Online car-hailing driving behavior intelligent warning method and system

Abstract

The invention belongs to the technical field of automobile driving safety, and particularly relates to an intelligent warning method and system for online car-hailing driving behaviors, and the method comprises the steps: obtaining a plurality of driving records of a plurality of drivers; acquiring nodes of a neighborhood range at any moment in each driving record of each driver; obtaining the fatigue degree of any driver in any driving record based on the vehicle speed of the driver and the distance change between the driver and the nodes in the neighborhood range; establishing a fatigue evolution model based on the fatigue degree change of each driving record of different drivers in different driving periods; based on the fatigue evolution model and the fatigue degree change of each driving record of the driver in the real-time driving period, obtaining a real-time warning index of each driver; the real-time warning indexes of the nodes in the adjacent areas are obtained through the Internet of Vehicles, and self warning and neighborhood warning are carried out on a driver. According to the invention, the accuracy and intelligence of online car-hailing driving behavior warning are improved.

Description

Technical Field

[0001] The present invention relates to the field of automobile driving safety technology. More specifically, the present invention relates to a method and system for intelligently warning online ride-hailing drivers about driving behavior. Background Art

[0002] With the increasing popularity of automobiles and the development of internet technology, online ride-hailing service platforms have emerged, allowing users to quickly hail a ride through mobile apps and enjoy convenient travel. However, as driving time increases, drivers can become fatigued, impacting their decision-making and making timely adjustments to their vehicles, leading to traffic accidents and subsequent chain reactions. Therefore, online ride-hailing service platforms need to monitor drivers' status and issue warnings for dangerous driving behaviors, thereby helping drivers focus and improving safety for both drivers and users.

[0003] In order to detect the driver's fatigue state, the driver's driving state is currently often characterized by vehicle operation data and driver physical condition data. In related technologies, for example, the Chinese patent document with authorization publication number CN107680338B discloses a fatigue driving detection method and device, which discloses the use of steering wheel speed as a representation of the driver's operating state of the vehicle, ensuring the effectiveness of driver fatigue driving detection. The Chinese patent document with authorization publication number CN105096528B discloses a fatigue driving detection method and system, which discloses the use of changes in the driver's facial expressions and body movements over time as a standard for detecting whether the driver is in a fatigue driving state.

[0004] However, different driving environments will affect the driver's state. For example, there will be significant differences in the driver's steering wheel speed, facial expressions and body movements in traffic congested sections and in smooth road sections, which may lead to misjudgment, affect the warning effect on the driver, and then affect the driver's psychology, leading to road traffic safety hazards. Summary of the Invention

[0005] In order to solve the technical problem of poor warning effect of online car-hailing driving behavior, the present invention provides solutions in the following aspects.

[0006] In a first aspect, the present invention provides an intelligent warning method for online car-hailing driving behavior, comprising: Acquire several driving records of several drivers, and divide all driving records of any driver into several driving periods, wherein the driving record is a vehicle speed-time series; acquire nodes in a neighborhood range at any time in each driving record of each driver, wherein the nodes include signal sources such as people and vehicles that can be acquired through the Internet of Vehicles; acquire the fatigue level of any driver in any driving record based on the change in the driver's vehicle speed and the distance between the driver and the nodes in the neighborhood; establish a fatigue evolution model based on the change in fatigue level of each driving record of different drivers in different driving periods; acquire the real-time warning index of each driver based on the fatigue evolution model and the change in fatigue level of each driving record of the driver in the real-time driving period; acquire the real-time warning index of the nodes in the neighborhood range, and warn the driver himself and the neighborhood.

[0007] The fatigue evolution model established by this invention provides a reference for all drivers' fatigue states, enables different driving time limits for different drivers, and enables real-time assessment of the driver's physical condition, improving the accuracy of warnings regarding driver behavior. This invention provides both self-warning and neighboring warnings, ensuring not only individual warnings but also the safety of the transportation network, reducing the risk of drivers entering traffic accidents under unpredictable circumstances and improving the effectiveness of warnings for ride-hailing drivers.

[0008] Preferably, the method of obtaining the fatigue level of any driver in any driving record includes: setting a safe distance and a danger range for the driver; recording the extreme speed point of the driving record as the danger point of the driving record; obtaining the moment when the speed closest to the extreme speed point of the driving record is the average speed, recording it as the key moment, and recording the key moments located to the left and right of the danger point of the driving record as the reaction point and regression point of the driving record; obtaining the judgment ability of each driver in each driving record and the reaction speed of each driver in each driving record based on the change in the driver's speed and the distance between the driver and the nodes in the adjacent area; comparing the reaction speed of each driver in each driving record with the judgment ability, and performing positive correlation normalization to obtain the fatigue level of each driver in each driving record.

[0009] The present invention determines the driver's fatigue level in each driving record based on the driver's vehicle speed and the change in distance from nodes in the neighborhood range, which can provide an effective basis for establishing a fatigue evolution model and can provide accurate data for judging the driver's status.

[0010] Preferably, obtaining the judgment ability of each driver in each driving record includes: Obtain the dangerous response ratio of each driver; ; Where, represents the judgment ability of the i-th driver in the c-th driving record; represents the dangerous reaction ratio of the i-th driver, represents the average proportion of dangerous reactions among all drivers; represents the mean distance between each dangerous point in the cth driving record of the i-th driver and the nearest node at the corresponding moment; represents the mean distance between all dangerous points in all driving records of the i-th driver and the nearest node at the corresponding moment; represents an exponential function with a natural constant as its base; Represents the normalization function.

[0011] The present invention obtains the driver's judgment ability in the driving record by analyzing the accuracy of the driver's reaction when facing neighboring vehicles, effectively analyzes the differences in the driver's own abilities, and makes the warning time for different drivers different, making the warning of online car-hailing driving behavior more intelligent and humane.

[0012] Preferably, obtaining the reaction speed of each driver in each driving record includes: obtaining all dangerous point moments in all driving records of each driver, and recording the proportion of dangerous point moments of the node closest to the driver within the dangerous range of the corresponding driver in all dangerous point moments as the dangerous reaction proportion of each driver.

[0013] Preferably, obtaining the reaction speed of each driver in each driving record includes: Obtain the proportion of insufficient response time of each driver in each driving record; ; Where, represents the reaction speed of the i-th driver in the c-th driving record; represents the time distance set from the reaction point of the i-th driver in the c-th driving record to the corresponding regression point, represents the time distance set from the reaction point to the corresponding regression point in all driving records of the i-th driver; represents the set of under-reaction time proportions of all driving records of the i-th driver, represents the set of under-reaction time proportions for all driving records of all drivers; Represents an exponential function with a natural constant as its base.

[0014] Preferably, obtaining the proportion of underreaction time of each driver in each driving record includes: obtaining the time distance from each reaction point in each driving record of each driver to the corresponding regression point; obtaining the proportion of time in which there are nodes within the danger range of the corresponding driver within the time range from all reaction points in each driving record to the corresponding regression point, which is recorded as the proportion of underreaction time of the corresponding driver in each driving record.

[0015] The present invention obtains the proportion of the driver's insufficient reaction time in each driving record, provides a basis for accurately judging the driver's reaction speed, and improves the accuracy of intelligent warnings.

[0016] Preferably, the establishment of the fatigue evolution model includes: taking the driver corresponding to the driving record with the lowest fatigue level as the benchmark driver, drawing a fatigue evolution coordinate system with time as the horizontal axis and fatigue level as the vertical axis, and drawing a line graph of the fatigue level of all driving records of each driver in each driving period in the fatigue evolution coordinate system, which is recorded as a line graph of each driving period; shifting the line graph of the fatigue level of all driving records in each driving period according to the fatigue level of different drivers in different driving periods, and obtaining the relative positions of all driving records in the fatigue evolution coordinate system; using the least squares method to perform polynomial fitting on the coordinates of the relative positions of all driving records in the fatigue evolution coordinate system, and using the fitting result as the fatigue evolution model.

[0017] Preferably, the method of displacing the line graph of fatigue levels of all driving records of each driving period to obtain the relative positions of all driving records in the fatigue evolution coordinate system includes: displacing the line graph of the second driving period of the benchmark driver as a whole in an area adjacent to the line graph of the first driving period, performing DTW matching with the line graph of the first driving period, and displacing the line graph of the second driving period to a position where the average DTW distance is minimum; and sequentially displacing the line graphs of the remaining driving periods of the benchmark driver and the line graphs of all driving periods of the remaining drivers in areas adjacent to the line graphs of all driving periods after the displacement, to obtain the relative positions of all driving records in the fatigue evolution coordinate system.

[0018] The present invention unifies the fatigue changes of different drivers in different driving periods, so that different drivers can determine their relative fatigue status in a unified fatigue evolution model, thereby avoiding the situation where the warning is not applicable to everyone due to the unified driving time limit.

[0019] Preferably, the real-time alert index of each driver satisfies the expression: ; Where, represents the real-time warning index of the i-th driver; represents the number of driving records of the i-th driver’s real-time driving period; represents the time from the hth driving record of the i-th driver's real-time driving period to the start of the real-time driving period; represents the fatigue level of the hth driving record of the i-th driver's real-time driving period; K represents the fatigue level threshold; Represents the normalization function.

[0020] In a second aspect, the present invention provides an intelligent warning system for online car-hailing driving behavior, comprising a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned intelligent warning method for online car-hailing driving behavior is implemented.

[0021] By adopting the above technical solution, the above-mentioned intelligent warning method for online car-hailing driving behavior is generated into a computer program and stored in a memory to be loaded and executed by a processor, thereby making a terminal device based on the memory and processor for easy use.

[0022] The beneficial effects of the present invention are: (1) The present invention establishes a fatigue evolution model that enables different drivers to determine the warning conditions for fatigue driving based on their own conditions, thereby improving the effectiveness of online ride-hailing driving behavior warnings; (2) The present invention uses the driver's speed as a feature to measure the driver's fatigue level, thus avoiding the problem of insufficient accuracy in judging the driver's fatigue level caused by the variability of features such as the driver's facial expression in existing methods; (3) The present invention uses the Internet of Vehicles to warn the driver himself and the neighbors, thereby improving the overall safety of road traffic and the driver's driving safety. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 1 is a flow chart schematically illustrating an intelligent warning method for online car-hailing driving behavior in the present invention; Figure 2 is a schematic diagram schematically showing a neighborhood node coordinate system of a driver; Figure 3 is a diagram schematically illustrating a driving record; Figure 4 is a schematic diagram schematically showing a line graph of two driving periods; Figure 5 is a displacement diagram schematically showing a line graph of two driving periods. DETAILED DESCRIPTION

[0024] The embodiment of the present invention discloses an intelligent warning method for online car-hailing driving behavior, referring to Figure 1 , including steps S1 to S4: S1: Obtain several driving records of several drivers. All driving records of any driver are divided into several driving periods. The driving record is a speed-time series. Obtain the nodes in the neighborhood range at any time in each driving record of each driver.

[0025] Specifically, several driving records of several drivers in a day are obtained from the online car-hailing service platform, and driving records with an interval less than a preset first threshold are recorded as belonging to the same driving period, so as to obtain several driving periods of any driver. The driving record is the speed-time sequence recorded by the online car-hailing service platform from the time the driver determines that the passenger gets on the car to the time the passenger gets off the car. It should be noted that the preset first threshold is set by the implementer based on the actual implementation situation. Since the mandatory rest time for fatigue driving must not be less than 20 minutes, the first threshold can be set to 20 minutes. When the time when the vehicle speed is equal to 0 between two driving records is greater than or equal to 20 minutes, the two driving records can be divided into two driving periods.

[0026] Set the driver's neighborhood range, obtain the node position within the driver's neighborhood range at any time in any driving record through the Internet of Vehicles, record the driver's driving direction as the positive direction of the Y axis, record the right side perpendicular to the driver's driving direction as the positive direction of the X axis, and use the driver as the coordinate origin to construct the driver's neighborhood node coordinate system, such as Figure 2 This is a schematic diagram of the coordinate system for the driver's neighborhood nodes. These nodes are people and vehicles that can be sensed by various sensing mechanisms, including millimeter-wave radar, cameras, inertial navigation, and vehicle networking. It should be noted that the driver's neighborhood range is set by implementers based on actual implementation circumstances; for example, it can be set to 50 meters.

[0027] At this point, several driving records of several drivers and the coordinate system of adjacent nodes at any time in each driving record of each driver are obtained.

[0028] S2: Based on the driver's speed and the change in the distance between the driver and the nodes in the adjacent range, obtain the fatigue level of any driver in any driving record.

[0029] It should be noted that when a driver is fatigued, his judgment ability will decline and his reaction will be slow. Therefore, when encountering an emergency, he will operate slowly and make mistakes. Therefore, the driver's fatigue level can be determined based on the driver's reaction speed and judgment ability when encountering an emergency.

[0030] It's important to further clarify that the first step is to identify an emergency. For example, if a nearby person or vehicle suddenly swerves or changes speed, the driver often needs to brake to maintain a safe distance, which manifests as a rapid change in speed. Judgment ability refers to the driver's accuracy in assessing the emergency—in other words, whether braking or other maneuvers successfully avoid the risk of a traffic accident and return to a safe distance from the person or vehicle causing the emergency. Reaction speed measures the time it takes from the appearance of an emergency person or vehicle in the driver's immediate vicinity to the return to a safe distance. Therefore, the driver's fatigue level for any given driving record can be determined based on the driver's speed and the distances to nearby people and vehicles acquired through the Internet of Vehicles.

[0031] Specifically, a safe distance is set for the driver, and a circular area with the safe distance as its radius is recorded as the driver's danger zone. It should be noted that the danger zone is the distance that ensures the driver has sufficient reaction time to make decisions and avoid traffic accidents. The safe distance can be set as the real-time vehicle speed multiplied by 5 seconds, and the driver's safe distance must be no less than 2 meters.

[0032] Preferably, the extreme point of the vehicle speed in the driving record is recorded as the dangerous point of the driving record; the moment when the vehicle speed is closest to the extreme point of the vehicle speed in the driving record as the average vehicle speed is recorded as the key moment, and the key moments to the left and right of the dangerous point of the driving record are recorded as the reaction point and regression point of the driving record. Figure 3 This is a schematic diagram of a driving record.

[0033] It should be noted that when a driver's judgment ability is high, they can proactively avoid people or vehicles that might enter the safe distance. Therefore, fewer people or vehicles may enter the safe distance during driving. Therefore, by comparing the number of nodes within the driver's danger zone at the time of the danger point in the driving records of different drivers, the driver's judgment ability can be determined. The farther the distance between the driver and the adjacent node at the time of the danger point, the stronger the driver's judgment ability. Conversely, the closer the distance between the driver and the adjacent node at the time of the danger point, the more likely a collision will occur, and the target driver's judgment ability is weaker.

[0034] Preferably, the judgment ability of each driver in each driving record is obtained based on the frequency of the node entering the safe distance at the dangerous point in the driver's driving record and the distance between the driver and the node entering the safe distance: All dangerous point moments in all driving records of each driver are obtained, and the proportion of dangerous point moments with the node closest to the driver within the corresponding driver's dangerous range in all dangerous point moments is recorded as the dangerous response proportion of each driver.

[0035] The judgment ability of the i-th driver in the c-th driving record satisfies the expression: ; Where, represents the judgment ability of the i-th driver in the c-th driving record; represents the dangerous reaction ratio of the i-th driver, represents the average proportion of dangerous reactions among all drivers; represents the mean distance between each dangerous point in the cth driving record of the i-th driver and the nearest node at the corresponding moment; represents the mean distance between all dangerous points in all driving records of the i-th driver and the nearest node at the corresponding moment; represents an exponential function with a natural constant as its base; Represents the normalization function.

[0036] Where, It indicates the relative judgment ability of the i-th driver relative to all drivers. The larger the value, the greater the proportion of dangerous reactions of the i-th driver, and the lower the relative judgment ability of the i-th driver. It represents the relative judgment ability of the i-th driver in the c-th driving record relative to all the i-th driver's driving records. The larger the value, the greater the distance between the i-th driver in the c-th driving record and the nearest node, and the i-th driver's judgment ability in the c-th driving record is relatively better. Therefore, the i-th driver has a stronger judgment ability in the c-th driving record.

[0037] Preferably, the reaction speed of each driver in each driving record is obtained based on the time distance between the danger point, reaction point, and regression point of the driver's driving record, as well as the time when the neighboring nodes of the driver's driving record are within the driver's danger range: Obtain the time distance from each reaction point to the corresponding regression point in each driving record of each driver; obtain the time proportion in which nodes exist within the corresponding driver's danger range within the time range from all reaction points in each driving record to the corresponding regression point, and record it as the time proportion of insufficient reaction of the corresponding driver in each driving record.

[0038] The reaction speed of the i-th driver in the c-th driving record satisfies the expression: ; Where, represents the reaction speed of the i-th driver in the c-th driving record; represents the time distance set from the reaction point of the i-th driver in the c-th driving record to the corresponding regression point, represents the time distance set from the reaction point to the corresponding regression point in all driving records of the i-th driver; represents the set of under-reaction time proportions of all driving records of the i-th driver, represents the set of under-reaction time proportions for all driving records of all drivers; Represents an exponential function with a natural constant as its base.

[0039] Where, It represents the ratio of the time distance from the reaction point to the corresponding regression point in the cth driving record of the i-th driver to the time distance from the reaction point to the corresponding regression point in all driving records. This value represents the relative reaction time of the i-th driver's cth driving record relative to all driving records. The larger the value, the longer the i-th driver's reaction time is relative to his own in the cth driving record, and the slower the i-th driver's reaction speed in the cth driving record. It represents the relative reaction speed of the i-th driver relative to all drivers. The larger the value, the larger the proportion of the i-th driver's insufficient reaction time. Therefore, the slower the i-th driver's relative reaction speed relative to all drivers, the slower the i-th driver's reaction speed in the c-th driving record.

[0040] It should be noted that the stronger the driver's judgment ability and the faster the reaction speed in the driving record, the lower the driver's fatigue level in the driving record.

[0041] Preferably, the fatigue level of the i-th driver in the c-th driving record satisfies the expression: ; Where, represents the fatigue level of the i-th driver in the c-th driving record; represents the reaction speed of the i-th driver in the c-th driving record; represents the judgment ability of the i-th driver in the c-th driving record; Represents the normalization function.

[0042] At this point, the fatigue level of each driver in each driving record is obtained.

[0043] S3: Based on the fatigue level changes of different drivers in the driving records of different driving periods, a fatigue evolution model is established; based on the fatigue evolution model and the fatigue level changes of the drivers in the real-time driving period, a real-time warning index of each driver is obtained.

[0044] It should be noted that different drivers have different physical conditions, which can lead to different changes in fatigue levels. For example, drivers with higher physical fitness and reactions will have lower initial fatigue levels, and their fatigue levels will change more slowly as their driving records increase. Therefore, if the fatigue levels of all drivers' driving records are regressed and fitted according to time, it will not be applicable to everyone. Considering that drivers with slower reactions can be compared to drivers with faster reactions who have been driving for a period of time, a unified fatigue evolution model can be established, and different drivers are at different stages in the fatigue evolution model. At the same time, considering that road safety will not change due to the driver's own physical condition, the threshold for reaching the fatigue level that requires warning should be consistent for all drivers. This means that different drivers can drive continuously for different periods of time. Therefore, the warning unit also needs to adaptively issue warnings based on the changes in fatigue levels in the driver's driving record.

[0045] Preferably, a fatigue evolution model is established based on the fatigue level changes of different drivers in different driving records at different driving periods: The driver corresponding to the driving record with the least fatigue is taken as the benchmark driver. A fatigue evolution coordinate system is drawn with time as the horizontal axis and fatigue level as the vertical axis. A line graph of the fatigue level of all driving records of each driver in each driving period is drawn in the fatigue evolution coordinate system, which is recorded as the line graph of each driving period. The line graph of the second driving period of the benchmark driver is displaced as a whole in the area adjacent to the line graph of the first driving period, and DTW matching is performed with the line graph of the first driving period. The line graph of the second driving period is displaced to the position with the minimum average DTW distance. The line graphs of the remaining driving periods of the benchmark driver and the line graphs of all driving periods of the remaining drivers are displaced in the area adjacent to the line graphs of all driving periods after the displacement, and the relative positions of all driving records in the fatigue evolution coordinate system are obtained. Figure 4 is a schematic diagram of a line graph for two driving periods. Figure 5 Schematic diagram of the displacement of the line graph for two driving periods.

[0046] The least squares method is used to perform a polynomial fit on the relative positions of all driving records within the fatigue evolution coordinate system, and the fitting result is used as the fatigue evolution model. It should be noted that while the driver is driving, obtaining all driving records for the current driving period can determine the driver's current relative fatigue level in the fatigue evolution model. When a driver is in poor physical condition, even if they have only driven for an hour, they may appear to be in a relatively fatigued state in the fatigue evolution model. Therefore, by establishing a unified fatigue evolution model, warning times can be adaptively determined for different drivers.

[0047] It should be noted that due to the complexity of road traffic, the driver's driving abnormalities are not only related to the driver himself. For example, if the driver brakes suddenly due to a ghosting, the driver's driving record may show a high level of fatigue, but it has not reached the fatigue level that should be warned. Therefore, it is necessary to combine the driver's entire driving record during the current driving period to determine the driver's real-time warning index.

[0048] Preferably, based on the fatigue evolution model and the fatigue level changes of each driving record during the driver's real-time driving period, the real-time warning index of each driver is obtained: In the fatigue evolution model, the fatigue levels of all driving records during the i-th driver's real-time driving period are annotated, and a fatigue threshold K is set. It should be noted that K is set by the implementer based on actual implementation circumstances. For example, K can be set to the fatigue level of the benchmark driver's driving records corresponding to the first four hours of driving during the first driving period.

[0049] The real-time alert index of the i-th driver satisfies the expression: ; Where, represents the real-time warning index of the i-th driver; represents the number of driving records of the i-th driver’s real-time driving period; represents the time from the hth driving record of the i-th driver's real-time driving period to the start of the real-time driving period; represents the fatigue level of the hth driving record of the i-th driver's real-time driving period; Represents the normalization function.

[0050] Where, represents the normalization of the relative driving duration of the hth driving record of the i-th driver's real-time driving period; represents the difference between the fatigue level of the hth driving record of the i-th driver's real-time driving period and the fatigue level threshold K; It means that the relative driving time of the driving record is used as the weight, and the difference between the fatigue level and fatigue level threshold of all driving records in the real-time driving period is weighted and averaged. The larger the value is, the higher the fatigue level of the driving record closer to the current one is, and the higher the driver's real-time warning index is.

[0051] At this point, the fatigue evolution model and the real-time warning index of any driver are obtained.

[0052] S4: Obtain the real-time warning index of the nodes in the neighborhood range and issue self-warning and neighborhood warning to the driver.

[0053] It should be noted that when a person engages in dangerous driving behavior, it often involves not only the individual and the car, but also pedestrians, other vehicles, and public facilities. The development of the Internet of Vehicles allows the flow of vehicle information and communication equipment information. When a neighboring vehicle exhibits abnormal driving behavior or the real-time warning index is high, the driver will be warned, which can effectively avoid the chain reaction scope of traffic accidents.

[0054] Specifically, the real-time warning index of the driver and all nodes in the driver's neighborhood is obtained, and the first-level warning threshold and the second-level warning threshold are set. When the driver's real-time warning index exceeds the first-level warning threshold, the driver is given a sound warning, a vibration warning through sound, the steering wheel, and the seat; when the driver's real-time warning index exceeds the second-level warning threshold, the driver is warned to prohibit continuing to accept orders. When the real-time warning index of the nodes in the driver's neighborhood exceeds the first-level warning threshold, the driver is warned to drive carefully and pay attention to the surroundings through sound; when the real-time warning index of the nodes in the driver's neighborhood exceeds the second-level warning threshold, the driver is warned through sound and the display screen that the neighboring nodes are behaving abnormally. It should be noted that the first-level warning threshold and the second-level warning threshold are set by the implementer according to the actual implementation situation. For example, the first-level warning threshold can be set to 0.5, and the first-level warning threshold can be set to 0.7.

[0055] At this point, the intelligent warning for online car-hailing driving behavior has been completed.

[0056] An embodiment of the present invention also discloses an intelligent warning system for online car-hailing driving behavior, including a processor and a memory, wherein the memory stores computer program instructions. When the computer program instructions are executed by the processor, an intelligent warning method for online car-hailing driving behavior according to the present invention is implemented.

[0057] The above system also includes other components well known to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.

[0058] While several embodiments of the present invention have been shown and described herein, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Numerous modifications, variations, and alternatives will occur to those skilled in the art without departing from the concept and spirit of the present invention. It should be understood that various alternatives to the embodiments of the present invention described herein may be employed in practicing the present invention.

Claims

1. An intelligent warning method for online car-hailing driving behavior, characterized in that: include: Acquire multiple driving records of multiple drivers, where all driving records of any driver are divided into multiple driving periods, and the driving records are a speed-time series; obtain nodes in a neighborhood range at any time in each driving record of each driver; Based on the driver's speed and the distance between the driver and the nodes in the neighborhood, the fatigue level of any driver in any driving record is obtained; A fatigue evolution model is established based on the fatigue level changes of different drivers in different driving records at different driving times; Based on the fatigue evolution model and the fatigue level changes of each driving record during the driver's real-time driving period, the real-time warning index of each driver is obtained; Obtain the real-time warning index of nodes in the neighborhood range to warn the driver himself and the neighborhood.

2. The intelligent warning method for online car-hailing driving behavior according to claim 1 is characterized in that: The step of obtaining the fatigue level of any driver in any driving record includes: Set the driver's safe distance and danger range; record the extreme speed point of the driving record as the danger point of the driving record; obtain the moment when the speed closest to the extreme speed point of the driving record is the average speed, record it as the key moment, and record the key moment located to the left and right of the danger point of the driving record as the reaction point and regression point of the driving record; obtain the judgment ability of each driver in each driving record and the reaction speed of each driver in each driving record based on the change in the driver's speed and the distance between the driver and the nodes in the adjacent area; compare the reaction speed of each driver in each driving record with the judgment ability, and perform positive correlation normalization to obtain the fatigue level of each driver in each driving record.

3. The intelligent warning method for online car-hailing driving behavior according to claim 2 is characterized in that: The obtaining of the judgment ability of each driver in each driving record includes: Obtain the dangerous response ratio of each driver; ； Where, represents the judgment ability of the i-th driver in the c-th driving record; represents the dangerous reaction ratio of the i-th driver, represents the average proportion of dangerous reactions among all drivers; represents the mean distance between each dangerous point in the cth driving record of the i-th driver and the nearest node at the corresponding moment; represents the mean distance between all dangerous points in all driving records of the i-th driver and the nearest node at the corresponding moment; represents an exponential function with a natural constant as its base; Represents the normalization function.

4. The intelligent warning method for online car-hailing driving behavior according to claim 3 is characterized in that: The obtaining of the reaction speed of each driver in each driving record includes: All dangerous point moments in all driving records of each driver are obtained, and the proportion of dangerous point moments with the node closest to the driver within the corresponding driver's dangerous range in all dangerous point moments is recorded as the dangerous response proportion of each driver.

5. The intelligent warning method for online car-hailing driving behavior according to claim 2 is characterized in that: The obtaining of the reaction speed of each driver in each driving record includes: Obtain the proportion of insufficient response time of each driver in each driving record; ； Where, represents the reaction speed of the i-th driver in the c-th driving record; represents the time distance set from the reaction point of the i-th driver in the c-th driving record to the corresponding regression point, represents the time distance set from the reaction point to the corresponding regression point in all driving records of the i-th driver; represents the set of under-reaction time proportions of all driving records of the i-th driver, represents the set of under-reaction time proportions for all driving records of all drivers; Represents an exponential function with a natural constant as its base.

6. The intelligent warning method for online car-hailing driving behavior according to claim 5 is characterized in that: The method of obtaining the proportion of insufficient response time of each driver in each driving record includes: Obtain the time distance from each reaction point to the corresponding regression point in each driving record of each driver; obtain the time proportion in which nodes exist within the corresponding driver's danger range within the time range from all reaction points in each driving record to the corresponding regression point, and record it as the time proportion of insufficient reaction of the corresponding driver in each driving record.

7. The intelligent warning method for online car-hailing driving behavior according to claim 1 is characterized in that: The establishment of the fatigue evolution model comprises: The driver corresponding to the driving record with the lowest fatigue level is taken as the benchmark driver. A fatigue evolution coordinate system is drawn with time as the horizontal axis and fatigue level as the vertical axis. A line graph of the fatigue level of all driving records of each driver in each driving period is drawn in the fatigue evolution coordinate system, which is recorded as the line graph of each driving period. According to the fatigue level of different drivers in different driving periods, the line graph of the fatigue level of all driving records in each driving period is shifted to obtain the relative positions of all driving records in the fatigue evolution coordinate system. The coordinates of the relative positions of all driving records in the fatigue evolution coordinate system are polynomially fitted using the least squares method, and the fitting results are used as the fatigue evolution model.

8. The intelligent warning method for online car-hailing driving behavior according to claim 7 is characterized in that: The step of shifting the fatigue level line graph of all driving records in each driving period to obtain the relative positions of all driving records in the fatigue evolution coordinate system includes: The line graph of the second driving period of the benchmark driver is displaced as a whole in the area adjacent to the line graph of the first driving period, and DTW matching is performed with the line graph of the first driving period. The line graph of the second driving period is displaced to the position with the minimum average DTW distance. The line graphs of the remaining driving periods of the benchmark driver and the line graphs of all driving periods of the remaining drivers are displaced in the area adjacent to the line graphs of all driving periods after the displacement, and the relative positions of all driving records in the fatigue evolution coordinate system are obtained.

9. The intelligent warning method for online car-hailing driving behavior according to claim 1 is characterized in that: The real-time alert index of each driver satisfies the expression: ； Where, represents the real-time warning index of the i-th driver; represents the number of driving records of the i-th driver’s real-time driving period; represents the time from the hth driving record of the i-th driver's real-time driving period to the start of the real-time driving period; represents the fatigue level of the hth driving record of the i-th driver's real-time driving period; K represents the fatigue level threshold; Represents the normalization function.

10. An intelligent warning system for online car-hailing driving behavior, characterized in that: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, an intelligent warning method for online car-hailing driving behavior according to any one of claims 1 to 9 is implemented.

Images