Operating driver safety analysis early warning method based on portrait data model

Abstract

The invention discloses an operating driver safety analysis and early warning method based on a figure portrait data model, and relates to the technical field of traffic safety. Comprising the following steps of obtaining driving behavior data of a running vehicle in real time, generating an overtime fatigue driving record table, constructing a driver posture safety quantitative evaluation model, training and verifying the model, and describing reaction degrees of a driver under different early warning types. According to the operator driver safety analysis and early warning method based on the figure portrait data model, the driving behavior data of the operating vehicle is obtained in real time, the quantitative evaluation index is determined by fully combining the driving posture characteristics, and the characteristics of scientificity, comprehensiveness, universality, high operability and the like are shown; the method has important practical significance for guaranteeing the personal safety of a driver of an operating vehicle and promoting the driving standardization and safety.

Description

Technical Field

[0001] The present invention relates to the field of traffic safety technology, and in particular to a safety analysis and early warning method for operating drivers based on a character portrait data model. Background Art

[0002] There are a large number of operating vehicles in my country, which undertake the vast majority of passenger and freight transportation in my country's road transport industry. With the continuous development of my country's national economy, the number of operating vehicles and the scale of the entire transportation industry will continue to expand, but the accidents caused by operating vehicles still bring huge disadvantages to the development of the industry in terms of quantity or severity. In recent years, mass casualties or accidents with huge social impacts caused by operating vehicles still occur frequently, seriously hindering the safe development of the road transport industry. It can be seen that how to improve and enhance the traffic safety level of operating vehicles has become a key problem restricting the healthy and rapid development of my country's passenger and freight transportation industry, and it is imperative to conduct research on it.

[0003] The invention patent with announcement number CN109649396A discloses a method for detecting the safety of drivers of commercial vehicles. This method establishes an initial detection model for the safety of commercial drivers based on the analytic hierarchy process according to the vehicle motion state data during driving and the data of the relative relationship parameters between the vehicle and the surrounding traffic environment, obtains the total score of the driver safety detection, and performs safety classification and judgment to evaluate the driver's safety. However, this method does not fully implement the monitoring responsibility of the transportation company, and it is impossible to monitor the vehicle as soon as it moves, and the driver fails to be warned and reminded in time to stop the violation during driving. In addition, the information of the violation will be lost during the transmission process, resulting in low detection accuracy in judging the driver's state and behavior. Summary of the invention

[0004] In view of the deficiencies in the prior art, the present invention provides a safety analysis and early warning method for operating drivers based on a character portrait data model, which solves the existing problems.

[0005] To achieve the above objectives, the present invention is implemented through the following technical solutions: a safety analysis and early warning method for operating drivers based on a character portrait data model, comprising the following steps:

[0006] Step 1: Collect and store driver personal information data, vehicle parking record data, active safety warning data and warning video record data;

[0007] Step 2: extract the data with the warning category of overtime fatigue in the active safety warning information detailed data table, and generate an overtime fatigue driving record table;

[0008] Step 3: Construct a quantitative evaluation model for driver posture safety;

[0009] Step 4: Using driving videos as data sets, the data sets are divided into training sets and test sets in a ratio of 4:1. The training set is input into the driver posture safety quantitative evaluation model for training. When the training error reaches the set threshold or the learning rate reaches the set upper limit, the training is stopped. Then the test set is input into the driver posture safety quantitative evaluation model to obtain the corresponding evaluation index.

[0010] Step 5: Characterize the driver's reaction degree under different warning types according to the evaluation indicators, cluster the indicators, and obtain the classification results of drivers under different warning types.

[0011] Preferably, in step one, the driver's personal information data, vehicle parking record data, active safety warning data and warning video record data are associated to generate a detailed data table of active safety warning information.

[0012] Preferably, the detailed data table of active safety warning information is specifically as follows: the active safety warning data is associated with the warning video recording data according to the warning number, the active safety warning data is associated with the driver's personal information data according to the driver's name, and the active safety warning data is associated with the vehicle parking record data according to the license plate number, and the active safety warning information detailed data table is constructed using the associated data.

[0013] Preferably, the overtime fatigue driving record table in step 2 includes the license plate number, driver's name, fatigue driving start time, fatigue driving end time and duration.

[0014] Preferably, the process of constructing the driver posture safety quantitative evaluation model in step 3 is:

[0015] 1) Using driving videos as data sets, we conduct correlation analysis of quantitative evaluation indicators based on quantitative evaluation indicators and evaluation postures Xi. The quantitative evaluation indicators include emergency response time Tr, sight deviation time Td, hand departure time Tl, and blinking frequency Fb. The evaluation postures Xi include normal driving X1, one-handed driving X2, gear manipulation X3, eating and drinking X4, picking up and putting things X5, and playing with mobile phones X6. The original data is processed by deviation standardization to obtain the original values ​​εij of each indicator and each posture, where i is the evaluation posture number and j is the evaluation indicator number.

[0016] 2) According to the original values ​​εij of each index and each posture, the fuzzy consistent judgment matrix R is established by using the fuzzy analytic hierarchy process, and the weight value of each quantitative evaluation index is calculated by the least squares method;

[0017] 3) According to the weight value, determine the optimal weight vector W, calculate the risk score Yi of each driving posture, and complete the establishment of the quantitative evaluation model of driver posture safety.

[0018] Preferably, in step 4, the accuracy of the evaluation index needs to be verified. If the accuracy requirement is not met, the data set is re-collected and the training step is repeated until the accuracy requirement is met.

[0019] Preferably, the evaluation indicators include:

[0020] Extract frames from the warning video and extract pictures;

[0021] Detect the generated images, extract the text in the images, and generate a detailed warning data table based on the text;

[0022] According to the data in the warning detailed data table, the reaction time, speed change before the warning and speed change after the warning are obtained.

[0023] Preferably, the warning types include forward collision warning, lane departure warning and distraction fatigue warning.

[0024] Beneficial Effects

[0025] The present invention provides a safety analysis and early warning method for operating drivers based on a character portrait data model.

[0026] Compared with the prior art, it has the following beneficial effects:

[0027] 1. The safety analysis and early warning method for operational drivers based on the character portrait data model extracts effective driver behavior characteristics through a large amount of real driving data, and obtains driving behavior thresholds of different categories of drivers by classifying the characteristics. The calculated thresholds can be further used in the driver driving habit evaluation system, achieving numerical precision classification that cannot be achieved by questionnaire surveys. At the same time, for different warning types, different indicators that can reflect the intensity and speed of driver reactions are proposed, avoiding the impact of drivers having essentially different reactions to different warning types.

[0028] 2. The safety analysis and early warning method for operating drivers based on the character portrait data model obtains the driving behavior data of operating vehicles in real time, fully combines the characteristics of driving posture to determine quantitative evaluation indicators, and exhibits the characteristics of scientificity, comprehensiveness, versatility, and strong operability. It has important practical significance for ensuring the personal safety of operating vehicle drivers and promoting the standardization and safety of driving. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 It is a schematic diagram of the method flow of the present invention;

[0030] Figure 2 Schematic diagram of the process of constructing the driver posture safety quantitative evaluation model of the present invention. DETAILED DESCRIPTION

[0031] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not 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.

[0032] See also Figure 1-2 , the present invention provides two technical solutions:

[0033] The first implementation method: a safety analysis and early warning method for operating drivers based on a character portrait data model, comprising the following steps:

[0034] Step 1: Collect and store driver personal information data, vehicle parking record data, active safety warning data and warning video record data.

[0035] The driver's personal information data, vehicle parking record data, active safety warning data and warning video record data are associated to generate a detailed data table of active safety warning information.

[0036] The active safety warning information detailed data table is specifically as follows: the active safety warning data is associated with the warning video record data according to the warning number, the active safety warning data is associated with the driver's personal information data according to the driver's name, and the active safety warning data is associated with the vehicle parking record data according to the license plate number, and the active safety warning information detailed data table is constructed using the associated data.

[0037] Step 2: Extract the data with the warning category of overtime fatigue in the active safety warning information detailed data table, and generate an overtime fatigue driving record table.

[0038] The excessive fatigue driving record form includes the license plate number, driver's name, fatigue driving start time, fatigue driving end time and duration.

[0039] Step 3: Construct a quantitative evaluation model for driver posture safety;

[0040] Step 4: Using driving videos as data sets, the data sets are divided into training sets and test sets in a ratio of 4:1. The training set is input into the driver posture safety quantitative evaluation model for training. When the training error reaches the set threshold or the learning rate reaches the set upper limit, the training is stopped. Then the test set is input into the driver posture safety quantitative evaluation model to obtain the corresponding evaluation indicators.

[0041] The accuracy of the evaluation indicators needs to be verified. If it does not meet the accuracy requirements, the data set is re-collected and the training steps are repeated until the accuracy requirements are met.

[0042] Step 5: Characterize the driver's reaction degree under different warning types according to the evaluation indicators, cluster the indicators, and obtain the classification results of drivers under different warning types. The warning types include forward collision warning, lane departure warning and distraction fatigue warning.

[0043] Evaluation indicators include:

[0044] Extract frames from the warning video and extract pictures;

[0045] Detect the generated images, extract the text in the images, and generate a detailed warning data table based on the text;

[0046] According to the data in the warning detailed data table, the reaction time, speed change before the warning and speed change after the warning are obtained.

[0047] In this implementation, the driving behavior data is collected in real time through the Ministry of Transport standard GPS equipment and smart video cameras installed on the operating vehicles, and uploaded to the supervision platform. Among them, the Ministry of Transport standard GPS equipment is based on the standards of the Ministry of Transport combined with 796 and various transmission standards; the smart video camera is ADAS, which is mainly used for front vehicle collision warning and lane departure warning, as well as collecting various illegal driving behaviors such as whether the driver smokes, makes phone calls, looks around, speeds, and drives while fatigued, and also collects whether the equipment is online and whether it is operating normally.

[0048] By analyzing the driving behavior data to see if there are any violations, if so, a voice warning message will be automatically sent to remind the driver to drive in accordance with regulations, and a voice warning message will be automatically sent or a phone call will be made to the driver to immediately stop the illegal and irregular behavior. The driver's behavior data will be automatically generated into a record report and sent to the unit to which the vehicle belongs within a preset time. It is determined whether the driver drives in accordance with the warning information. If not, when the driver has not stopped the illegal and irregular behavior after receiving the prompt information, manual intervention will be carried out, and the administrator of the supervision platform will be automatically notified to handle it and the information of the vehicle's illegal behavior will be sent to the unit to which the vehicle belongs and / or the local traffic police management department, forcing the driver to stop the illegal and irregular behavior.

[0049] In addition, when the operating vehicle engages in speeding, fatigue driving, or even the speed suddenly drops to zero, the data node is transmitted back three times in 1 minute from 0 to 60 seconds, that is, the data is transmitted back to the supervision platform once every 20 seconds. When the supervision platform analyzes the data sent back by the departmental standard GPS equipment and smart video camera installed on the vehicle, if the operating vehicle exceeds the speed limit, drives fatigued, or even the speed suddenly drops to zero (that is, when the vehicle is involved in an accident or collision), or if the vehicle moves abnormally or abnormally, the supervision platform will automatically sound an alarm through the external speaker to remind management personnel to pay attention and handle it.

[0050] After collecting and analyzing the data sent back by the department-standard GPS equipment and intelligent video cameras on the vehicles, a set of safety management electronic templates based on national standards are formed, which automatically generate record reports from driving behavior data, including violation handling reports, mileage statistics, vehicle maintenance management reports, and road list matching vehicle operation route reports; as well as various standardized safety documents required by the state, including enterprise quality reputation assessment, a set of tables and 16 safety standardization items, forming a set of standardized and process-based safety management documents for the whole station. From problem alarm, automatic problem recording, online problem processing (including sending voice, making phone calls or issuing TTS text-to-speech notifications, etc.), until the end of the dangerous driving behavior, the generated record report will be sent to the unit to which the vehicle belongs within the preset time, thus forming a closed-loop management. Improve the efficiency and accuracy of enterprise safety management. Through the third-party supervision platform, discrete industry data and information resources can be interconnected and shared, which can strengthen the collaborative supervision capabilities of key road transport operating vehicles, strengthen the ability to control vehicles from the source, strengthen real-time monitoring of vehicles during the supervision process, and strengthen the ability to punish illegal and irregular vehicles and companies afterwards, so as to achieve closed-loop collaborative management of pre-emptive prevention, in-process strong control and post-event punishment.

[0051] Second implementation method: The process of constructing a quantitative evaluation model for driver posture safety is as follows:

[0052] 1) Using driving videos as data sets, we use quantitative evaluation indicators and evaluation posture X i , and conduct correlation analysis on quantitative evaluation indicators. The quantitative evaluation indicators include the emergency response time T r , sight deviation time T d , the time when the hand leaves T l and blink frequency F b The evaluation posture Xi includes normal driving X1, one-handed driving X2, gear manipulation X3, eating and drinking X4, picking up and putting things X5 and playing with mobile phones X6, and the original data is processed by deviation standardization to obtain the original value ε of each index and each posture ij , where i is the evaluation posture number and j is the evaluation index number;

[0053] 2) According to the original value ε of each indicator and each posture ij , the fuzzy consistent judgment matrix R is established by using fuzzy analytic hierarchy process, and the weight value of each quantitative evaluation index is calculated by using the least square method;

[0054] 3) According to the weight value, determine the optimal weight vector W and calculate the risk score Y of each driving posture i , and complete the establishment of a quantitative evaluation model for driver posture safety.

[0055] In this embodiment, SPSS software is used to perform correlation analysis of quantitative evaluation indicators, 1≤j≤4, from small to large, representing quantitative evaluation indicators T r , T d , T l and F b .

[0056] In this embodiment, the action emergency response time T r The time required for the driver to recover from the dangerous posture to the normal posture after the driver realizes the dangerous situation, and the sight deviation time T d is the total time the driver’s sight is away from the road during the posture duration, and the hand-off time T l is the total time the driver’s hands are away from the driving position during the posture duration, and the blink frequency F b is the blinking frequency of the driver during the posture duration. The original data is normalized to make the result fall into the interval [0,1], and the original value ε of each index and each posture is obtained. ij , quantitative cluster analysis was performed.

[0057] In this embodiment, the least square method is used to calculate the weight value of each quantitative evaluation index, that is, to solve the following constraint programming problem:

[0058]

[0059]

[0060] In the formula, min z is the minimum error value, n is the total number of quantitative evaluation indicators, a is the parameter, 0 <a≤0.5,W i is the weight value of the quantitative evaluation index of sequence number i, W j is the weight value of the quantitative evaluation index of sequence number j, r ij is the relative importance membership value of the quantitative evaluation index number i and the quantitative evaluation index number j. The parameter a in the above formula is traversed every 0.04 to 0.06 in the value range to calculate the weight value of each quantitative evaluation index.

[0061] In this embodiment, each driving posture risk score Y i The calculation formula is as follows:

[0062]

[0063] Where W i is the weight value of the quantitative evaluation index.

[0064] Meanwhile, the contents not described in detail in this specification belong to the prior art known to those skilled in the art, and the model parameters of each electrical appliance are not specifically limited, and conventional equipment can be used.

[0065] It should be noted that, in this article, relational terms such as first and second, etc. are only used 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", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including 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.

[0066] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A safety analysis and early warning method for operating drivers based on a character portrait data model, characterized in that: The following steps are involved: Step 1: Collect and store driver personal information data, vehicle parking record data, active safety warning data and warning video record data; Step 2: extract the data with the warning category of overtime fatigue in the active safety warning information detailed data table, and generate an overtime fatigue driving record table; Step 3: Construct a quantitative evaluation model for driver posture safety; Step 4: Using driving videos as data sets, the data sets are divided into training sets and test sets in a ratio of 4:

1. The training set is input into the driver posture safety quantitative evaluation model for training. When the training error reaches the set threshold or the learning rate reaches the set upper limit, the training is stopped. Then the test set is input into the driver posture safety quantitative evaluation model to obtain the corresponding evaluation index. Step 5: Characterize the driver's reaction degree under different warning types according to the evaluation indicators, cluster the indicators, and obtain the classification results of drivers under different warning types.

2. The method for safety analysis and early warning of operating drivers based on a character portrait data model according to claim 1 is characterized in that: In the step 1, the driver's personal information data, vehicle parking record data, active safety warning data and warning video record data are associated to generate a detailed data table of active safety warning information.

3. The method for safety analysis and early warning of operating drivers based on character portrait data model according to claim 2 is characterized in that: The active safety warning information detailed data table is specifically as follows: the active safety warning data is associated with the warning video record data according to the warning number, the active safety warning data is associated with the driver's personal information data according to the driver's name, and the active safety warning data is associated with the vehicle parking record data according to the license plate number, and the active safety warning information detailed data table is constructed using the associated data.

4. The method for safety analysis and early warning of operating drivers based on a character portrait data model according to claim 1, characterized in that: The overtime fatigue driving record table in step 2 includes the license plate number, the driver's name, the fatigue driving start time, the fatigue driving end time and the duration.

5. The method for safety analysis and early warning of operating drivers based on character portrait data model according to claim 1 is characterized in that: The process of constructing the driver posture safety quantitative evaluation model in step 3 is as follows: 1) Using driving videos as data sets, we conduct correlation analysis of quantitative evaluation indicators based on quantitative evaluation indicators and evaluation postures Xi. The quantitative evaluation indicators include emergency response time Tr, sight deviation time Td, hand departure time Tl, and blinking frequency Fb. The evaluation postures Xi include normal driving X1, one-handed driving X2, gear manipulation X3, eating and drinking X4, picking up and putting things X5, and playing with mobile phones X6. The original data is processed by deviation standardization to obtain the original values ​​εij of each indicator and each posture, where i is the evaluation posture number and j is the evaluation indicator number. 2) According to the original values ​​εij of each index and each posture, the fuzzy consistent judgment matrix R is established by using the fuzzy analytic hierarchy process, and the weight value of each quantitative evaluation index is calculated by the least squares method; 3) According to the weight value, determine the optimal weight vector W, calculate the risk score Yi of each driving posture, and complete the establishment of the quantitative evaluation model of driver posture safety.

6. The method for safety analysis and early warning of operating drivers based on character portrait data model according to claim 1, characterized in that: In step 4, the accuracy of the evaluation index needs to be verified. If the accuracy requirement is not met, the data set is re-collected and the training steps are repeated until the accuracy requirement is met.

7. The method for safety analysis and early warning of operating drivers based on character portrait data model according to claim 1 is characterized by: The evaluation indicators include: Extract frames from the warning video and extract pictures; Detect the generated images, extract the text in the images, and generate a detailed warning data table based on the text; According to the data in the warning detailed data table, the reaction time, speed change before the warning and speed change after the warning are obtained.

8. The method for safety analysis and early warning of operating drivers based on character portrait data model according to claim 1, characterized in that: The warning types include forward collision warning, lane departure warning and distraction warning.

Images