A driver physiological fatigue driving monitoring system based on behavior data collection

Abstract

The present application belongs to the technical field of driving fatigue monitoring, and particularly relates to a driver physiological fatigue driving monitoring system based on behavior data collection. The method fixes a multi-angle collection mechanism on a steering wheel steering column cover through a mounting mechanism, and collects driver facial behavior data, vehicle forward driving data, vehicle control data, vehicle position data and night fatigue risk data after the vehicle is started. The processor filters and performs feature engineering on the facial image, generates features such as squinting, blinking times, lowering head, yawning and the like, and forms a first fatigue judgment result by using a logistic regression, clustering or K-means model; generates features such as swing, deflection, deviation and uncorrected deviation from the vehicle control data, and forms a vehicle driving state judgment result by using an anomaly detection, random forest or DBSCAN model; simultaneously judges the physiological fatigue risk in combination with the continuous motion duration and the night time slice statistical result, and alarms and stores data when abnormal.

Description

Technical Field

[0001] This invention belongs to the field of driver fatigue monitoring technology, specifically relating to a driver physiological fatigue monitoring system based on behavioral data collection. Background Technology

[0002] Driver fatigue monitoring technology falls under the field of vehicle active safety and driver state monitoring. It is primarily used to monitor changes in the driver's state during vehicle operation and to alert the driver when there is a risk of fatigue, thereby reducing driving safety hazards caused by decreased driver attention, slow reaction time, or prolonged continuous driving. With the increasing use of long-distance transportation, night driving, vehicles engaged in continuous operations, and certain special-purpose vehicles, driver fatigue monitoring has become an important component of vehicle safety assistance systems.

[0003] Current methods for monitoring driver fatigue typically assess the risk of driver fatigue by collecting information on the driver's facial expression, vehicle trajectory, or driving time. For example, some methods rely on changes in facial features such as closing eyes, blinking, looking down, and yawning to determine fatigue levels, while others depend on vehicle trajectory, directional deviations, continuous driving time, or abnormal driving behavior. While these methods can reflect the driver's state to some extent, in real-world driving environments, the driver's facial expression is easily affected by vehicle vibrations, glare from strong lights, windshield reflections, and changes in camera angle. Furthermore, the vehicle's driving behavior is easily influenced by road conditions, turning, lane changes, and other normal driving actions. Therefore, relying on a single criterion makes it difficult to reliably and accurately reflect the driver's true physiological fatigue level.

[0004] Most existing methods for monitoring driver fatigue focus on real-time status recognition after the driver has taken to the road. This typically involves assessing driver fatigue based on facial features, vehicle trajectory, or continuous driving duration. However, these methods struggle to provide timely advance risk assessments for drivers who haven't had sufficient rest at night, have traveled long distances overnight, or have been to multiple locations at night, and then continue driving the following day. Such drivers may not initially exhibit obvious facial fatigue characteristics or abnormal driving trajectories, but they already pose a physiological risk of fatigue. Relying solely on real-time facial recognition or continuous driving time tracking can easily lead to delayed warnings.

[0005] In some special vehicles or vehicles operating continuously, there may be situations where the vehicle is parked but the engine is not turned off. If the sole criterion for judging whether the driver is resting is whether the vehicle is off, it is easy to mistakenly assume that the driver is still driving even when the vehicle is stationary, thus affecting the accuracy of judging continuous driving time. Furthermore, vehicle-mounted fatigue monitoring equipment usually needs to be installed in front of the driver's face. If the data collection device rotates with the steering wheel, has insufficient installation stability, or the wiring harness shakes during driving vibrations, it will also affect the stability of the driver's facial behavior data collection.

[0006] Existing driver fatigue monitoring technologies still suffer from several problems, including insufficient criteria for judging drivers' physiological fatigue, difficulty in identifying the risk of insufficient rest at night, susceptibility to misjudgment or omission in complex driving scenarios, and inadequate stability of data collection equipment. As a result, the reliability of driver fatigue monitoring results still needs to be improved. Summary of the Invention

[0007] To address the problems existing in the prior art, the present invention aims to provide a driver physiological fatigue driving monitoring system based on behavioral data collection, which can solve the above-mentioned problems.

[0008] To achieve the above objectives, the present invention provides the following technical solution: A driver physiological fatigue monitoring system based on behavioral data acquisition is implemented using a driver physiological fatigue monitoring system. The driver physiological fatigue monitoring system includes an installation mechanism, a shock absorption and fixing mechanism, a multi-angle acquisition mechanism, a wiring harness anti-sway mechanism, a processor, a locator, a driving recorder, a vehicle operation data acquisition module, and a data interaction module. The monitoring system includes the following steps: S1. Fix the mounting mechanism to the steering column cover of the vehicle steering wheel, so that the multi-angle acquisition mechanism is installed at the steering column cover and does not rotate with the steering wheel. S2. After the vehicle starts, the control box supplies power to the system and performs a self-test. After the system self-test is normal, it enters the driver fatigue driving monitoring state. S3. Adjust the acquisition orientation of the multi-angle acquisition mechanism according to the driver's body shape, and lock and fix the adjusted acquisition angle through the shock absorption and fixing mechanism so that the multi-angle acquisition mechanism can face the driver's face to collect data. S4. Real-time video of the driver's face is captured by a multi-angle acquisition device to form driver's facial behavior data. The driver's facial behavior data is then transmitted to the processor. The processor filters, extracts features, and identifies fatigue in the driver's facial behavior data to form the first fatigue judgment result. S5. The vehicle forward driving data collected by the dashcam is transmitted to the processor, and the vehicle control data collected by the vehicle control data acquisition module is transmitted to the processor. The processor determines whether the vehicle is driving in a straight line, whether there is a directional deviation, whether there is an S-shaped driving, and whether there is a control abnormality based on the vehicle forward driving data and vehicle control data, and forms a vehicle driving status judgment result. It also filters out deviations caused by turning or changing lanes based on the actual road conditions. S6. The vehicle location data collected by the locator is transmitted to the processor. The processor determines whether the vehicle is in motion based on the vehicle location data, and calculates the driver's continuous driving time based on the duration of the vehicle being in motion, thus forming a second fatigue judgment result. S7. The processor obtains nighttime fatigue risk data through the data interaction module. The nighttime fatigue risk data includes vehicle location data, driver identification data, vehicle license plate collection data and / or roadside video collection data during the nighttime period. The processor statistically analyzes the nighttime fatigue risk data according to a preset time slice to form a nighttime insufficient rest risk judgment result. S8. The processor combines the first fatigue assessment result, the vehicle driving status assessment result, the second fatigue assessment result, and the nighttime insufficient rest risk assessment result to determine whether the driver is in a state of fatigued driving or has a physiological fatigue risk. When the first fatigue assessment result shows that the driver's facial behavior data indicates fatigue, or the second fatigue assessment result shows that the vehicle has been in continuous motion for 4 hours, or the nighttime insufficient rest risk assessment result shows that the driver has insufficient rest at night and continues to drive the motor vehicle during the daytime period after the nighttime period ends, the processor determines that the driver is in a state of fatigued driving or has a physiological fatigue risk. When the vehicle driving status assessment result shows that the vehicle deviates from its direction, drives in an S-shape, or has abnormal handling when not turning or changing lanes, the processor uses the vehicle driving status assessment result as an auxiliary basis for judging abnormal driving. S9. When the processor determines that the driver is in a state of fatigue or there is a risk of physiological fatigue, the control box controls the buzzer to sound an alarm; after the vehicle has finished driving, the data generated during the monitoring process is stored.

[0009] Furthermore, in step S4, the process of forming the first fatigue judgment result includes: continuously acquiring driver facial images through a multi-angle acquisition mechanism; transmitting the acquired driver facial images as driver facial behavior data to a processor; filtering the driver facial behavior data to reduce the impact of vehicle vibration, strong light glare, and windshield reflection on the image data; performing feature engineering processing on the filtered driver facial behavior data to generate one or more of the following features: squinting feature, blinking frequency feature, head-down feature, and yawning feature; inputting the generated facial features into a pre-trained fatigue recognition model, and determining whether the driver facial behavior data corresponds to a fatigue state based on the output of the pre-trained fatigue recognition model; when the driver facial behavior data corresponds to a fatigue state, forming a first fatigue judgment result indicating that the driver is fatigued.

[0010] Furthermore, the pre-trained fatigue recognition model includes one or more of the following: logistic regression model, clustering model, and K-means model; when the fatigue probability output by the logistic regression model reaches a preset threshold, or when the clustering model or K-means model classifies the facial features corresponding to the driver's facial behavior data into the fatigue state category, the processor generates a first fatigue judgment result indicating that the driver is fatigued.

[0011] Furthermore, in step S5, the process of forming the vehicle driving status judgment result includes: collecting the driving scene in front of the vehicle through the driving recorder; the processor judging whether the vehicle is driving in a straight line based on the driving scene in front of the vehicle; when the vehicle deviates from its direction or drives in an S-shape, the processor further judges whether the deviation from its direction or the S-shape is caused by turning or changing lanes in combination with the actual road conditions; when the deviation from its direction or the S-shape is not caused by turning or changing lanes, the processor forms a driving abnormality judgment result and uses the driving abnormality judgment result as an auxiliary basis for the comprehensive judgment of fatigue driving.

[0012] Furthermore, in step S5, the vehicle handling data includes one or more of the following: steering wheel rotation torque data, steering wheel angle data, and vehicle lateral offset data; the processor performs feature engineering processing on the vehicle handling data to generate one or more of the following: hand-off swing feature, large deflection feature, random swing feature, and uncorrected deviation feature, and inputs the generated vehicle handling features into a pre-trained vehicle anomaly recognition model, using the pre-trained vehicle anomaly recognition model to detect whether there are anomalies in the vehicle handling data in real time, forming a vehicle driving state judgment result; the pre-trained vehicle anomaly recognition model includes one or more of the following: anomaly detection model, random forest model, and DBSCAN model.

[0013] Furthermore, in step S6, the process of forming the second fatigue judgment result includes: collecting vehicle position data in real time through the locator; the processor determining whether the vehicle is in motion based on the vehicle position data; when the vehicle is continuously in motion, the processor accumulates the continuous motion time of the vehicle; when the vehicle stops moving, the processor uses the cessation of vehicle movement as the basis for the start of rest; when the continuous motion time of the vehicle reaches 4 hours, the processor forms a second fatigue judgment result indicating that the driver is in a state of fatigued driving.

[0014] Furthermore, when calculating the continuous movement time of the vehicle, the processor uses whether the vehicle has stopped moving as the rest judgment condition, rather than whether the vehicle has turned off, so that the monitoring system can be applied to vehicles that cannot be turned off; when the vehicle has stopped moving but has not turned off, the processor takes the moment when the vehicle stops moving as the rest start time; when the vehicle enters the moving state again, the processor recalculates the continuous movement time of the vehicle.

[0015] Furthermore, in step S7, the nighttime period is from 0:00 to 5:00, and the preset time slice is a series of consecutive time periods, each time slice lasting 20 minutes. Within each time slice, the processor counts one or more of the following: vehicle travel distance, number of administrative regions the vehicle passes through, number of driver face scans, number of vehicle license plate scans, and number of scan locations. When the vehicle travels 200 kilometers during the nighttime period, and the vehicle's location crosses multiple administrative regions, and the driver's identity verification result shows the same driver, the processor generates a nighttime sleep deprivation risk assessment result. Alternatively, when the driver's face is scanned multiple times and at multiple locations during the nighttime period, the processor generates a nighttime sleep deprivation risk assessment result. Or, when the vehicle license plate is scanned multiple times and at multiple locations during the nighttime period, and the driver's identity verification result shows the same driver, the processor generates a nighttime sleep deprivation risk assessment result.

[0016] Furthermore, in steps S3 and S4, the multi-angle acquisition mechanism includes a monitoring camera, a ball head, an infrared polarizer, and a data cable; the ball head is fixed to the bottom of the monitoring camera; when adjusting the acquisition orientation, the ball head rotates within the shock absorption and fixing mechanism so that the monitoring camera faces the driver's face; the infrared polarizer is installed on the front side of the monitoring camera to reduce the influence of direct sunlight, oncoming headlights, and reflected light from the windshield on the acquisition of driver's facial behavior data; the data cable connects the monitoring camera and the control box for power supply and data transmission.

[0017] Furthermore, the shock absorption and fixing mechanism includes a damping spring, a mounting base, a ball groove, a compression groove, and a locking screw; three damping springs are evenly fixed to the bottom of the fixing cylinder; the mounting base is installed on top of the three damping springs; the ball groove is opened on the top of the mounting base and is adapted to the ball head; multiple compression grooves are evenly opened on the top of the mounting base; the locking screw is screwed onto the surface of the mounting base; when the acquisition orientation of the monitoring camera is adjusted, the locking screw is screwed up and moves upward to compress the top of the mounting base, causing the ball groove to contract and clamp the ball head, thereby keeping the acquisition angle of the monitoring camera fixed.

[0018] Furthermore, the mounting mechanism includes an arc-shaped base plate, a high-damping silicone pad, a rigid seat, a fixing cylinder, a control box, and a buzzer; the arc-shaped base plate is adapted to the curved surface of the steering column cover and is bonded and fixed to the steering column cover; the high-damping silicone pad is glued to the top of the arc-shaped base plate; the rigid seat is glued to the top of the high-damping silicone pad; the fixing cylinder is located at the top center of the rigid seat; the control box is fixed to the rear side of the fixing cylinder; and the buzzer is fixed to one side of the control box.

[0019] Furthermore, the anti-sway mechanism for the wiring harness includes a fixed rod, a slide block, a U-shaped groove, a rotating wheel, an arc-shaped groove, and a compression spring; two fixed rods are vertically fixed to the upper surface of the control box and symmetrically arranged; the slide block is slidably installed between the two fixed rods; the U-shaped groove is opened at the center of the upper surface of the slide block; the rotating wheel is rotatably installed in the U-shaped groove; the arc-shaped groove is opened on the surface of the rotating wheel and is adapted to the data cable; the compression spring is sleeved on the lower surface of the fixed rod and applies an upward thrust to the slide block, so that the rotating wheel applies an upward elastic support force to the data cable through the arc-shaped groove, thereby keeping the data cable taut during the monitoring of camera angle adjustment and vehicle driving vibration.

[0020] Compared with the prior art, the beneficial effects of the present invention are: This invention simultaneously collects driver facial behavior data, vehicle forward driving data, vehicle handling data, vehicle position data, and nighttime fatigue risk data. The processor then generates a first fatigue assessment result, a vehicle driving status assessment result, a second fatigue assessment result, and a nighttime insufficient rest risk assessment result. This allows the assessment of driver physiological fatigue risk to no longer rely on a single data source. In situations where individual assessments of driver facial status, vehicle driving status, or continuous driving time are easily affected by environmental and road factors, this invention can utilize multiple types of behavioral data for comprehensive judgment, improving the comprehensiveness and reliability of fatigue driving monitoring results.

[0021] In the process of judging facial behavior data, this invention filters, performs feature engineering on, and pre-trains a fatigue recognition model to generate facial features such as squinting, blinking frequency, head tilting, and yawning. The first fatigue judgment result is formed based on the model output. At the same time, the multi-angle acquisition mechanism, together with the shock absorption and fixing mechanism, infrared polarizer and wiring harness anti-sway mechanism, can reduce the impact of vehicle vibration, strong light glare, windshield reflection and wiring harness shaking on image acquisition, thereby improving the stability of driver facial behavior data acquisition and recognition.

[0022] In the process of judging the vehicle driving status, this invention collects forward driving data through a driving recorder and collects one or more of the following through a vehicle control data acquisition module: steering wheel rotation torque data, steering wheel angle data, and vehicle lateral deviation data. The processor further generates vehicle control characteristics such as hand-off swing, large deflection, random swing, and uncorrected deviation, and filters out normal deviations caused by turning or lane changing in combination with actual road conditions. Thus, abnormal directional deviations, S-shaped driving, or abnormal control can be used as auxiliary judgment criteria to reduce the risk of misjudging normal driving operations as fatigue abnormalities.

[0023] In the process of determining continuous driving time, this invention determines whether the vehicle is in motion based on vehicle position data, and uses whether the vehicle stops moving as the rest judgment condition, rather than whether the vehicle is turned off. For vehicles that cannot be turned off or need to be kept running for a long time, it can still determine whether the driver is continuously driving based on the actual movement of the vehicle, thereby improving the applicability and accuracy of continuous driving time determination.

[0024] This invention acquires nighttime fatigue risk data through a data interaction module and statistically analyzes information such as vehicle travel distance, cross-administrative region situations, number of driver face scans, number of vehicle license plate scans, and number of scan locations during nighttime periods according to preset time slices. When a driver engages in long-distance, large-scale movement at night, frequent and multi-location data collection, or continues to drive during the day after insufficient nighttime rest, the processor can generate a nighttime rest deficiency risk assessment result and use it as a basis for judging physiological fatigue risk, thereby overcoming the problem that existing technologies only identify fatigue in real time during driving and are unable to pre-judge the risk of nighttime rest deficiency. Attached Figure Description

[0025] Figure 1 This is a front view schematic diagram of the installation structure of the present invention; Figure 2 This is a schematic diagram of the installation three-dimensional structure of the present invention; Figure 3 This is a three-dimensional structural diagram of the mounting mechanism and camera of the present invention; Figure 4 For the present invention Figure 3 A schematic diagram of the cross-sectional structure; Figure 5 This is a three-dimensional structural diagram of the installation mechanism of the present invention; Figure 6 This is a three-dimensional structural diagram of the mounting base of the present invention; Figure 7 This is a three-dimensional structural diagram of the camera of the present invention; Figure 8 This is a three-dimensional structural diagram of the wire harness anti-sway mechanism of the present invention; Figure 9 This is a system flowchart of the present invention.

[0026] The attached diagram lists the components represented by each number as follows: 1. Mounting mechanism; 11. Arc-shaped base plate; 12. High-damping silicone pad; 13. Rigid seat; 14. Fixing cylinder; 15. Control box; 16. Buzzer; 2. Vibration damping and fixing mechanism; 21. Damping spring; 22. Mounting base; 23. Ball groove; 24. Extrusion groove; 25. Locking screw; 3. Multi-angle acquisition mechanism; 31. Monitoring camera; 32. Ball head; 33. Infrared polarizer; 34. Data cable; 4. Wiring harness anti-sway mechanism; 41. Fixing rod; 42. Slide block; 43. U-shaped groove; 44. Rotary wheel; 45. Arc-shaped groove; 46. Compression spring. Detailed Implementation

[0027] To make the objectives and advantages of this invention clearer, the invention will be specifically described below with reference to embodiments. It should be understood that the following text is merely used to describe one or more specific embodiments of the invention and does not strictly limit the scope of protection specifically claimed by the invention.

[0028] Example 1:

[0029] See Figure 1 , Figure 2 and Figure 9 As shown, a driver physiological fatigue monitoring system based on behavioral data acquisition is implemented using a driver physiological fatigue monitoring system. The system includes an installation mechanism 1, a shock absorption and fixing mechanism 2, a multi-angle acquisition mechanism 3, a wiring harness anti-sway mechanism 4, a processor, a locator, a driving recorder, a vehicle control data acquisition module, and a data interaction module. The installation mechanism 1 is fixed to the steering column cover of the vehicle steering wheel. The multi-angle acquisition mechanism 3 is positioned above the installation mechanism 1 via the shock absorption and fixing mechanism 2. The wiring harness anti-sway mechanism 4 is mounted on the installation mechanism 1. The processor is connected to the multi-angle acquisition mechanism 3, the locator, the driving recorder, the vehicle control data acquisition module, and the data interaction module.

[0030] See Figure 1 , Figure 2 , Figure 3 and Figure 9 As shown, the mounting mechanism 1 is installed on the steering column cover of the vehicle's steering wheel, positioning the multi-angle acquisition mechanism 3 in the acquisition area in front of the driver's face, and the multi-angle acquisition mechanism 3 does not rotate with the steering wheel. After the vehicle is started, the control box 15 supplies power to the system and performs a self-test. After the system self-test is normal, it enters the driver fatigue driving monitoring state. Before entering the monitoring state, the acquisition orientation of the multi-angle acquisition mechanism 3 is adjusted according to the driver's body shape and sitting posture, so that the multi-angle acquisition mechanism 3 can face the driver's face to acquire images, and the adjusted acquisition angle is locked and fixed by the shock absorption and fixing mechanism 2.

[0031] See Figure 7 and Figure 9 As shown, after entering the monitoring state, the multi-angle acquisition unit 3 captures real-time images of the driver's face, forming driver facial behavior data, which is then transmitted to the processor. The processor filters, extracts features, and identifies fatigue in the driver's facial behavior data, generating a first fatigue judgment result. The driving recorder collects forward driving data and transmits it to the processor, while the vehicle control data acquisition module collects vehicle control data and transmits it to the processor. Based on the forward driving data and vehicle control data, the processor determines whether the vehicle is traveling in a straight line, whether there is a directional deviation, whether there is S-shaped driving, and whether there are any abnormal controls, generating a vehicle driving status judgment result. The locator collects vehicle position data and transmits it to the processor. Based on the vehicle position data, the processor determines whether the vehicle is in motion and calculates the driver's continuous driving time based on the duration of the vehicle being in motion, generating a second fatigue judgment result. The data interaction module acquires nighttime fatigue risk data and transmits it to the processor. The processor statistically analyzes the nighttime fatigue risk data according to preset time slices, generating a nighttime insufficient rest risk judgment result.

[0032] See Figure 9As shown, the processor combines the first fatigue assessment result, the vehicle driving status assessment result, the second fatigue assessment result, and the nighttime insufficient rest risk assessment result to determine whether the driver is in a state of fatigued driving or has a physiological fatigue risk. When the first fatigue assessment result shows that the driver's facial behavior data indicates fatigue, or the second fatigue assessment result shows that the vehicle has been in continuous motion for 4 hours, or the nighttime insufficient rest risk assessment result shows that the driver has not had enough rest at night and continues to drive the motor vehicle during the daytime period after the nighttime period ends, the processor determines that the driver is in a state of fatigued driving or has a physiological fatigue risk. When the vehicle driving status assessment result shows that the vehicle deviates from its direction, drives in an S-shape, or has abnormal handling when not turning or changing lanes, the processor uses the vehicle driving status assessment result as an auxiliary basis for judging driving abnormalities. When the processor determines that the driver is in a state of fatigued driving or has a physiological fatigue risk, the control box 15 controls the buzzer 16 to sound an alarm. After the vehicle stops driving, the system stores the data generated during the monitoring process.

[0033] Example 2:

[0034] See Figure 1 , Figure 2 , Figure 3 and Figure 5 As shown, based on Embodiment 1, the mounting mechanism 1 is used to fix the driver's physiological fatigue monitoring system to the vehicle steering column cover. The mounting mechanism 1 includes an arc-shaped base plate 11, a high-damping silicone pad 12, a rigid seat 13, a fixing cylinder 14, a control box 15, and a buzzer 16. The arc-shaped base plate 11 is adapted to the curved surface of the vehicle steering column cover, and is adhered and fixed to the steering column cover, allowing the mounting mechanism 1 to conform to the vehicle steering column cover to form a basic mounting position.

[0035] See Figure 3 , Figure 4 and Figure 5 As shown, the high-damping silicone pad 12 is bonded to the top of the arc-shaped base plate 11, and the rigid seat 13 is bonded to the top of the high-damping silicone pad 12. The arc-shaped base plate 11, the high-damping silicone pad 12, and the rigid seat 13 form a vibration-damping mounting structure in the vertical direction. During vehicle operation, the vibration generated at the steering column cover is first transmitted to the arc-shaped base plate 11, and then buffered by the high-damping silicone pad 12 before being transmitted to the rigid seat 13, thereby reducing the impact of vehicle vibration transmitted to the multi-angle acquisition mechanism 3.

[0036] See Figure 1 , Figure 2 , Figure 3 and Figure 9As shown, the fixing cylinder 14 is located at the top center of the rigid seat 13. The fixing cylinder 14 is used to accommodate part of the shock absorption and fixing mechanism 2 and to provide an installation base for the multi-angle acquisition mechanism 3. The control box 15 is fixed to the rear side of the fixing cylinder 14, and the buzzer 16 is fixed to one side of the control box 15. The control box 15 is used to supply power to the system and cooperate with the processor to output alarm control. The buzzer 16 is used to issue an alarm when the processor determines that the driver is in a state of fatigue or has a risk of physiological fatigue.

[0037] Example 3:

[0038] See Figure 3 , Figure 4 , Figure 5 and Figure 6 As shown, based on Embodiment 1 or Embodiment 2, the shock absorption and fixing mechanism 2 is used to provide shock absorption support for the multi-angle acquisition mechanism 3 and to lock and fix the adjusted acquisition angle of the multi-angle acquisition mechanism 3. The shock absorption and fixing mechanism 2 includes a damping spring 21, a mounting base 22, a ball groove 23, a compression groove 24, and a locking screw 25. The three damping springs 21 are evenly fixed to the bottom of the fixing cylinder 14, and the mounting base 22 is installed on the top of the three damping springs 21. The mounting base 22 is adapted to the internal dimensions of the fixing cylinder 14, and the upper part of the mounting base 22 extends beyond the fixing cylinder 14.

[0039] See Figure 4 , Figure 6 and Figure 7 As shown, a ball groove 23 is formed on the top of the mounting base 22, and the ball groove 23 is adapted to accommodate the ball head 32 of the multi-angle acquisition mechanism 3. Multiple compression grooves 24 are evenly formed on the top of the mounting base 22 and spaced around the ball groove 23. The compression grooves 24 provide deformation space for the ball groove 23 to contract when the top of the mounting base 22 is compressed. A locking screw 25 is screwed onto the surface of the mounting base 22.

[0040] See Figure 3 , Figure 4 and Figure 6 As shown, when adjusting the acquisition orientation of the multi-angle acquisition mechanism 3 according to the driver's body shape, the ball head 32 rotates within the ball groove 23, and the multi-angle acquisition mechanism 3 adjusts to a position facing the driver's face along with the ball head 32. After the acquisition orientation adjustment is completed, the locking screw 25 is screwed upward, and the locking screw 25 presses against the top of the mounting base 22, causing the ball groove 23 to contract and clamp the ball head 32. At the same time, the gaps of the multiple compression grooves 24 decrease as the ball groove 23 contracts, thereby maintaining a fixed acquisition angle for the multi-angle acquisition mechanism 3. When the vehicle is in motion, the damping spring 21 supports the mounting base 22, reducing the transmission of vehicle vibrations to the multi-angle acquisition mechanism 3.

[0041] Example 4:

[0042] See Figure 3 , Figure 4 , Figure 7 and Figure 9 As shown, based on any of Embodiments 1 to 3, the multi-angle acquisition mechanism 3 is used to acquire driver facial behavior data. The multi-angle acquisition mechanism 3 includes a monitoring camera 31, a ball head 32, an infrared polarizer 33, and a data cable 34. The ball head 32 is fixed to the bottom of the monitoring camera 31 and is installed in a ball groove 23, allowing the monitoring camera 31 to rotate relative to the mounting base 22 via the ball head 32. When adjusting the acquisition orientation, the monitoring camera 31 drives the ball head 32 to rotate within the ball groove 23 so that the monitoring camera 31 faces the driver's face.

[0043] See Figure 7 As shown, an infrared polarizer 33 is installed on the front side of the monitoring camera 31. The infrared polarizer 33 is located in the light-receiving direction of the monitoring camera 31 and is used to reduce the influence of direct sunlight, oncoming headlights, and reflected light from the windshield on the acquisition of driver facial behavior data. The data cable 34 is connected between the monitoring camera 31 and the control box 15 and is used to power the monitoring camera 31 and transmit driver facial behavior data.

[0044] See Figure 7 and Figure 9 As shown, the monitoring camera 31 continuously acquires images of the driver's face and transmits these images as driver facial behavior data to the processor. The processor filters the driver facial behavior data to reduce the impact of vehicle vibration, glare from strong light, and windshield reflection on the image data. The processor then performs feature engineering on the filtered driver facial behavior data, generating one or more of the following features: squinting feature, blinking frequency feature, head-down feature, and yawning feature. Subsequently, the processor inputs the generated facial features into a pre-trained fatigue recognition model and determines whether the driver's facial behavior data corresponds to a fatigue state based on the output of the pre-trained fatigue recognition model. When the driver's facial behavior data corresponds to a fatigue state, the processor generates a first fatigue judgment result indicating that the driver is fatigued.

[0045] See Figure 9 As shown, the pre-trained fatigue recognition model includes one or more of the following: logistic regression model, clustering model, and K-means model. When the fatigue probability output by the logistic regression model reaches a preset threshold, or when the clustering model or K-means model classifies the facial features corresponding to the driver's facial behavior data into the fatigue state category, the processor generates a first fatigue judgment result indicating that the driver is fatigued.

[0046] Example 5:

[0047] See Figure 8 and Figure 9As shown, based on any of Embodiments 1 to 4, the wiring harness anti-sway mechanism 4 is used to provide elastic support for the data cable 34 during the adjustment of the monitoring camera 31 angle and vehicle driving vibration. The wiring harness anti-sway mechanism 4 includes a fixed rod 41, a slide 42, a U-shaped groove 43, a rotating wheel 44, an arc-shaped groove 45, and a compression spring 46. The two fixed rods 41 are vertically fixed to the upper surface of the control box 15 and are symmetrically arranged. The slide 42 is slidably installed between the two fixed rods 41 and can slide up and down along the setting direction of the fixed rods 41.

[0048] See Figure 8 As shown, a U-shaped groove 43 is formed at the center of the upper surface of the slide 42. The rotating wheel 44 is rotatably mounted in the U-shaped groove 43. An arc-shaped groove 45 is formed on the surface of the rotating wheel 44 and is adapted to the data cable 34. A compression spring 46 is sleeved on the lower surface of the fixing rod 41 and applies an upward thrust to the slide 42, causing the slide 42 to drive the rotating wheel 44 to press against the data cable 34. The rotating wheel 44 applies an upward elastic support force to the data cable 34 through the arc-shaped groove 45.

[0049] See Figure 7 and Figure 8 As shown, when the monitoring camera 31 adjusts its angle, the data cable 34 is pulled or relaxed as the monitoring camera 31 rotates. The rotating wheel 44 provides rolling support for the data cable 34 through the arc-shaped groove 45, and the compression spring 46 compensates for the slack of the data cable 34 through the slide 42, keeping the data cable 34 taut during adjustment. During vehicle operation, the wiring harness anti-sway mechanism 4 continues to provide elastic support for the data cable 34, ensuring that the data cable 34 maintains a relatively stable connection under vehicle vibration conditions.

[0050] Example 6:

[0051] See Figure 9 As shown, based on any of Embodiments 1 to 5, the dashcam and vehicle control data acquisition module are used in conjunction with the processor to form a vehicle driving status judgment result. The dashcam captures the driving scene in front of the vehicle and transmits the forward driving data to the processor. The processor determines whether the vehicle is driving in a straight line based on the driving scene in front of the vehicle. When the vehicle deviates from its direction or drives in an S-shape, the processor further determines, in conjunction with the actual road conditions, whether the deviation or S-shape is caused by turning or changing lanes. When the deviation or S-shape is not caused by turning or changing lanes, the processor forms a driving anomaly judgment result and uses this driving anomaly judgment result as an auxiliary basis for the comprehensive judgment of fatigue driving.

[0052] See Figure 9As shown, the vehicle control data acquisition module collects vehicle control data and transmits it to the processor. Vehicle control data includes one or more of the following: steering wheel torque data, steering wheel angle data, and vehicle lateral deviation data. The processor performs feature engineering on the vehicle control data, generating one or more of the following: hand-off swing features, large deflection features, random swing features, and uncorrected deviation features. These generated vehicle control features are then input into a pre-trained vehicle anomaly detection model. The pre-trained model is used to detect anomalies in the vehicle control data in real time, forming a vehicle driving state judgment result.

[0053] See Figure 9 As shown, the pre-trained vehicle anomaly recognition model includes one or more of the following: anomaly detection model, random forest model, and DBSCAN model. When the processor determines that the vehicle has deviated from its course, driven in an S-shape, or exhibited abnormal handling in non-turning or non-lane-changing situations based on the vehicle's forward driving data and vehicle handling data, the processor uses the vehicle driving state judgment result as an auxiliary basis for judging driving anomalies, and combines it with the first fatigue judgment result, the second fatigue judgment result, and the risk judgment result of insufficient nighttime rest in the subsequent comprehensive judgment process.

[0054] Example 7:

[0055] See Figure 9 As shown, based on any one of Embodiments 1 to 6, the locator is used to collect vehicle location data and transmit the vehicle location data to the processor. The processor determines whether the vehicle is in motion based on the vehicle location data. When the vehicle is continuously in motion, the processor accumulates the continuous motion time of the vehicle; when the vehicle stops moving, the processor uses the cessation of vehicle movement as the basis for the start of a rest period; when the continuous motion time of the vehicle reaches 4 hours, the processor generates a second fatigue judgment result indicating that the driver is in a state of fatigued driving.

[0056] See Figure 9 As shown, when calculating the continuous driving time of the vehicle, the processor uses whether the vehicle has stopped moving as the rest condition, rather than whether the vehicle has been turned off. When the vehicle has stopped moving but the engine is not turned off, the processor takes the moment when the vehicle stops moving as the start time of the rest; when the vehicle starts moving again, the processor recalculates the continuous driving time. In this way, the processor can calculate the driver's continuous driving time based on the actual driving state of the vehicle.

[0057] Example 8:

[0058] See Figure 9As shown, based on any one of Embodiments 1 to 7, the data interaction module is used to acquire nighttime fatigue risk data and transmit the nighttime fatigue risk data to the processor. The nighttime fatigue risk data includes vehicle location data, driver identification data, vehicle license plate data, and / or roadside video data collected during the nighttime period. The nighttime period is from 0:00 to 5:00. The processor statistically analyzes the nighttime fatigue risk data according to preset time slices, where each preset time slice is a continuous series of time periods, and each time slice is 20 minutes long.

[0059] See Figure 9 As shown, the processor counts one or more of the following within each time slice: vehicle travel distance, number of administrative regions the vehicle passes through, number of driver face scans, number of vehicle license plate scans, and number of scan locations. When a vehicle travels 200 kilometers during the nighttime period, its location crosses multiple administrative regions, and the driver identification result shows the same driver, the processor generates a risk assessment result for insufficient nighttime rest.

[0060] See Figure 9 As shown, when a driver's face is collected frequently and at multiple locations during nighttime hours, the processor generates a risk assessment result for insufficient nighttime rest. Alternatively, when a vehicle license plate is collected frequently and at multiple locations during nighttime hours, and the driver identification result shows the same driver, the processor generates a risk assessment result for insufficient nighttime rest. When the risk assessment result for insufficient nighttime rest indicates that the driver is not getting enough sleep at night, and the driver continues to drive the motor vehicle during the daytime hours after the nighttime period ends, the processor uses this risk assessment result for insufficient nighttime rest as a basis for assessing the risk of physiological fatigue.

[0061] Example 9:

[0062] See Figures 1 to 9 As shown, based on any one of Embodiments 1 to 8, a typical usage process of the present invention is as follows: First, the arc-shaped base plate 11 is adhered and fixed to the steering column cover of the vehicle steering wheel, so that the mounting mechanism 1 is fixed at the steering column cover; the high-damping silicone pad 12 is located between the arc-shaped base plate 11 and the rigid seat 13, the fixing cylinder 14 is set at the top center of the rigid seat 13, the control box 15 is fixed on the rear side of the fixing cylinder 14, and the buzzer 16 is fixed on one side of the control box 15. The shock absorption and fixing mechanism 2 is set inside the fixing cylinder 14, and the multi-angle acquisition mechanism 3 is installed in front of the wiring harness anti-sway mechanism 4 at the corresponding acquisition position through the ball head 32 and the ball groove 23.

[0063] See Figure 3 , Figure 4 , Figure 6 and Figure 7As shown, after the vehicle starts, the control box 15 supplies power to the system and performs a self-test. After the system self-test is normal, the acquisition orientation of the monitoring camera 31 is adjusted according to the driver's body shape. During adjustment, the monitoring camera 31 drives the ball head 32 to rotate within the ball groove 23 until the monitoring camera 31 faces the driver's face. After adjustment, the locking screw 25 is screwed up and moves upward. The locking screw 25 presses against the top of the mounting base 22, causing the ball groove 23 to contract and clamp the ball head 32, thereby fixing the acquisition angle of the monitoring camera 31.

[0064] See Figure 7 , Figure 8 and Figure 9 As shown, the monitoring camera 31 captures real-time images of the driver's face and transmits the driver's facial behavior data to the processor via the data cable 34. During the wiring harness connection process, the compression spring 46 pushes the slide 42 upward, and the slide 42 drives the rotating wheel 44 to apply an upward elastic support force to the data cable 34 through the arc-shaped groove 45, keeping the data cable 34 taut during the adjustment of the monitoring camera 31's angle and vehicle vibration. The processor filters, performs feature engineering processing, and performs model recognition on the driver's facial behavior data to form the first fatigue judgment result.

[0065] See Figure 9 As shown, the dashcam transmits forward driving data to the processor, and the vehicle control data acquisition module transmits vehicle control data to the processor. The processor determines the vehicle's driving status based on whether the vehicle is traveling in a straight line, whether there is directional deviation, whether it is driving in an S-shape, and whether there are any abnormal controls. It also filters out deviations caused by turning or lane changes based on actual road conditions. The locator transmits vehicle position data to the processor. The processor determines whether the vehicle is in motion based on the vehicle position data and calculates the driver's continuous driving time to form a second fatigue assessment result. The data interaction module transmits nighttime fatigue risk data to the processor. The processor statistically analyzes the nighttime fatigue risk data according to preset time slices to form a nighttime insufficient rest risk assessment result.

[0066] See Figure 9As shown, the processor comprehensively judges the first fatigue assessment result, the vehicle driving status assessment result, the second fatigue assessment result, and the nighttime insufficient rest risk assessment result. When the first fatigue assessment result shows that the driver's facial behavior data indicates fatigue, or the second fatigue assessment result shows that the vehicle has been in continuous motion for 4 hours, or the nighttime insufficient rest risk assessment result shows that the driver has not had enough rest at night and continues to drive the motor vehicle during the daytime after the nighttime period ends, the processor determines that the driver is in a state of fatigued driving or has a physiological fatigue risk. When the vehicle driving status assessment result shows that the vehicle deviates from its direction, drives in an S-shape, or has abnormal handling when not turning or changing lanes, the processor uses the vehicle driving status assessment result as an auxiliary basis for judging driving abnormalities. When the processor determines that the driver is in a state of fatigued driving or has a physiological fatigue risk, the control box 15 controls the buzzer 16 to sound an alarm. After the vehicle finishes driving, the system stores the data generated during the monitoring process.

[0067] The working principle of this invention is as follows: In use, first fix the mounting mechanism 1 to the steering column cover of the vehicle's steering wheel, so that the multi-angle acquisition mechanism 3 is located at the steering column cover, and ensure that the multi-angle acquisition mechanism 3 does not rotate with the steering wheel during vehicle driving. During installation, the arc-shaped base plate 11 is adapted to the curved surface of the steering column cover and is bonded and fixed. The high-damping silicone pad 12 is located between the arc-shaped base plate 11 and the rigid seat 13. The fixing cylinder 14 is set at the top center of the rigid seat 13. The control box 15 is fixed to the rear side of the fixing cylinder 14, and the buzzer 16 is fixed to one side of the control box 15. This allows the multi-angle acquisition mechanism 3 to be kept in a stable acquisition position in front of the driver's face.

[0068] After the vehicle starts, the control box 15 supplies power to the system and performs a self-test. Once the system self-test is normal, it enters the driver fatigue monitoring state. Subsequently, the acquisition orientation of the multi-angle acquisition mechanism 3 is adjusted according to the driver's body shape and sitting posture. During adjustment, the monitoring camera 31 rotates within the shock absorption and fixing mechanism 2 via the ball head 32 at the bottom, so that the monitoring camera 31 faces the driver's face. After the acquisition orientation of the monitoring camera 31 is adjusted, the locking screw 25 is screwed up, and the locking screw 25 presses against the top of the mounting base 22, causing the ball groove 23 on the top of the mounting base 22 to contract. At the same time, the gaps between the multiple compression grooves 24 are reduced, thereby clamping the ball head 32 through the ball groove 23, keeping the acquisition angle of the monitoring camera 31 fixed.

[0069] During the adjustment of the monitoring camera 31 angle, the data cable 34 will be pulled or loosened as the monitoring camera 31 angle changes. At this time, the cable harness anti-sway mechanism 4 provides elastic support for the data cable 34. Specifically, two fixed rods 41 are vertically fixed to the upper surface of the control box 15, and the slide 42 is slidably installed between the two fixed rods 41. A U-shaped groove 43 is opened in the center of the upper surface of the slide 42, and the rotating wheel 44 is rotatably installed in the U-shaped groove 43. The arc-shaped groove 45 is adapted to the data cable 34, and the compression spring 46 is sleeved on the lower surface of the fixed rod 41 and applies an upward thrust to the slide 42, so that the rotating wheel 44 applies an upward elastic support force to the data cable 34 through the arc-shaped groove 45. In this way, the data cable 34 is kept taut during the adjustment of the monitoring camera 31 angle and the vibration of the vehicle, reducing the impact of the loosening of the data cable 34 on power supply and data transmission.

[0070] During vehicle operation, the arc-shaped base plate 11, the high-damping silicone pad 12, and the rigid seat 13 form a vibration damping mounting structure. Vehicle vibration is first buffered by the high-damping silicone pad 12. At the same time, three damping springs 21 are evenly fixed at the bottom of the fixing cylinder 14, and the mounting seat 22 is installed on top of the three damping springs 21. When vehicle vibration continues to be transmitted upward, the damping springs 21 further reduce the impact of vibration on the mounting seat 22 and the monitoring camera 31, so that the monitoring camera 31 can relatively stably collect the driver's facial image during driving.

[0071] Once in monitoring mode, the monitoring camera 31 captures real-time images of the driver's face, generating facial behavior data. An infrared polarizer 33 is installed on the front of the monitoring camera 31 to reduce the impact of direct sunlight, oncoming headlights, and windshield reflections on the driver's facial behavior data acquisition during the data collection process. The driver's facial behavior data acquired by the monitoring camera 31 is transmitted to the processor via data cable 34. The processor filters the driver's facial behavior data to reduce the impact of vehicle vibration, glare from strong light, and windshield reflections on the image data.

[0072] Subsequently, the processor performs feature engineering on the filtered driver facial behavior data, generating one or more of the following features: squinting, blinking frequency, head-down posture, and yawning. These generated facial features are then input into a pre-trained fatigue recognition model. The pre-trained fatigue recognition model includes one or more of logistic regression, clustering, and K-means models. When the fatigue probability output by the logistic regression model reaches a preset threshold, or when the clustering or K-means model categorizes the facial features corresponding to the driver's facial behavior data into a fatigue state category, the processor generates a first fatigue judgment result indicating that the driver is fatigued.

[0073] Meanwhile, the dashcam collects forward driving data and transmits it to the processor; the vehicle control data acquisition module collects vehicle control data and transmits it to the processor. The processor determines whether the vehicle is traveling in a straight line based on the forward driving image. When the vehicle deviates from its course or veers in an S-shape, the processor further analyzes the actual road conditions to determine if the deviation or S-shape is caused by turning or changing lanes. If the deviation or S-shape is not caused by turning or changing lanes, the processor generates a driving anomaly judgment result, which serves as an auxiliary basis for a comprehensive assessment of fatigue driving.

[0074] Vehicle handling data includes one or more of the following: steering wheel torque data, steering wheel angle data, and vehicle lateral deviation data. The processor performs feature engineering on the vehicle handling data to generate one or more of the following: hand-off swing features, large yaw features, random swing features, and uncorrected deviation features. These generated vehicle handling features are then input into a pre-trained vehicle anomaly detection model. The pre-trained vehicle anomaly detection model includes one or more of the following: anomaly detection model, random forest model, and DBSCAN model. The processor uses the pre-trained vehicle anomaly detection model to detect anomalies in the vehicle handling data in real time and combines this with the vehicle's forward driving data to form a vehicle driving state judgment result.

[0075] Simultaneously, the locator collects vehicle location data in real time and transmits the data to the processor. The processor determines whether the vehicle is in motion based on the location data. When the vehicle is continuously in motion, the processor accumulates the continuous motion time. When the vehicle stops moving, the processor uses the cessation of motion as the start of a rest period, rather than whether the vehicle is turned off. When the vehicle is still running but has stopped moving, the processor uses the moment of cessation of motion as the start of a rest period. When the vehicle resumes motion, the processor recalculates the continuous motion time. When the continuous motion time reaches 4 hours, the processor generates a second fatigue assessment result indicating that the driver is experiencing fatigue.

[0076] In addition, the processor acquires nighttime fatigue risk data through a data interaction module. This data includes vehicle location data, driver identification data, vehicle license plate data, and / or roadside video data collected during the nighttime period. The nighttime period is from 0:00 to 5:00. The processor statistically analyzes the nighttime fatigue risk data according to preset time slices, which are several consecutive time intervals, each lasting 20 minutes. Within each time slice, the processor statistically analyzes one or more of the following: vehicle travel distance, number of administrative regions passed by the vehicle, number of driver facial recognition scans, number of vehicle license plate scans, and number of data collection locations.

[0077] When a vehicle travels 200 kilometers or more during the nighttime period, spans multiple administrative regions, and the driver identification result shows the same driver, the processor generates a nighttime sleep deprivation risk assessment result. Alternatively, when the driver's face is collected frequently and at multiple locations during the nighttime period, the processor generates a nighttime sleep deprivation risk assessment result. Or, when the vehicle license plate is collected frequently and at multiple locations during the nighttime period, and the driver identification result shows the same driver, the processor generates a nighttime sleep deprivation risk assessment result. When the nighttime sleep deprivation risk assessment result indicates that the driver is experiencing nighttime sleep deprivation, and the driver continues to drive the motor vehicle during the daytime period after the nighttime period ends, the processor uses this nighttime sleep deprivation risk assessment result as the basis for assessing physiological fatigue risk.

[0078] Subsequently, the processor combines the first fatigue assessment result, the vehicle driving status assessment result, the second fatigue assessment result, and the nighttime insufficient rest risk assessment result to determine whether the driver is in a state of fatigued driving or has a physiological fatigue risk. Specifically, the first fatigue assessment result reflects whether the driver's facial behavior data corresponds to a fatigue state; the second fatigue assessment result reflects whether the driver's continuous driving time meets the fatigue risk criteria; the nighttime insufficient rest risk assessment result reflects whether the driver is at risk of continuing to drive during the day after insufficient nighttime rest; and the vehicle driving status assessment result serves as an auxiliary basis for judging driving abnormalities.

[0079] When the first fatigue assessment result shows that the driver's facial behavior data indicates fatigue, or the second fatigue assessment result shows that the vehicle has been in continuous motion for 4 hours, or the nighttime sleep deprivation risk assessment result shows that the driver has not had enough rest at night and continues to drive the motor vehicle during the daytime after the nighttime period ends, the processor determines that the driver is in a state of fatigued driving or has a risk of physiological fatigue. When the vehicle driving status assessment result shows that the vehicle deviates from its direction, drives in an S-shape, or has abnormal handling when not turning or changing lanes, the processor uses the vehicle driving status assessment result as an auxiliary basis for judging driving abnormalities, and combines the first fatigue assessment result, the second fatigue assessment result, and the nighttime sleep deprivation risk assessment result to make a comprehensive judgment on the driver's status.

[0080] When the processor determines that the driver is in a state of fatigued driving or has a risk of physiological fatigue, the processor outputs an alarm command to the control box 15, and the control box 15 controls the buzzer 16 to sound an alarm to remind the driver to take rest measures in time. After the vehicle finishes driving, the system stores the driver's facial behavior data, vehicle driving status judgment results, vehicle continuous movement duration data, night fatigue risk data, night insufficient rest risk judgment results, and fatigue driving judgment results generated during the monitoring process, thereby completing a driver physiological fatigue driving monitoring process.

[0081] The above description is merely a preferred embodiment of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention. Structures, devices, and operating methods not specifically described or explained in this invention are implemented according to conventional methods in the art unless otherwise specified or limited.

Claims

1. A driver physiological fatigue monitoring system based on behavioral data acquisition, characterized in that, The system employs a driver physiological fatigue monitoring system, which includes an installation mechanism (1), a shock absorption and fixing mechanism (2), a multi-angle acquisition mechanism (3), a wiring harness anti-sway mechanism (4), a processor, a locator, a driving recorder, a vehicle operation data acquisition module, and a data interaction module. The monitoring system includes the following steps: S1. Fix the installation mechanism (1) on the steering column cover of the vehicle steering wheel so that the multi-angle acquisition mechanism (3) is installed at the steering column cover of the steering wheel and does not rotate with the steering wheel; S2. After the vehicle starts, the control box (15) supplies power to the system and performs a self-test. After the system self-test is normal, it enters the driver fatigue driving monitoring state. S3. Adjust the acquisition orientation of the multi-angle acquisition mechanism (3) according to the driver's body shape, and lock and fix the adjusted acquisition angle through the shock absorption and fixing mechanism (2) so that the multi-angle acquisition mechanism (3) can face the driver's face to collect data. S4. The driver's face is captured in real time by the multi-angle acquisition mechanism (3) to form driver's facial behavior data, and the driver's facial behavior data is transmitted to the processor. The processor filters, extracts features and identifies fatigue in the driver's facial behavior data to form the first fatigue judgment result. S5. The vehicle forward driving data collected by the dashcam is transmitted to the processor, and the vehicle control data collected by the vehicle control data acquisition module is transmitted to the processor. The processor determines whether the vehicle is driving in a straight line, whether there is a directional deviation, whether there is an S-shaped driving, and whether there is a control abnormality based on the vehicle forward driving data and vehicle control data, and forms a vehicle driving status judgment result. It also filters out deviations caused by turning or changing lanes based on the actual road conditions. S6. The vehicle location data collected by the locator is transmitted to the processor. The processor determines whether the vehicle is in motion based on the vehicle location data, and calculates the driver's continuous driving time based on the duration of the vehicle being in motion, thus forming a second fatigue judgment result. S7. The processor obtains nighttime fatigue risk data through the data interaction module. The nighttime fatigue risk data includes vehicle location data, driver identification data, vehicle license plate collection data and / or roadside video collection data during the nighttime period. The processor statistically analyzes the nighttime fatigue risk data according to a preset time slice to form a nighttime insufficient rest risk judgment result. S8. The processor combines the first fatigue judgment result, the vehicle driving status judgment result, the second fatigue judgment result, and the nighttime insufficient rest risk judgment result to determine whether the driver is in a state of fatigued driving or has a physiological fatigue risk. When the first fatigue assessment result shows that the driver's facial behavior data indicates fatigue, or the second fatigue assessment result shows that the vehicle has been in continuous motion for 4 hours, or the nighttime insufficient rest risk assessment result shows that the driver has insufficient rest at night and continues to drive the motor vehicle during the daytime after the nighttime period ends, the processor determines that the driver is in a state of fatigued driving or has a physiological fatigue risk; when the vehicle driving status assessment result shows that the vehicle deviates from its direction, drives in an S-shape, or has abnormal handling when not turning or changing lanes, the processor uses the vehicle driving status assessment result as an auxiliary basis for judging driving abnormalities. S9. When the processor determines that the driver is in a state of fatigue or has a risk of physiological fatigue, the control box (15) controls the buzzer (16) to issue an alarm; after the vehicle finishes driving, the data generated during the monitoring process is stored.

2. The driver physiological fatigue monitoring system based on behavioral data acquisition according to claim 1, characterized in that: In step S4, the process of forming the first fatigue judgment result includes: continuously acquiring driver facial images through a multi-angle acquisition mechanism (3); transmitting the acquired driver facial images as driver facial behavior data to the processor; filtering the driver facial behavior data to reduce the impact of vehicle vibration, strong light glare, and windshield reflection on the image data; performing feature engineering on the filtered driver facial behavior data to generate one or more of the following features: squinting feature, blinking frequency feature, head-down feature, and yawning feature; inputting the generated facial features into a pre-trained fatigue recognition model and judging whether the driver facial behavior data corresponds to a fatigue state based on the output of the pre-trained fatigue recognition model; when the driver facial behavior data corresponds to a fatigue state, forming the first fatigue judgment result indicating that the driver is fatigued.

3. The driver physiological fatigue monitoring system based on behavioral data acquisition according to claim 2, characterized in that: The pre-trained fatigue recognition model includes one or more of the following: logistic regression model, clustering model, and K-means model. When the fatigue probability output by the logistic regression model reaches a preset threshold, or when the clustering model or K-means model classifies the facial features corresponding to the driver's facial behavior data into the fatigue state category, the processor generates a first fatigue judgment result indicating that the driver is fatigued.

4. The driver physiological fatigue monitoring system based on behavioral data acquisition according to claim 1, characterized in that: In step S5, the process of forming the vehicle driving status judgment result includes: collecting the driving scene in front of the vehicle through the driving recorder; the processor judging whether the vehicle is driving in a straight line based on the driving scene in front of the vehicle; when the vehicle deviates from its direction or drives in an S-shape, the processor further judges whether the deviation from its direction or the S-shape is caused by turning or changing lanes in combination with the actual road conditions; when the deviation from its direction or the S-shape is not caused by turning or changing lanes, the processor forms a driving abnormality judgment result and uses the driving abnormality judgment result as an auxiliary basis for the comprehensive judgment of fatigue driving.

5. A driver physiological fatigue monitoring system based on behavioral data acquisition according to claim 1, characterized in that: In step S5, the vehicle handling data includes one or more of the following: steering wheel rotation torque data, steering wheel angle data, and vehicle lateral offset data. The processor performs feature engineering on the vehicle handling data to generate one or more of the following: hand-off swing feature, large deflection feature, random swing feature, and uncorrected deviation feature. The generated vehicle handling features are then input into a pre-trained vehicle anomaly recognition model. The pre-trained vehicle anomaly recognition model is used to detect whether there are any anomalies in the vehicle handling data in real time, thereby forming a vehicle driving state judgment result. The pre-trained vehicle anomaly recognition model includes one or more of the following: anomaly detection model, random forest model, and DBSCAN model.

6. A driver physiological fatigue monitoring system based on behavioral data acquisition according to claim 1, characterized in that: In step S6, the process of forming the second fatigue judgment result includes: real-time acquisition of vehicle position data through the locator; the processor determining whether the vehicle is in motion based on the vehicle position data; when the vehicle is continuously in motion, the processor accumulates the continuous motion time of the vehicle; when the vehicle stops moving, the processor uses the cessation of vehicle movement as the basis for the start of rest; when the continuous motion time of the vehicle reaches 4 hours, the processor forms a second fatigue judgment result indicating that the driver is in a state of fatigued driving.

7. A driver physiological fatigue monitoring system based on behavioral data acquisition according to claim 6, characterized in that: When calculating the duration of continuous vehicle movement, the processor uses whether the vehicle stops moving as the rest condition, rather than whether the vehicle is turned off, so that the monitoring system can be applied to vehicles that cannot be turned off. When the vehicle is still running but has stopped moving, the processor takes the moment when the vehicle stops moving as the start of the rest period; when the vehicle starts moving again, the processor recalculates the duration of continuous movement.

8. A driver physiological fatigue monitoring system based on behavioral data acquisition according to claim 1, characterized in that: In step S7, the nighttime period is from 0:00 to 5:00, and the preset time slice is a series of consecutive time periods, with each time slice lasting 20 minutes. The processor counts one or more of the following in each time slice: vehicle travel distance, number of administrative regions passed by the vehicle, number of driver face captures, number of vehicle license plate captures, and number of capture locations. When a vehicle travels 200 kilometers during the nighttime period, and its location spans multiple administrative regions, and the driver's identity verification result shows that it is the same driver, the processor generates a risk assessment result for insufficient nighttime rest; or, when the driver's face is collected multiple times and at multiple locations during the nighttime period, the processor generates a risk assessment result for insufficient nighttime rest. Alternatively, when a vehicle's license plate is collected frequently and at multiple locations during the nighttime period, and the driver's identity verification results show that it is the same driver, the processor will generate a risk assessment result for insufficient nighttime rest.

9. A driver physiological fatigue monitoring system based on behavioral data acquisition according to claim 1, characterized in that: In steps S3 and S4, the multi-angle acquisition mechanism (3) includes a monitoring camera (31), a ball head (32), an infrared polarizer (33), and a data cable (34); the ball head (32) is fixed to the bottom of the monitoring camera (31); when adjusting the acquisition orientation, the ball head (32) is rotated within the shock absorption and fixing mechanism (2) so that the monitoring camera (31) faces the driver's face; the infrared polarizer (33) is installed on the front side of the monitoring camera (31) to reduce the influence of direct sunlight, oncoming headlights, and windshield reflection on the acquisition of driver's facial behavior data; the data cable (34) is connected between the monitoring camera (31) and the control box (15) for power supply and data transmission.

10. A driver physiological fatigue monitoring system based on behavioral data acquisition according to claim 9, characterized in that: The shock absorption and fixing mechanism (2) includes a damping spring (21), a mounting base (22), a ball groove (23), an extrusion groove (24), and a locking screw (25); three damping springs (21) are evenly fixed to the bottom of the fixing cylinder (14); the mounting base (22) is installed on the top of the three damping springs (21); the ball groove (23) is opened on the top of the mounting base (22) and is adapted to the ball head (32); multiple extrusion grooves (24) are evenly opened on the top of the mounting base (22); the locking screw (25) is screwed onto the surface of the mounting base (22); After the acquisition orientation of the monitoring camera (31) is adjusted, the locking screw (25) is screwed up and pressed against the top of the mounting base (22), causing the ball groove (23) to contract and clamp the ball head (32), thereby keeping the acquisition angle of the monitoring camera (31) fixed; the mounting mechanism (1) includes an arc-shaped base plate (11), a high-damping silicone pad (12), a rigid seat (13), a fixing cylinder (14), a control box (15), and a buzzer (16); the arc-shaped base plate (11) is adapted to the curved surface of the steering column cover and is bonded and fixed to the steering column cover; the high-damping silicone pad (12) is bonded to the top of the arc-shaped base plate (11); the rigid seat (13) is bonded to the top of the high-damping silicone pad (12); the fixing cylinder (14) is set at the top center of the rigid seat (13); the control box (15) is fixed to the rear side of the fixing cylinder (14); the buzzer (16) is fixed to one side of the control box (15); the The anti-sway mechanism (4) for the wire harness includes a fixed rod (41), a slide (42), a U-shaped groove (43), a rotating wheel (44), an arc-shaped groove (45), and a compression spring (46). The two fixed rods (41) are vertically fixed on the upper surface of the control box (15) and are symmetrically arranged. The slide (42) is slidably installed between the two fixed rods (41). The U-shaped groove (43) is opened at the center of the upper surface of the slide (42). The rotating wheel (44) is rotatably installed in the U-shaped groove (43). The arc-shaped groove (45) is opened on the surface of the rotating wheel (44) and is adapted to the data cable (34). The compression spring (46) is sleeved on the lower surface of the fixed rod (41) and applies an upward thrust to the slide (42), so that the rotating wheel (44) applies an upward elastic support force to the data cable (34) through the arc-shaped groove (45), thereby keeping the data cable (34) taut during the monitoring of camera (31) angle adjustment and vehicle driving vibration.

Images