference range 1, the training threshold difference will be multiplied by the duration conversion coefficient to obtain the set duration of the test item. At the same time, "Train the test items according to the set duration" will be generated and sent together with the set duration to the corresponding students' and coaches' smart terminals. This can avoid unreasonable teaching and training caused by practicing the test items too much or too little time.

[0068] If the training threshold difference falls within difference range 2, a prompt label of "Unqualified, re-learn the training for this test item" is generated.

[0069] The present invention formulates training strategies according to the total training value of each test item of the students, and generates corresponding prompt labels so that the coach can understand the learning progress of the students, avoid students with high learning quality spending too much time practicing mature test items, resulting in a waste of resources, and avoid students with low learning quality spending insufficient time on immature test items.

[0070] The above models are all custom models, and the weight coefficients in the models are pre-customized according to actual conditions and stored in the database. Example 2

[0071] The present invention provides a driving training method based on artificial intelligence, comprising the following steps:

[0072] S1: The trainee's fingerprint and face authentication is performed through the vehicle-mounted timing terminal. After successful authentication, the authentication information is collected. At the same time, the vehicle-mounted timing terminal and the corresponding training car are positioned in real time to obtain positioning information, and the moving speed of the corresponding training car is obtained. When the moving speed of the training car is greater than the set speed threshold, the timing starts, and when the moving speed is equal to zero, the timing stops. At the same time, the training car information of the corresponding training car in this time period is collected;

[0073] S2: Compare the real-time position coordinates of the vehicle timing terminal and the corresponding training car. If the position coordinates of the two coincide, analyze the face video of the main driver to obtain the single real-time duration. If the single real-time duration is less than or equal to the set duration threshold, the timing is invalid, and a corresponding timing invalidation instruction is generated and sent to the corresponding student's smart terminal; if the single real-time duration is greater than the set duration threshold, analyze the driving data of the main driver's seat and the training data within the preset range of the training car to obtain the effective class hours and driving training value respectively;

[0074] S3: Develop training strategies based on the total training value of each test item for the trainees. Example 3

[0075] The present invention provides a computer-readable storage medium. When the instructions in the storage medium are executed by a processor of a terminal, the terminal is enabled to execute the above-mentioned driving training method based on artificial intelligence.

[0076] It is understandable that if the method in the above embodiment is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application is essentially or the part that contributes to the prior art or all or part of the technical solution can be embodied in the form of a software product, and the computer software product is stored in a storage medium to execute all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0077] Since the embodiments of the readable storage medium part correspond to the embodiments of the method part, the embodiments and beneficial effects of the readable storage medium part can be found in the description of the embodiments of the method part, which will not be repeated here.

[0078] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to only specific implementation methods. Obviously, many modifications and changes can be made according to the content of this specification. This specification selects and specifically describes these embodiments in order to better explain the principles and practical applications of the present invention, so that those skilled in the art can understand and use the present invention well. The present invention is limited only by the claims and their full scope and equivalents.

Claims

1. A driving training method based on artificial intelligence, characterized in that: The following steps are involved: The trainee's fingerprint and face authentication is performed through the vehicle-mounted timing terminal. After successful authentication, the authentication information is collected. At the same time, the vehicle-mounted timing terminal and the corresponding training car are located in real time to obtain positioning information, and the moving speed of the corresponding training car is obtained. When the moving speed of the training car is greater than the set speed threshold, the timing starts, and when the moving speed is equal to zero, the timing stops. At the same time, the training car information of the corresponding training car in this time period is collected; The real-time position coordinates of the vehicle timing terminal and the corresponding training car are compared. If the position coordinates of the two coincide, the face video of the main driver is analyzed to obtain the single real-time duration. If the single real-time duration is less than or equal to the set duration threshold, the timing is invalid, and the corresponding timing invalidation instruction is generated and sent to the corresponding student's smart terminal; if the single real-time duration is greater than the set duration threshold, the driving data of the main driver's seat and the training data within the preset range of the training car are analyzed to obtain the effective class hours and driving training value respectively; the specific process of analyzing the driving training value of the student is as follows: Sort the single training values ​​of each test item in chronological order, determine the size of two adjacent single training values, if the previous single training value is smaller than the subsequent single training value, calculate the difference between the two and mark it as the training improvement value, sum up all the training improvement values ​​to get the total training improvement value, sum up all the single training values ​​and take the average to get the average training value, mark the largest single training value as the maximum training value, then convert the total training improvement value, average training value, maximum training value and number of training failures into straight line lengths according to a certain ratio, construct two perpendicular line segments with intersecting center points with the straight line lengths corresponding to the total training improvement value and the average training value, and connect the endpoints of the two line segments in sequence to get Rhombus, randomly select a point on any side of the rhombus, take the point as the starting point, draw a height with a length equal to the length of the straight line corresponding to the maximum training value and perpendicular to the rhombus, construct a quadrangular prism with the rhombus and the height, select the top surface of the quadrangular prism, draw a preset circle at the center of the top surface, select the center of the preset circle, take the center of the circle as the starting point, draw a height with a length equal to the length of the straight line corresponding to the number of training failures inside the quadrangular prism, construct a cylinder with the preset circle and the height, dig out the entity corresponding to the cylinder on the quadrangular prism, calculate the entity volume of the remaining quadrangular prism and mark the value of its entity volume as the total training value of each test item, sum up the total training values ​​of all the test items that the trainee has trained to obtain the trainee's driving training value; Formulate training strategies based on the total training value of each test item of the trainees; The specific process of analyzing the driving data of the main driver's seat is as follows: The number of clutch-pressing moments is counted to obtain the total clutch-pressing times, and the displacement of each clutch-pressing is compared with a preset displacement threshold. If the displacement of the clutch-pressing is less than the displacement threshold, the displacement difference between the two is calculated to obtain a bias threshold, and several bias threshold coefficients are set. Each bias threshold coefficient corresponds to a value range, and the bias threshold is matched with all value ranges to obtain the corresponding bias threshold coefficient. The bias threshold and the bias threshold coefficient matched to the corresponding bias threshold coefficient are analyzed to obtain the total bias system value. Analyze the number of flameouts and flameout moments: if the number of flameouts is greater than the set number threshold, sort the flameout moments in chronological order, calculate the time difference between two adjacent flameout moments to get the flameout interval duration, sort the flameout interval duration values ​​in chronological order, substitute the sorted values ​​into the line graph to get the duration points corresponding to the flameout interval duration values, connect two adjacent duration points to get the flameout interval line, calculate the slope of the flameout interval line, if the slope is negative, mark the slope as the first slope, and set all the first Sum the slopes and take the absolute value to get the total value of the line drop; draw the duration threshold line in the line graph, identify the closed area below the duration threshold line and enclosed by the duration threshold line, mark it as the low threshold area, calculate the area of ​​all low threshold areas and sum them, and mark the summed value as the low threshold total value; substitute it into model 1 to output the extinction base value; sum up all the single real time durations, partial system total values ​​and extinction base values ​​corresponding to the students to obtain the total real time duration, partial system base value and extinction base total value, analyze the total real time duration, partial system base value and extinction base total value to obtain the effective class hours; For the model , where XF is the flameout base value, XJ1 is the value of the total line drop value, DY2 is the value of the low threshold total value, PJ3 is the average value of all flameout intervals, Pmin is the value of the minimum flameout interval, XH5 is the value of the number of flameouts, cy is the set number threshold, η1, η2, η3, η4, η5 and η6 are all preset weight factors.

2. The driving training method based on artificial intelligence according to claim 1, characterized in that: The specific process of analyzing the main driver's face video is as follows: divide the main driver's face video into several frames of pictures, sort the pictures in chronological order, compare the facial features in the picture with the facial pictures in the authentication information and determine whether they are the same person. If they are the same person, mark the picture as a same-person picture. If they are not the same person, mark the picture as a different-person picture. Then, start from the moment of the first same-person picture and then count backward in sequence. If there is no different-person picture, mark the duration of the face video as a single real-time duration. If a different-person picture appears, calculate the time difference between the previous same-person picture and the first same-person picture to obtain the divided time duration. Then, take the first same-person picture after the different-person picture as the starting time, and then perform the same statistics until several frames of pictures are counted, and sum up all the divided time durations to obtain the single real-time duration.

3. The driving training method based on artificial intelligence according to claim 1, characterized in that: The specific process of analyzing the training data within the preset range of the training vehicle is as follows: The training data within the preset range of the training car is identified and divided into project data of each test item, and the project data of each test item is parsed to obtain project training information, wherein the project training information includes the body position of the training car corresponding to each exercise, the project boundary line and the number; The item training information is analyzed to obtain the single training value and the number of training failures for each test item.

4. The driving training method based on artificial intelligence according to claim 1, characterized in that: The specific process of formulating training strategies based on the total training value of each test item of the trainees is as follows: Compare the total training value of the test item with the preset training threshold value corresponding to the test item: If the total training value is greater than or equal to the training threshold, a prompt label is generated, indicating that no strategy formulation is required and that the next test item training should be continued, and the prompt label is sent to the corresponding smart terminals of the student and the coach; If the total training value is less than the training threshold, the difference between the two is calculated to obtain the training threshold difference, and the training threshold difference is compared with the difference range, which includes difference range 1 and difference range 2; If the training threshold difference belongs to the difference range 1, the training threshold difference is multiplied by the duration conversion coefficient to obtain the set duration of the test item, and at the same time, training for the test item based on the set duration is generated and sent together with the set duration to the corresponding smart terminals of the trainee and the coach; If the training threshold difference falls within difference range 2, a failure prompt label is generated, indicating that the training of the test item should be relearned.

5. A driving training device based on artificial intelligence, characterized in that: The artificial intelligence-based driving training method applied to any one of claims 1 to 4 comprises: The vehicle-mounted timing terminal is used to perform fingerprint and face authentication on the trainees. After successful authentication, the authentication information is collected. At the same time, the vehicle-mounted timing terminal and the corresponding training car are positioned in real time to obtain positioning information, and the moving speed of the corresponding training car is obtained. When the moving speed of the training car is greater than the set speed threshold, the timing starts, and when the moving speed is equal to zero, the timing stops. At the same time, the training car information of the corresponding training car in the time period is collected; An information analysis module is used to analyze the authentication information, positioning information and training vehicle information to obtain the effective time; The driving training analysis module is used to analyze the training data within the preset range of the training car corresponding to the student to obtain the student's driving training value, and then analyze the student's driving training value to obtain the total training value of each test item of the student; The training strategy module is used to formulate training strategies based on the total training value of each test item of the trainees.

6. A computer-readable storage medium, characterized in that: When the instructions in the storage medium are executed by a processor of a terminal, the terminal is enabled to execute the method according to any one of claims 1 to 4.