A driving behavior-based abnormality early warning method and system

Abstract

The application provides a driving behavior-based abnormality early warning method and system, which is applied to the field of data processing; the application realizes multi-level judgment on vehicle driving behavior by combining driver individual driving habits, road types and traffic control information; firstly, based on the driving identity information pre-recorded by the central controller, real-time driving parameters are detected and compared with individual habits; only the behaviors obviously deviating from the habits are subjected to subsequent analysis, thus avoiding "one-size-fits-all" misjudgment; in combination with road types and traffic control information, driving behavior data are generated through a prediction model, and it is further judged whether the behavior is reasonable, thus ensuring the pertinence of early warning; finally, abnormal driving behavior is compared with the group driving behavior baseline, and individual difference dynamic correction is realized, so that on the basis of fully considering the individual differences of drivers, the false alarm rate is effectively reduced, the abnormal behavior is timely warned, and the driving safety and early warning practicability are improved.

Description

Technical Field

[0001] This invention relates to the field of data processing, and in particular to an anomaly warning method and system based on driving behavior. Background Technology

[0002] Most existing abnormal driving behavior warning systems are based on a unified driving behavior model for abnormal identification. For example, they may identify abnormalities by setting a fixed threshold (such as determining abnormality if the acceleration exceeds a certain value during emergency braking) or by using a pre-trained general driving behavior classifier to identify dangerous driving patterns.

[0003] However, there are significant individual differences among drivers. For example, some drivers have a long-term habit of sudden braking and rapid lane changes, but this is not considered dangerous driving in the true sense. Other drivers may have potential risks even if they accelerate slightly. Summary of the Invention

[0004] This invention aims to address the problem of how to fully consider individual differences among drivers in abnormal warnings and avoid misjudgments and false alarms caused by a "one-size-fits-all" approach, and provides an abnormal warning method and system based on driving behavior.

[0005] The present invention employs the following technical means to solve the technical problem:

[0006] This invention provides an anomaly warning method based on driving behavior, comprising:

[0007] Based on the driver identity information pre-collected by the vehicle's central control system, the vehicle's driving parameters during the driving process are detected. Specifically, the driving parameters include the current acceleration, the rate of change of vehicle speed, and the brake pedal depth.

[0008] Determine whether the driving parameters match the driving habits corresponding to the driving identity information;

[0009] If not, then based on the driving road type pre-identified by the vehicle central control system, the traffic control information of the vehicle within a preset range is obtained, and the traffic control information is input into the prediction model pre-trained by the vehicle central control system in real time. Through the prediction model, the driving behavior data of the vehicle is generated. The driving road type specifically includes fast passage roads, urban passage roads, and complex environment roads. The traffic control information specifically includes road traffic restriction, speed time control, and emergency control.

[0010] Determine whether the driving behavior data conforms to the driving indicators of the driving identity information;

[0011] If the conditions are not met, the group driving behavior reference data pre-collected from the driving platform by the vehicle's central control system is compared with the driving behavior data. The difference value between the vehicle and the group baseline is calculated, and the corresponding abnormal driving behavior is identified in the driving behavior data. Based on the abnormal driving behavior, an abnormal warning prompt is generated for the driver through the vehicle. The group driving behavior reference data specifically includes vehicle speed distribution, acceleration range, and lane change frequency.

[0012] Furthermore, before the step of generating the driving behavior data of the vehicle, the method further includes:

[0013] Based on the communication data pre-sent by the vehicle's central control system to the driving platform, the vehicle's location information is obtained through the driving platform;

[0014] Determine whether the location information is within the geographical range pre-recorded by the vehicle's central control system;

[0015] If so, then based on the traffic control requirements preset in the geographical area, driving behavior rules for the vehicle are constructed. According to the driving behavior rules, corresponding violations are marked during the vehicle's operation. The violations are dynamically presented to the vehicle through the vehicle's central control system. Specifically, the dynamic presentation includes dashboard image reminders, voice prompts, and steering wheel and seat vibrations.

[0016] Furthermore, the step of obtaining traffic control information for the vehicle within a preset range based on the driving road type pre-identified by the vehicle's central control system also includes:

[0017] Based on the control type of the traffic control information, the vehicle central control system generates the control content of the vehicle within the preset range, wherein the control type specifically includes dynamic control and static control;

[0018] Determine whether the vehicle has left the preset range;

[0019] If not, the control content is dynamically mapped, and the duration of each control rule's impact on the vehicle within a preset driving time window is calculated based on the vehicle's real-time driving path. Based on the duration of the impact, invalid control information of the control content on the vehicle is adaptively filtered.

[0020] Furthermore, the step of pre-collecting group driving behavior reference data from the driving platform based on the vehicle's central control system also includes:

[0021] Based on the pre-established communication connection between the vehicle central control and the driving platform, the source channels of the group driving behavior reference data are identified, wherein the source channels specifically include data from official traffic management platforms, data collected by car manufacturers themselves, and data from Internet travel platforms;

[0022] Determine whether the data format of the source channels is consistent;

[0023] If not, the temporal continuity of the group driving behavior reference data is obtained, and based on the temporal continuity, missing data values ​​of the group driving behavior reference data are detected, a corresponding missing ratio is generated, and the dynamic timeliness of the group driving behavior reference data is corrected in real time based on the missing ratio.

[0024] Furthermore, the step of determining whether the driving parameters match the driving habits corresponding to the driving identity information also includes:

[0025] Based on the external driving environment pre-detected by the vehicle's central control system, external environment information corresponding to the external driving environment is constructed, wherein the external environment information specifically includes meteorological and natural environment, road traffic conditions, and dynamic traffic participants;

[0026] Determine whether the external environment information belongs to a preset normal driving scenario;

[0027] If not, then based on the environmental type of the external driving environment, the vehicle's deviation behavior is continuously collected within a preset driving time window, the trigger frequency of the deviation behavior is identified, and the deviation behavior is dynamically marked as an observation state based on the trigger frequency.

[0028] Furthermore, the step of determining whether the driving behavior data conforms to the driving indicators of the driving identity information further includes:

[0029] Based on the preset upper limit of the deviation value of the driving indicator, the real-time deviation value of the driving behavior data is collected, wherein the real-time deviation value specifically includes speed deviation rate, acceleration deviation and vehicle distance deviation;

[0030] Determine whether the real-time deviation value exceeds the upper limit of the deviation value;

[0031] If so, the additional time period of the real-time deviation value is identified, the deviation values ​​of each indicator of the driving behavior data are weighted and summed to generate a comprehensive deviation score, and a tolerance range of the driving behavior data is constructed based on the comprehensive deviation score. The additional time period specifically includes daytime, nighttime, peak hours and off-peak hours, and the tolerance range specifically includes compliance with the indicator, slight deviation and severe deviation.

[0032] Furthermore, before the step of detecting the vehicle's driving parameters during driving based on the driver's identity information pre-recorded by the vehicle's central control system, the method further includes:

[0033] Based on the user's identity identifier entered in the vehicle, the user's identity data is read from the vehicle's central control system;

[0034] Determine whether the identity data is included in the vehicle's central control system;

[0035] If not, the preset driving text of the vehicle's central control is activated to verify the user's driving. Based on the verification result, a driving profile of the user driving the vehicle is constructed. The user's driving event habits are retained in the driving profile. The driving event habits specifically include rapid acceleration, sudden braking, emergency lane change, continuous speeding, and lane departure.

[0036] The present invention also provides an abnormal warning system based on driving behavior, comprising:

[0037] The detection module is used to detect the driving parameters of the vehicle during driving based on the driver identity information pre-recorded in the vehicle's central control system. The driving parameters specifically include the current acceleration, the rate of change of vehicle speed, and the brake pedal depth.

[0038] The judgment module is used to determine whether the driving parameters match the driving habits corresponding to the driving identity information;

[0039] The execution module is used to, if not, obtain traffic control information of the vehicle within a preset range based on the driving road type pre-identified by the vehicle central control system, input the traffic control information into the prediction model pre-trained by the vehicle central control system in real time, and generate driving behavior data of the vehicle through the prediction model. The driving road type specifically includes fast passage roads, urban passage roads, and complex environment roads, and the traffic control information specifically includes road traffic restriction, speed time control, and emergency control.

[0040] The second judgment module is used to determine whether the driving behavior data conforms to the driving indicators of the driving identity information;

[0041] The second execution module is used to, if the conditions are not met, compare the group driving behavior reference data with the driving behavior data based on the group driving behavior reference data pre-collected from the driving platform by the vehicle central control, calculate the difference value between the vehicle and the group baseline, identify the corresponding abnormal driving behavior in the driving behavior data, and generate an abnormal warning prompt for the driver through the vehicle based on the abnormal driving behavior. The group driving behavior reference data specifically includes vehicle speed distribution, acceleration range, and lane change frequency.

[0042] Furthermore, it also includes:

[0043] The acquisition module is used to acquire the vehicle's location information through the driving platform based on the communication data pre-sent by the vehicle's central control system to the driving platform;

[0044] The third judgment module is used to determine whether the location information is within the geographical range pre-recorded by the vehicle's central control system;

[0045] The third execution module is used to, if so, construct driving behavior rules for the vehicle based on the preset traffic control requirements of the geographical area, mark the corresponding violations during the vehicle's operation according to the driving behavior rules, and dynamically present the violations to the vehicle through the vehicle's central control system. The dynamic presentation specifically includes dashboard image reminders, voice prompts, and steering wheel and seat vibrations.

[0046] Furthermore, the execution module also includes:

[0047] The generation unit is used to generate the control content of the vehicle within the preset range based on the control type of the traffic control information through the vehicle central control system, wherein the control type specifically includes dynamic control and static control;

[0048] The judgment unit is used to determine whether the vehicle has driven out of the preset range;

[0049] The execution unit is used to dynamically map the control content if no, calculate the duration of the impact of each control rule on the vehicle within a preset driving time window based on the real-time driving path of the vehicle, and adaptively filter invalid control information of the control content on the vehicle based on the duration of the impact.

[0050] This invention provides an abnormal warning method and system based on driving behavior, which has the following beneficial effects:

[0051] This invention achieves multi-level judgment of vehicle driving behavior by combining individual driver driving habits, road type, and traffic control information. First, based on the driver's identity information pre-collected by the central control system, real-time driving parameters are detected and compared with individual habits. Only behaviors that significantly deviate from habits are further analyzed to avoid "one-size-fits-all" misjudgments. Combining road type and traffic control information, driving behavior data is generated through a predictive model to further determine whether the behavior is reasonable and ensure the targeted nature of the warning. Finally, abnormal driving behaviors are compared with the group driving behavior baseline to achieve dynamic correction of individual differences. Thus, while fully considering the individual differences of drivers, it effectively reduces the false alarm rate and ensures timely warning of abnormal behaviors, improving driving safety and the practicality of the warning. Attached Figure Description

[0052] Figure 1 This is a flowchart illustrating an embodiment of the abnormal warning method based on driving behavior of the present invention;

[0053] Figure 2 This is a structural block diagram of an embodiment of the abnormal warning system based on driving behavior of the present invention. Detailed Implementation

[0054] It should be understood that the specific embodiments described herein are merely illustrative of the invention and are not intended to limit the invention. The realization of the purpose, functional features, and advantages of the invention will be further explained in conjunction with the embodiments and with reference to the accompanying drawings.

[0055] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of the present invention, and not all of them. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0056] Reference Appendix Figure 1 An abnormal warning method based on driving behavior in one embodiment of the present invention includes:

[0057] S1: Based on the driver identity information pre-collected by the vehicle's central control system, detect the vehicle's driving parameters during the driving process, wherein the driving parameters specifically include the current acceleration, vehicle speed change rate, and brake pedal depth;

[0058] S2: Determine whether the driving parameters match the driving habits corresponding to the driving identity information;

[0059] S3: If not, then based on the driving road type pre-identified by the vehicle central control, obtain the traffic control information of the vehicle within a preset range, input the traffic control information into the prediction model pre-trained by the vehicle central control in real time, and generate the driving behavior data of the vehicle through the prediction model. The driving road type specifically includes fast passage roads, urban passage roads and complex environment roads, and the traffic control information specifically includes road traffic restriction, speed time control and emergency control.

[0060] S4: Determine whether the driving behavior data conforms to the driving indicators of the driving identity information;

[0061] S5: If not, based on the group driving behavior reference data pre-collected from the driving platform by the vehicle central control, the group driving behavior reference data is compared with the driving behavior data, the difference value between the vehicle and the group baseline is calculated, the corresponding abnormal driving behavior is identified in the driving behavior data, and an abnormal warning prompt is generated for the driver through the vehicle according to the abnormal driving behavior. The group driving behavior reference data specifically includes vehicle speed distribution, acceleration range and lane change frequency.

[0062] In this embodiment, the system, based on the driver identity information pre-recorded in the vehicle's central control unit, detects the vehicle's driving parameters during operation using built-in sensors. These parameters specifically include current acceleration, rate of change of vehicle speed, and brake pedal depth. The system then determines whether these driving parameters match the driving habits corresponding to the driver identity information, and executes the corresponding steps accordingly. For example, when the system determines that the vehicle's driving parameters match the driving habits corresponding to the driver identity information, the system considers the driver's current operation consistent with their long-established driving characteristics, belonging to their individual normal driving mode. The system sets the current state to "conforming to driving habits," does not trigger an abnormal warning process, and stores the real-time driving parameters as normal sample data. For long-term optimization of the subsequent prediction model, the matching driving parameters are bound and stored with the driver's identity information. If such behavior is recorded multiple times in different environments, the stability of the prediction model can be further enhanced. Furthermore, subsequent computationally expensive prediction model calculations and group baseline comparisons are no longer performed; only low-power real-time monitoring is maintained, saving system resources. For example, when the system determines that the vehicle's driving parameters do not match the driving habits corresponding to the driver's identity information, the system considers the driver's current operation inconsistent with their long-established driving characteristics, detecting some rare driving behaviors. The system will activate the camera devices in the vehicle's central control unit, using these cameras to identify the driving road type, specifically including expressways. For roads classified as general traffic, urban traffic, and roads in complex environments, the system acquires traffic control information for vehicles within a pre-defined range on the driving platform. This traffic control information includes road traffic restrictions, speed limits during specific times, and emergency controls. This information is then input in real-time into a pre-trained prediction model in the vehicle's central control system. The prediction model generates driving behavior data. When driving parameters do not match the driving habits corresponding to the driver's identity information, the system can promptly identify rare or abnormal driver behaviors. This deviation detection method, based on individual characteristics, avoids the "one-size-fits-all" problem caused by applying a uniform threshold to all drivers in traditional methods. This allows for more accurate detection of potential dangerous behaviors and effectively improves the sensitivity of driving behavior recognition. The system enhances sensitivity and personalization. By incorporating pre-identified road types and traffic control information within a preset range, the system can perform secondary verification of deviation behaviors within the environmental context. This avoids misjudging behavioral changes caused by external environmental factors as abnormalities, making the warning results more consistent with actual driving scenarios and improving the rationality and practicality of the system's judgment. Furthermore, by inputting traffic control information into the prediction model and generating driving behavior data, the system can not only dynamically reconstruct the driver's behavioral characteristics in specific environments but also provide a reliable input basis for subsequent anomaly judgment and warnings. Through intelligent analysis of the model, the system can optimize the warning effect under the dual dimensions of individual differences and environmental constraints, thereby effectively reducing the false alarm rate and improving the overall driving safety assurance capability.The system then determines whether the driving behavior data conforms to the driving indicators of the driver's identity information and executes the corresponding steps accordingly. For example, when the system determines that the vehicle's driving behavior data conforms to the driving indicators of the driver's identity information, the system considers that the driver's current behavior is not only normal at the driving parameter level, but also conforms to the individual driving indicators after correction by road type, traffic control information, and prediction model. This indicates that the driving state is safe and stable, and the system will not trigger an abnormal warning, but will continue to monitor in real time. At the same time, the driving behavior data that conforms to the driving indicators will be stored in the individual driving profile as a reference for subsequent model optimization and personalized baseline updates. Furthermore, if such behavior is verified as reasonable multiple times in different environments, the system can gradually increase the weight of this behavior pattern, making the model closer to the driver's real habits. This eliminates the need to perform subsequent complex steps such as group baseline comparison, reducing computational resource consumption and improving system efficiency. Conversely, when the system determines that the vehicle's driving behavior data does not conform to the driving indicators of the driver's identity information, the system considers the driver's current behavior to be abnormal at the driving parameter level and the driving state to be unstable. The system will compare the group driving behavior reference data pre-collected from the driving platform by the vehicle's central control system with the driving behavior data and calculate... The difference between the vehicle and the group baseline is used to classify abnormal driving behaviors in the driving behavior data. Based on different abnormal driving behaviors, the system generates abnormal warning prompts for the driver. When the driving behavior data fails to meet the driving indicators of the driver's identity information, the system can determine that the driving state is unstable. At this time, the group driving behavior reference data is introduced as a baseline to further verify whether there is a real dangerous behavior in the case of individual abnormality, thereby avoiding misjudgment caused by relying solely on individual models and improving the accuracy of abnormal behavior identification. At the same time, by calculating the difference between the vehicle's driving behavior data and the group baseline, the system can quantify the degree of deviation of the driver's behavior from the normal range of the group, and classify different types of abnormal driving behaviors based on this. This method avoids classifying all deviations as abnormal indiscriminately, helps to distinguish between minor deviations and serious dangerous behaviors, and makes the abnormal judgment more hierarchical and reasonable. After completing the abnormal behavior classification, the system can generate graded warning prompts according to different abnormality types, reminding the driver through visual, audible, or vibration means. This graded warning avoids the interference caused by excessive reminders and can intervene in time when there is real danger, improving the driver's perception and response to risks, thereby effectively ensuring vehicle driving safety.

[0063] It should be noted that, based on the driving road type pre-identified by the vehicle's central control system, traffic control information for the vehicle within a preset range is obtained. This traffic control information is then input into the pre-trained prediction model of the vehicle's central control system in real time. Through this prediction model, driving behavior data of the vehicle is generated. A specific example is as follows:

[0064] Example 1: Speed ​​limit control scenario on urban main roads.

[0065] Road recognition: The system uses the central control map and positioning module to identify that the current vehicle is traveling on "urban traffic roads" and locates the specific road segment as XX main road;

[0066] Traffic control information is obtained by retrieving traffic control information provided by the traffic management platform within a 2-kilometer radius of the vehicle's current location.

[0067] "During weekday morning rush hour (7:00–9:00), the speed limit on this section of road is 50 km / h." The current time is 7:30, and the rule is in effect.

[0068] Driving parameter detection,

[0069] Real-time vehicle speed: 65km / h;

[0070] Acceleration: 1.2 m / s² 2 (The vehicle is in a state of slight acceleration);

[0071] Brake pedal depth: 0 (driver not applying the brakes);

[0072] The data fusion input model uses the aforementioned detection values ​​and control information as feature variables to input into the pre-trained prediction model at the central control station; the input variables include:

[0073] Vehicle speed = 65km / h;

[0074] Speed ​​limit rule = 50km / h;

[0075] Road type = urban road;

[0076] Time = 7:30 (within the effective speed limit range);

[0077] Acceleration = 1.2 m / s² 2 ;

[0078] Braking depth = 0;

[0079] The predictive model calculates the speed by first comparing the vehicle speed with the speed limit and concluding that the speed is 15 km / h higher than the speed limit. Then, it combines the acceleration and the lack of braking to determine that the speeding is likely to continue.

[0080] Driving behavior data generation, model output results

[0081] Driving behavior type: speeding;

[0082] Risk level: Medium;

[0083] Triggering condition: Vehicle speed 65km / h > speed limit 50km / h;

[0084] Trend analysis: The vehicle is in a state of continuous acceleration, which increases the risk;

[0085] The recommended action from the vehicle's central control system at this moment is to prompt the driver to slow down immediately.

[0086] Example 2: Highway construction control scenario.

[0087] Road recognition: The system identifies that the current vehicle is traveling on "fast-passage roads" (highways);

[0088] Traffic control information was obtained from the system within 5 kilometers in front of the vehicle: "Due to construction, the right lane is closed, and all vehicles need to change lanes in advance."

[0089] Driving parameter detection,

[0090] Real-time vehicle speed: 95km / h;

[0091] Current lane: Right lane;

[0092] Acceleration: 0.3 m / s² 2 (Basically uniform speed);

[0093] Data fusion input model, input variables:

[0094] Vehicle speed = 95km / h;

[0095] Current lane position = right;

[0096] Traffic control information = Right lane closed;

[0097] Road type = Expressway;

[0098] The predictive model compares the vehicle's lane position with the construction control rules and finds that the vehicle is currently in a state of "about to enter a controlled and closed lane"; combined with the vehicle speed of 95km / h, it is determined that there will be a strong risk if the vehicle does not change lanes.

[0099] Driving behavior data generation, model output results

[0100] Driving behavior type: Lane occupancy risk;

[0101] Risk level: High;

[0102] Triggering condition: Lane location does not comply with traffic control requirements;

[0103] The recommended action from the vehicle's central control system at this moment is to prompt the driver to immediately change lanes to the left.

[0104] Example 3: Emergency control scenario on mountain roads.

[0105] Road recognition: The system identifies that the vehicle is currently traveling on "complex environment roads" (two-way two-lane road sections in mountainous areas);

[0106] Traffic control information was obtained 3 kilometers ahead of the vehicle, showing the following real-time control information: "Due to a landslide, this section of the road is temporarily closed and vehicles are prohibited from passing."

[0107] Driving parameter detection,

[0108] Real-time vehicle speed: 40km / h;

[0109] Acceleration: –0.2m / s² 2 (The driver is slowing down slightly);

[0110] Brake pedal depth: 20% (light braking);

[0111] Data fusion input model, input variables:

[0112] Vehicle speed = 40km / h;

[0113] Road type = Complex mountain roads;

[0114] Traffic control rule = Road ahead is closed;

[0115] Vehicle status = decelerating;

[0116] The predictive model identifies a "no passage" rule ahead of the vehicle and, combined with the fact that the vehicle is currently decelerating, determines that the driver's behavior is reasonable, but the route still needs to be replanned.

[0117] Driving behavior data generation, output results:

[0118] Driving behavior type: Normal deceleration but need to detour;

[0119] Risk level: Low (because the driver is already braking);

[0120] The recommended action from the vehicle's central control system at this moment is to prompt the driver to stop in advance and automatically plan an alternative route.

[0121] It should be added that, based on the group driving behavior reference data pre-collected from the driving platform by the vehicle's central control system, the group driving behavior reference data is compared with the driving behavior data to calculate the difference value between the vehicle and the group baseline. Corresponding abnormal driving behaviors are then identified in the driving behavior data. Based on these abnormal driving behaviors, an abnormal warning prompt is generated for the driver through the vehicle. Specifically:

[0122] 1. Select group data in the same scenario. Based on the scene tags already identified by the vehicle (road type, time period, weather / road surface, area and speed limit level, traffic density, vehicle type, etc.), select group driving behavior reference data in the same scenario from the driving platform to form a group baseline sample set N≥500.

[0123] 2. Construct a population baseline and establish statistical baselines for core indicators:

[0124] mean μ v Standard deviation σ v , quantile P5 / P50 / P95,

[0125] Acceleration range μ a , σ a (Can be divided into major axis / lateral axis),

[0126] Lane change frequency (per kilometer): μ lc , σ lc ,

[0127] Brake pedal depth distribution: quantiles or mean / standard deviation

[0128] Simultaneously estimate the multidimensional covariance matrix (for joint deviation determination);

[0129] 3. Process the judged driving behavior data (time window). Within a fixed time window W (e.g., 30 seconds or 1 km driving distance), extract feature vectors with the same caliber as the group baseline from the real-time vehicle flow. The formula is as follows:

[0130] 4. Perform standardized deviation calculation for a single indicator (Z score / quantile ratio). The formula for the normal approximation indicator is as follows:

[0131]

[0132] For skewed / upper bound indices (such as braking depth), quantile positions are used: q brake =rank(brake) p95 )∈[0,1];

[0133] 5. Calculate the multidimensional dissimilarity value (Mahanobis distance) and extract key continuous features. Calculate the Mahalanobis distance D 2 =(x k -μ) T-1 (x k -μ), under degrees of freedom k (eigennumber), with Use distribution thresholds to determine significance (e.g., 95th / 99th percentile);

[0134] 6. Conduct a comprehensive difference score (0–100), and assign the tail probability p of each single indicator. i Significance of p with Mahalanobis distance M Fusion: The weights satisfy w M +∑w i =1, the larger the value, the more abnormal it is;

[0135] 7. Anomaly type classification (based on rules + statistical thresholds), with rules taking priority. If a "hard violation" exists (such as exceeding P99.5 or violating mandatory controls), it is directly marked as the corresponding anomaly type, and statistical stratification is performed:

[0136] S < 40: Normal / Slight deviation

[0137] 40≤S<70: Moderately abnormal (suggestive level)

[0138] S≥70: Significantly abnormal (warning / intervention level)

[0139] Simultaneously record the "dominant anomaly factor" (which one contributes the most);

[0140] 8. Early warning generation and anti-shake mechanism: Mapping anomaly types to tiered early warnings.

[0141] Low: Visual cue

[0142] Chinese: Voice + HUD

[0143] High: Seat / steering wheel vibration and coordinated lane keeping / torque limiting, etc.

[0144] Jitter suppression / hysteresis: Abnormalities require continuous T. hold The system will only upgrade if the error occurs repeatedly within M windows (e.g., 2–5 seconds) or if the error is repeated within M windows, thus avoiding false alarms due to transient noise.

[0145] 9. Write back the results and learn, recording the degree of difference, triggering factors and environmental labels; if it is judged as "reasonable deviation" multiple times, enter the individual model adaptive process (not described in detail in this step);

[0146] Specific examples are as follows:

[0147] Example A: Urban traffic, weekday morning rush hour, sunny, dry road surface.

[0148] Group baseline (same scenario N=12,000),

[0149] μ v =38km / h, σ v =8,

[0150] μ a =0.6m / s 2 , σ a=0.4 (using longitudinal acceleration converted from P95 caliber to equivalent mean / variance for approximation),

[0151] μ lc =0.20 / km, σ lc =0.15 / km,

[0152] The covariance matrix ∑ is approximately diagonal after estimation;

[0153] Vehicle time window characteristics (approximately 30 seconds),

[0154] ν avg =58km / h, a p95 =1.4m / s 2 lc perkm =0.70 / km, brake p95 =65% (P95 of the population = 55%)

[0155] Single indicator Z / quantile,

[0156] z v = (58-38) / 8 = 2.50,

[0157] z a =(1.4-0.6) / 0.4=2.00,

[0158] z lc =(0.70-0.20) / 0.15=3.33,

[0159] Braking position:

[0160] Mahalanobis distance (using a diagonal approximation, for illustration)

[0161] Degrees of freedom k = 3 Since 21.34 > 16.27, it is judged as a highly significant deviation (p < 0.05). M <0.001),

[0162] Overall score (with weighted examples)

[0163] Let w M =0.5, w ν =0.2, w a =0.15, w lc =0.15,

[0164] Single indicator tail probability approximation:

[0165] Therefore, S≈100×(0.5(1-0.001)+0.2(1-0.006)+0.15(1-0.023)+0.15(1-0.0004))≈96.7,

[0166] Conclusion: Significant anomaly (high level), with dominant factors being: lane change frequency, high speed, and high braking depth. The warning level is high (voice + HUD + steering wheel vibration), and the recommendation is to "reduce speed and minimize frequent lane changes."

[0167] Example B: Fast-moving traffic, highway, light rain at night, slippery conditions.

[0168] Group baseline (N=8,500 in the same scenario),

[0169] μ v =92km / h, σ v =12,

[0170] μ a =0.8m / s 2 , σ a =0.5,

[0171] μ lc =0.12 / km, σ lc =0.08 / km,

[0172] Vehicle time window characteristics.

[0173]

[0174] Mahalanobis distance (illustrated)

[0175] D 2 ≈(-0.58) 2 +1.6 2 +2.875 2 ≈0.34 + 2.56 + 8.27 = 11.17

[0176]

[0177] If the value falls between the 95% and 99% thresholds (slightly below 0.99), it is considered a medium-to-high deviation.

[0178] Taking into account both the overall score and the early warning, with the same weighting as above, we calculate S≈78 (illustrated).

[0179] Therefore, the conclusion is: moderate to high abnormality, mainly caused by frequent lane changes, the warning level is medium (voice + HUD), and it is recommended to "reduce lane changes and maintain a stable speed". If the threshold is exceeded for two consecutive windows, the warning level will be upgraded.

[0180] Example C: Complex environment (mountain curves, multiple sudden braking incidents involving a group of people).

[0181] The baseline for the group is that on mountainous winding roads, after rainfall, the group's brake pedal depth P95 is higher (e.g., P95 = 70%), and the average vehicle speed is lower (μ). ν =32, σ ν =7),

[0182] Vehicle time window characteristics.

[0183]

[0184] brake p95 =68% (lower than the group P95 = 70%)

[0185]

[0186] The judgment is that all individual items are within the reasonable range of the group, and the Mahalanobis distance is...

[0187] Therefore, the conclusion is that it is not abnormal. Even if the model indicates that there are "many sudden braking" scenarios, no warning should be issued in this scenario to effectively avoid false alarms.

[0188] In summary, in the above example scenarios, the correct scenario was selected before comparison. The difference value was given by using the tail probability of a single indicator plus multidimensional Mahalanobis distance. At the same time, rule priority and statistical stratification were used to determine the anomaly type. In addition, noise was reduced by combining shake suppression and graded warning. Three typical results were shown: significant anomaly, marginal anomaly, and reasonable deviation.

[0189] In this embodiment, before step S3 of generating the vehicle's driving behavior data, the method further includes:

[0190] S301: Based on the communication data pre-sent by the vehicle central control to the driving platform, obtain the vehicle's location information through the driving platform;

[0191] S302: Determine whether the location information is within the geographical range pre-recorded by the vehicle's central control system;

[0192] S303: If so, then according to the traffic control requirements preset in the geographical area, the driving behavior rules of the vehicle are constructed, and according to the driving behavior rules, the corresponding violations are marked during the driving process of the vehicle, and the violations are dynamically presented to the vehicle through the vehicle central control. The dynamic presentation specifically includes dashboard image reminders, voice prompts and steering wheel seat vibrations.

[0193] In this embodiment, the system obtains the vehicle's location information through the driving platform based on communication data pre-sent by the vehicle's central control unit. The system then determines whether the location information is within the geographical area pre-recorded by the vehicle's central control unit and executes corresponding steps accordingly. For example, if the system determines that the vehicle's location information is not within the geographical area pre-recorded by the vehicle's central control unit, the system considers the vehicle to be in an unfamiliar or uncovered area. Its driving environment may be inconsistent with the driver's long-term driving habits and lacks corresponding group baseline or control rule data support. The system will increase the trigger threshold of the warning algorithm to avoid excessive warnings in unknown environments, only issuing alerts for "strong rule-based risks" (such as speeding, running red lights, and driving against traffic), while not immediately warning for minor deviations. Simultaneously, the system calls real-time traffic control information, road type characteristics, and group data for the new area through the driving platform or cloud. If the platform lacks readily available driving reference data, it temporarily uses nationwide / regional baseline data as a substitute and records the driving data in the new region, marking it as a "new region behavior sample." If the vehicle repeatedly operates in the region, a corresponding extended dataset of individual driving habits is gradually built and stored in the regional database of the vehicle's central control system. For example, when the system determines that the vehicle's location information is within the pre-collected regional range of the vehicle's central control system, the system will consider the vehicle to be in a known area. The driving environment is consistent with the driver's long-term driving habit data. The system will construct driving behavior rules for the vehicle based on the pre-set traffic control requirements of the regional range. Based on these driving behavior rules, the system will mark the corresponding violations during the vehicle's operation and dynamically present different violations to the vehicle through the vehicle's central control system. The dynamic presentation includes dashboard image reminders, voice prompts, and steering wheel and seat vibrations.When the vehicle's location is within the area recorded by the central control system, it indicates that the system has grasped the road characteristics, traffic rules, and driving environment of that area. Based on this premise, the system establishes driving behavior rules that fully reflect the driver's long-term habits in familiar areas, making the judgment of violations or abnormal behaviors more accurate. This avoids false alarms or unnecessary warnings due to environmental differences, improving the practicality and reliability of the warning system. Simultaneously, relying on traffic control requirements within the area, the system can automatically compare and verify driving behavior during vehicle operation. Once behaviors such as speeding, illegal lane changes, or illegal overtaking occur, the system can mark and output violation information in real time, displayed on the dashboard. Multimodal presentation methods, including image alerts, voice prompts, and seat / steering wheel vibrations, ensure that drivers receive timely warning signals regardless of their attention level, reducing risks caused by driver negligence. Furthermore, within a known geographical area, the system can gradually optimize driving behavior rules by combining individual drivers' historical behavior data with traffic control requirements. For example, for common driving habits such as minor speeding or sudden acceleration, the system can distinguish between "reasonable deviations" and "genuine violations" through long-term data recording and comparison. This dynamic adjustment not only enhances the personalization of warnings but also provides stable training samples and data support for subsequent cross-regional expansion of driving models.

[0194] In this embodiment, step S3, which involves obtaining traffic control information for the vehicle within a preset range based on the road type pre-identified by the vehicle's central control system, further includes:

[0195] S31: Based on the control type of the traffic control information, the vehicle central control system generates the control content of the vehicle within the preset range, wherein the control type specifically includes dynamic control and static control;

[0196] S32: Determine whether the vehicle has left the preset range;

[0197] S33: If not, the control content is dynamically mapped, and the duration of the influence of each control rule on the vehicle within a preset driving time window is calculated based on the real-time driving path of the vehicle. Based on the duration of the influence, the invalid control information of the control content on the vehicle is adaptively filtered.

[0198] In this embodiment, the system generates control measures based on traffic control information, specifically dynamic and static control. Through the vehicle's central control system, it generates control measures for the vehicle within a pre-defined range. The system then determines whether the vehicle has left the pre-defined range and executes corresponding steps accordingly. For example, if the system determines that the vehicle has not left the pre-defined range, it considers the vehicle's behavior still subject to established traffic restrictions. The system-generated control measures (whether dynamic controls such as temporary road closures or time-based speed limits, or static controls such as fixed restricted areas) remain valid, and the system maintains current control requirements, such as speed limits, no-entry restrictions, or detour prompts. During vehicle movement, the system continuously compares the vehicle's driving parameters with the control measures. The system requires that potential violations be flagged. If the control is dynamic (e.g., road closures due to accidents, temporary speed limits), the system continuously updates rules from the traffic platform to avoid outdated alerts due to changes in control conditions. If the control is static (e.g., fixed restricted areas), the existing rules remain unchanged, only monitoring for vehicles entering at risk of violation. Furthermore, if a vehicle's behavior violates control measures (e.g., exceeding dynamic speed limits, approaching a restricted area), the system immediately issues a warning via dashboard image, voice prompts, or steering wheel / seat vibration. For example, if the system determines that a vehicle has left a pre-defined area, it considers the traffic restrictions on the vehicle's behavior needing to be updated. The system dynamically maps the control measures to the vehicle's behavior based on the vehicle's position. The system calculates the duration of each traffic control rule's impact on the vehicle within a pre-set driving time window based on the vehicle's real-time driving path. Based on different impact durations, it adaptively filters out invalid control information. When a vehicle leaves the preset area, the system automatically refreshes traffic restrictions and dynamically maps the control content. This mechanism prevents vehicles from continuing to be restricted by outdated control information that is no longer applicable (e.g., triggering a speeding warning even after leaving a speed-limited area), thereby reducing redundant warnings and error messages, ensuring the accuracy and real-time nature of warning information, and increasing driver trust in the system. Furthermore, the system not only simply removes controls from old areas but also calculates the impact of each control rule within the driving time window based on the vehicle's real-time driving path. The system adaptively filters rules based on their duration of impact, ensuring that short-term invalid control information (such as speed limits for completed temporary construction projects) is no longer prompted, while rules still within their effective time periods (such as peak-hour restrictions and long-term no-entry zones) are retained. This achieves refined and differentiated control, improving the system's usability in complex traffic scenarios. Furthermore, by dynamically filtering and refreshing control information, drivers receive prompts highly relevant to their current driving route, rather than being distracted by a large number of irrelevant or invalid rules. This optimization not only helps drivers focus on genuine traffic restrictions and reduce the risk of violations but also reduces unnecessary reminders that distract the driver's attention. Thus, while ensuring safety, it also improves the overall driving experience and the user-friendliness of human-computer interaction.

[0199] It should be noted that the control content is dynamically mapped. Based on the vehicle's real-time driving path, the duration of each control rule's impact on the vehicle within a preset driving time window is calculated. Based on the duration of the impact, invalid control information of the control content for the vehicle is adaptively filtered. Specifically:

[0200] First, based on the vehicle's real-time location and driving path prediction results, traffic control rules are mapped to road segments that the vehicle may pass through in the future. This mapping relationship not only considers the current location but also combines navigation paths, historical driving trajectories, and road connectivity to ensure that the system can identify potential control points in advance and calculate the duration of impact. For each mapped control rule, the system combines the vehicle's current speed, estimated driving speed range, and distance to the control point to calculate the effective duration of the control rule's effect on the vehicle within a preset time window (such as the next 10 minutes or 5 kilometers of travel). If the vehicle is expected to enter the control area within the time window, the system calculates the continuous constraint time that the vehicle may be subject to within that area. If the vehicle enters outside the time window or the control ends, the impact duration is 0, and invalid control information is adaptively filtered out. When the impact duration is 0 or less than the system's set threshold (such as less than 5 seconds, which has no actual impact on driving behavior), the control rule is determined to be invalid for the vehicle, and the system will remove it to avoid redundant prompts. Control content that is still within the effective duration is retained, and warnings or prompts are output in real time to ensure that the system prompts are consistent with the actual constraint situation of the vehicle.

[0201] Specific examples are as follows:

[0202] Scenario: City road, vehicles are traveling on the main road, the system is set to a time window of the next 10 minutes.

[0203] 1. Current vehicle status

[0204] Real-time speed: 60km / h

[0205] Driving direction: East

[0206] Distance from construction site: 3km

[0207] Current time: 9:55

[0208] System geographical scope: Core urban area; control information has been collected.

[0209] 2. Traffic control information (control type),

[0210] Speed ​​limit 50km / h (static control, effective for the entire section)

[0211] The right lane will be closed due to construction (dynamic traffic control, construction distance 3km, construction expected to last 20 minutes).

[0212] Peak hour traffic restriction rules (dynamic control, certain license plate numbers are prohibited from entering a specific area from 10:00 to 11:00).

[0213] 3. Dynamically map and control content.

[0214] The system maps three rules to the vehicle's driving path for the next 10 minutes:

[0215] Speed ​​limit 50km / h → Current road, effective immediately.

[0216] Construction closure → 3km later, estimated arrival time is 3 minutes.

[0217] Peak hour traffic restriction → Expected to enter in 15 minutes, outside of the designated time window.

[0218] 4. Calculate the duration of the impact of each control rule.

[0219] Speed ​​limit 50km / h: Effective time ≥10 minutes (continuously effective throughout the entire time window),

[0220] Construction closure: Effective time = Remaining construction time - Arrival time = 20 - 3 = 17 minutes → Effective within the time window.

[0221] Peak hour traffic restriction: If the effective time is 0 (exceeding the time window), the restriction is invalid.

[0222] 5. Adaptively filter invalid control information.

[0223] The system filters out the effective control content for vehicles within the specified time window:

[0224] Retained: Speed ​​limit 50km / h, construction closure.

[0225] Removed: Peak hour traffic restrictions (currently will not affect driving).

[0226] 6. System dynamic output,

[0227] Speed ​​limit reminder: The instrument panel displays "Speed ​​limit 50km / h" and can be accompanied by a voice prompt "Please maintain the speed limit".

[0228] Construction notice: When you are 1km away from the construction site, a voice prompt will say, "Construction ahead, right lane closed, please change lanes in advance," and the steering wheel or seat may vibrate slightly to remind you.

[0229] Peak hour traffic restrictions: No notice will be given to avoid misleading drivers.

[0230] In summary, in the examples above, drivers only receive valid traffic control information within the current route and time window, avoiding interference from future or irrelevant control rules. The system can calculate the duration of construction impact in advance, reasonably arrange warning times, improve safety, and eliminate short-term invalid rules to reduce driver distraction or repeated prompts.

[0231] In this embodiment, step S5, based on the group driving behavior reference data pre-collected from the driving platform by the vehicle central control system, further includes:

[0232] S51: Based on the pre-established communication connection between the vehicle central control and the driving platform, identify the source channels of the group driving behavior reference data, wherein the source channels specifically include data from official traffic management platforms, data collected by car manufacturers themselves, and data from Internet travel platforms;

[0233] S52: Determine whether the data format of the source channels is uniform;

[0234] S53: If not, then obtain the temporal continuity of the group driving behavior reference data, detect the missing data values ​​of the group driving behavior reference data based on the temporal continuity, generate the corresponding missing ratio, and correct the dynamic timeliness of the group driving behavior reference data in real time based on the missing ratio.

[0235] In this embodiment, the system identifies the source channels of group driving behavior reference data based on the pre-established communication connection between the vehicle's central control system and the driving platform. These source channels specifically include data from official traffic management platforms, data collected by vehicle manufacturers, and data from internet-based mobility platforms. The system then determines whether the data formats of these source channels are consistent, and executes corresponding steps accordingly. For example, when the system determines that the data formats of the source channels for group driving behavior reference data are consistent, it considers that the data can be directly interfaced and processed without additional data cleaning or conversion, resulting in high data compatibility and integrity. The system directly merges and aggregates data from different channels to build a baseline database covering a wider range of drivers. Simultaneously, it directly calls data in a consistent format into the prediction model for difference calculation and group baseline verification, omitting the format conversion step, reducing processing latency, and improving real-time performance. This allows for rapid comparison of individual driving behavior with group reference data, improving the accuracy and response efficiency of warnings. Conversely, when the system determines that the data formats of the source channels for group driving behavior reference data are inconsistent, it considers that the data from these source channels may have data delays or missing data. The system then obtains the temporal continuity of this group driving behavior reference data and, based on different temporal continuity, detects... The system identifies missing values ​​in the group driving behavior reference data and generates corresponding missing proportions. Based on these missing proportions, the dynamic timeliness of the group driving behavior reference data is corrected in real time. When the system detects that the data formats from different sources are inconsistent, it analyzes the temporal continuity of the data and identifies missing values ​​to calculate the missing proportion in real time. This method avoids distortion of the group baseline caused by data delays or missing values, ensuring that the data input to the prediction model always maintains high integrity and reliability, thereby avoiding false warnings or misjudgments caused by data gaps. Furthermore, the system not only detects missing values ​​but also dynamically corrects the timeliness of the group driving behavior reference data based on the missing proportion. This dynamic correction mechanism ensures that the group baseline can be adaptively updated even when the data source fluctuates, improving the system's compatibility with multi-source heterogeneous data and making it more suitable for high-frequency calculations and warning needs in real-time driving scenarios. Through correction based on the missing proportion, the group reference data can more realistically reflect the distribution characteristics of driving behavior, avoiding comparisons between individual driving behavior and "distorted" group data, thus reducing the false alarm rate. This mechanism can significantly improve the accuracy of abnormal warnings, ensuring that the warning information is both consistent with real-time driving scenarios and has practical reference value, enhancing the system's practicality in complex road environments.

[0236] It should be noted that the time continuity of the group driving behavior reference data is obtained. Based on the time continuity, missing data values ​​in the group driving behavior reference data are detected, and a corresponding missing ratio is generated. Based on the missing ratio, the dynamic timeliness of the group driving behavior reference data is corrected in real time. A specific example is as follows:

[0237] Scenario assumptions: The system collects reference data on group driving behavior at a sampling frequency of 1Hz (i.e., once per second) for a duration of 1 minute. The sampled fields include vehicle speed (km / h) and acceleration (m / s²). 2 Lane change frequency (times / minute) and lane change frequency are data from official traffic management platforms, vehicle manufacturers' self-collected terminals, and internet mobility platforms, respectively.

[0238] 1. Data volume calculation: Based on a sampling frequency of 1Hz, 60 time points × 3 fields = 180 data values ​​should be collected per minute, which constitutes a complete reference data matrix for group driving behavior;

[0239] 2. Regarding the actual data collected, system detection revealed inconsistencies in data from different sources.

[0240] Official traffic management platform: Data is available for all 60 time points (complete).

[0241] Automakers' own procurement terminals: only received data from 48 time points (12 seconds without a record).

[0242] The ride-hailing platform has vehicle speed data for all 60 time points, but acceleration data is missing for some time points (6 in total).

[0243] At this point, the overall data matrix actually received:

[0244] Vehicle speed: 60 points complete.

[0245] Acceleration: 54 points (6 missing)

[0246] Lane change frequency: 48 points (12 missing)

[0247] The total number of valid data points is 60 + 54 + 48 = 162, which is 18 fewer than the required 180.

[0248] 3. Missing item detection and missing item ratio calculation: The system first checks the temporal continuity based on the timestamp sequence.

[0249] Between 20 and 32 seconds, no data was uploaded from the automaker's own data collection terminal (a continuous 12-second gap).

[0250] At 40s, 50s, and 55s, the ride-hailing platform lacks acceleration data.

[0251] Calculate the missing percentage:

[0252] Acceleration field: 6 ÷ 60 = 10% missing.

[0253] Lane change field: 12 ÷ 60 = 20% missing.

[0254] Overall missing percentage: 18 ÷ 180 = 10%;

[0255] 4. Dynamic and timely correction: The system sets a threshold based on the proportion of missing data.

[0256] Less than 10% → direct interpolation correction.

[0257] 10%~30% → Interpolation correction and reduction of the field's weight.

[0258] Greater than 30% → Marked as low confidence, not included in core baseline calculation.

[0259] In this example, the acceleration field (missing 10%) can be corrected using linear interpolation.

[0260] For example, at t=40s, the vehicle speed is 52km / h, but the acceleration is missing; at t=41s, the acceleration is 1.5m / s². 2 Based on the speed difference and time interval, the system calculates that the acceleration at t = 40s is approximately 1.4m / s². 2 ,

[0261] Lane change field (missing 20%): Completed using weighted average; for example, if 4 lane changes were detected in the past 48 seconds, with an average frequency of ≈0.083 times / s, the missing interval will be filled in using this average, and the lane change weight will be reduced in subsequent baseline calculations;

[0262] 5. Correction results and dynamic adjustment: After correction, the group driving behavior reference data is restored to a complete set of 60 time points. However, the system also marks "20% missing, corrected" for "lane change frequency". When compared with individual driving behavior, it automatically reduces its contribution coefficient to the difference calculation (e.g., from 0.3 to 0.2). The resulting group driving baseline is more realistic and robust, which utilizes the corrected data and avoids model deviation caused by excessive missing proportion.

[0263] In summary, in the examples above, the system ensures the temporal continuity of the reference data through missing data detection and interpolation correction. At the same time, the correction process is completed within seconds, without affecting real-time driving warnings. Furthermore, the missing data ratio and weight adjustment mechanism reduces the impact of missing data on the accuracy of the group baseline.

[0264] In this embodiment, step S2, which determines whether the driving parameters match the driving habits corresponding to the driving identity information, further includes:

[0265] S21: Based on the external driving environment pre-detected by the vehicle's central control system, construct the external environment information corresponding to the external driving environment, wherein the external environment information specifically includes meteorological and natural environment, road traffic conditions, and dynamic traffic participants;

[0266] S22: Determine whether the external environment information belongs to a preset normal driving scenario;

[0267] S23: If not, then based on the environmental type of the external driving environment, continuously collect the vehicle's deviation behavior within a preset driving time window, identify the trigger frequency of the deviation behavior, and dynamically mark the deviation behavior as an observation state based on the trigger frequency.

[0268] In this embodiment, the system constructs external environment information corresponding to the external driving environment on a pre-installed display screen in the vehicle, based on the external driving environment pre-detected by the vehicle's central control system. This external environment information specifically includes weather conditions, road traffic conditions, and dynamic traffic participants. The system then determines whether this external environment information belongs to a pre-set normal driving scenario and executes the corresponding steps accordingly. For example, when the system determines that the external environment information corresponding to the external driving environment belongs to a pre-set normal driving scenario, the system considers the current external environment to be within the normal range preset by the vehicle's central control system, without severe weather, special road restrictions, or abnormal traffic behavior. The system will then maintain a normal monitoring mode, monitoring acceleration and vehicle speed. The system routinely compares driving parameters such as rate of change and brake pedal depth without triggering additional warnings or interventions. Simultaneously, it aligns external environmental information with driving habit data corresponding to the driver's identity information, dynamically verifying whether the driver's behavior conforms to an individualized baseline. If there is no significant deviation, it continues to maintain stable monitoring and continuously presents routine environmental information on the vehicle's display screen in a concise and intuitive manner, including normal weather icons, traffic flow indicators, and regular traffic distribution, to help the driver maintain environmental awareness. For example, if the system determines that the external environmental information corresponding to the external driving environment does not belong to the pre-set routine driving scenario, the system will assume that the vehicle's central control system cannot understand the current external environment in real time, potentially leading to adverse conditions. In response to weather conditions, special road restrictions, or abnormal traffic behavior, the system continuously collects vehicle deviation behavior data within a pre-set driving time window, based on the type of external driving environment. It identifies the trigger frequency of these deviations and dynamically marks them as pending observation based on different trigger frequencies. When the external environment deviates from typical driving scenarios, the system proactively identifies potential uncertainties and compensates for the central control system's insufficient perception of the external environment by continuously collecting deviation behavior data. This allows the vehicle to maintain dynamic monitoring of driver behavior even in complex or abnormal environments, enhancing the system's adaptability and robustness. Furthermore, the system does not immediately trigger a warning upon detecting deviation behavior; instead, it first statistically analyzes the deviation data. By marking the trigger frequency as "to be observed," this delayed judgment mechanism effectively avoids false alarms caused by occasional operations (such as sudden braking or lane changing). It ensures that the risk judgment process is only initiated when the deviation behavior shows a high frequency or continuous trend, thereby improving the accuracy and practicality of abnormal warnings. Furthermore, through dynamic marking based on trigger frequency, the system can form a hierarchical risk observation mechanism. It can not only monitor the development trend of deviation behavior in real time, but also dynamically adjust the trigger threshold according to individual driver differences. This method makes the warning system more flexible in the face of uncertain environments, enabling it to detect real risks in a timely manner while avoiding excessive intervention, ultimately improving overall driving safety and user experience.

[0269] It should be noted that, based on the type of the external driving environment, the vehicle's deviation behavior is continuously collected within a preset driving time window. The trigger frequency of the deviation behavior is identified, and based on the trigger frequency, the deviation behavior is dynamically marked as an observation state. A specific example is as follows:

[0270] Scenario assumption:

[0271] Vehicle type: Passenger car

[0272] Route: Urban expressway

[0273] Travel time: 17:30–17:35

[0274] External driving environment type: Complex environment.

[0275] Weather: Light rain, slippery roads, visibility approximately 150m.

[0276] Road: Some sections under construction, speed limit 40km / h.

[0277] Dynamic traffic participants: Surrounding vehicles frequently change lanes, and occasionally pedestrians or bicycles approach the edge of the lane.

[0278] System Settings:

[0279] Time window: 5 minutes

[0280] Lane departure monitoring parameters: emergency braking (brake pedal depth > 70%), lane departure (vehicle lateral deviation from lane centerline > 0.3m), sudden sharp steering (steering wheel angle change rate > 20° / s);

[0281] Step 1: Continuously collect deviation behavior data. The system collects driving parameters in real time over 5 minutes, as shown in Table 1 below.

[0282] Table 1:

[0283] time Emergency braking incident Lane departure event Sudden turning event 0:00–1:00 1 2 0 1:00–2:00 2 3 1 2:00–3:00 1 2 0 3:00–4:00 2 3 1 4:00–5:00 2 2 1

[0284] Cumulative number of events:

[0285] Emergency braking = 1 + 2 + 1 + 2 + 2 = 8 times.

[0286] Lane departure = 2 + 3 + 2 + 3 + 2 = 12 times.

[0287] Sudden turning = 0 + 1 + 0 + 1 + 1 = 3 times;

[0288] Step 2: Calculate the trigger frequency.

[0289] Trigger frequency = Number of events ÷ Time window (minutes)

[0290] Emergency braking frequency = 8 ÷ 5 = 1.6 times / minute

[0291] Lane departure frequency = 12 ÷ 5 = 2.4 times / minute

[0292] Sudden turning frequency = 3 ÷ 5 = 0.6 times / minute;

[0293] Step 3: Dynamically mark the status based on the trigger frequency.

[0294] System-defined thresholds:

[0295] Low risk (<1 time / minute) → Mark as "Normal"

[0296] Medium risk (1–2 times / minute) → Mark as "Under observation"

[0297] High risk (>2 times / minute) → Mark as "abnormal"

[0298] Judgment result:

[0299] Emergency braking: 1.6 times / minute → To be observed.

[0300] Lane departure: 2.4 times / minute → Abnormal

[0301] Sudden turning: 0.6 times / minute → Normal;

[0302] Step 4: System Response and Processing

[0303] Behavior to be observed (sudden braking)

[0304] The system continues to monitor the number of emergency braking events over the next 5 minutes.

[0305] If the frequency continues to increase, it will be escalated to abnormal behavior, triggering an alert.

[0306] Abnormal behavior (lane departure),

[0307] The system immediately triggers vehicle warnings, such as flashing on the dashboard, slight vibration of the steering wheel, or a voice prompt to "keep your lane".

[0308] The data is also recorded for subsequent driving behavior analysis.

[0309] Normal behavior (sudden change of direction),

[0310] The system will not intervene, but will only record the behavior as a driving data reference;

[0311] In summary, as illustrated in the examples above, the system can quantify the degree of driver deviation in complex environments through trigger frequency statistics, which is more reliable than single-event judgment, avoids misjudgments caused by occasional operations, dynamically marks the state to be observed, reduces the false alarm rate, detects high-frequency deviation behaviors in advance, improves the accuracy of warnings, and supports individualized driving behavior analysis to achieve flexible and hierarchical risk management.

[0312] In this embodiment, step S4, which determines whether the driving behavior data matches the driving indicators of the driving identity information, further includes:

[0313] S41: Based on the preset upper limit of the deviation value of the driving indicator, collect the real-time deviation value of the driving behavior data, wherein the real-time deviation value specifically includes speed deviation rate, acceleration deviation and vehicle distance deviation;

[0314] S42: Determine whether the real-time deviation value exceeds the upper limit of the deviation value;

[0315] S43: If so, identify the additional time period of the real-time deviation value, perform weighted summation on the deviation values ​​of each indicator of the driving behavior data, generate a comprehensive deviation score, and construct the tolerance range of the driving behavior data based on the comprehensive deviation score. The additional time period specifically includes daytime, nighttime, peak hours and off-peak hours, and the tolerance range specifically includes compliance with indicators, slight deviation and severe deviation.

[0316] In this embodiment, the system collects real-time deviation values ​​of driving behavior data based on a pre-set upper limit for deviation values ​​in the driving indicators. These real-time deviation values ​​specifically include speed deviation rate, acceleration deviation, and distance deviation. The system then determines whether these real-time deviation values ​​exceed the upper limit and executes corresponding steps accordingly. For example, if the system determines that the real-time deviation values ​​of the driving behavior data do not exceed the upper limit, the system considers the vehicle's current driving behavior to be within the safe range allowed by the driving indicators, and the driver's speed, acceleration, and distance from the vehicle in front all conform to the predetermined individualized baseline or group reference standard. The system will continue to collect and monitor driving behavior data in real time, maintaining the dynamic nature of the deviation values. The system tracks the vehicle's deviation without triggering any abnormal warnings. It compares the real-time deviation value with driving indicators and continuously updates the data on the display screen or in the background log for subsequent trend analysis and predictive model optimization. For deviations within the upper limit, the system presents the data as regular information, such as speed indicators or distance status, through the dashboard or app without interfering with driver operation. However, if the system determines that the real-time deviation of the driving behavior data exceeds the upper limit, it considers the current driving behavior unsafe and identifies additional time periods for the real-time deviation value. These additional time periods include daytime, nighttime, peak hours, and off-peak hours. The deviation values ​​of the indicators are weighted and summed to generate a comprehensive deviation score. Based on different comprehensive deviation scores, a tolerance range for driving behavior data is constructed. The tolerance range specifically includes compliance with the indicators, minor deviation, and severe deviation. When the real-time deviation value exceeds the preset upper limit, the system not only detects a single deviation event but also combines additional time periods to weight and sum the deviation values ​​of each indicator to generate a comprehensive deviation score. This method can comprehensively consider the impact of different environmental conditions on driving behavior, improve the accuracy of judging abnormal driving behavior, and avoid misjudging driving risks due to a single indicator or instantaneous event. Furthermore, based on the comprehensive deviation score, the system constructs a tolerance range including compliance with the indicators, minor deviation, and severe deviation. Through this hierarchical management, driving behavior is no longer a simple binary judgment of "normal / abnormal," but rather a risk level that can be differentiated based on the degree of deviation. This allows the warning system to respond flexibly, taking suggestive measures for minor deviations and immediately triggering strong interventions or warnings for serious deviations. Furthermore, additional time periods are included in the scoring calculation, enabling the system to adjust the deviation judgment criteria according to the driving characteristics of different time periods. For example, when visibility is limited at night, a slightly higher speed deviation is allowed, while in peak-hour congestion, the requirements for vehicle spacing deviation are more stringent. This method achieves personalized and environmentally adaptable driving behavior analysis, improving the system's practicality and safety assurance capabilities under complex and variable road conditions.

[0317] It should be noted that, for the additional time period used to identify the real-time deviation value, the deviation values ​​of each indicator in the driving behavior data are weighted and summed to generate a comprehensive deviation score. Based on the comprehensive deviation score, a tolerance range for the driving behavior data is constructed. A specific example is as follows:

[0318] The scenario is assumed to be,

[0319] Vehicle type: Passenger car

[0320] Route: Urban roads

[0321] Travel time: Nighttime peak hours (18:00–18:05, 5 minutes).

[0322] External environment: Low visibility at night, traffic congestion, and heavy vehicle traffic.

[0323] Driving indicator deviation limit

[0324] Speed ​​deviation rate ≤10%,

[0325] Acceleration deviation ≤ 1.0 m / s² 2 ,

[0326] Vehicle spacing deviation ≤ 0.5m

[0327] Additional time-based weighting (nighttime peak hours),

[0328] Speed ​​deviation rate w1 = 0.4w_1 = 0.4w1 = 0.4,

[0329] Acceleration deviation w2 = 0.3w_2 = 0.3w2 = 0.3,

[0330] The vehicle spacing deviation w3 = 0.3w_3 = 0.3w3 = 0.3;

[0331] Step 1: Collect real-time deviation values. The system collects driving behavior indicators every second, generating a 5-minute time series (300 seconds). For simplicity, the average value per minute is shown in Table 2 below.

[0332] Table 2:

[0333]

[0334] Step 2: Calculate the weighted overall deviation score.

[0335] Formula for overall deviation score per minute:

[0336] The overall deviation score is calculated as follows: w1 × speed deviation rate + w2 × acceleration deviation + w3 × vehicle spacing deviation. The results are shown in Table 3 below.

[0337] Table 3:

[0338] minute Overall deviation score 0-1 0.48+0.30.9+0.3*0.4＝3.2+0.27+0.12＝3.59 1-2 0.411+0.31.1+0.3*0.6＝4.4+0.33+0.18＝4.91 2-3 0.412+0.31.2+0.3*0.7＝4.8+0.36+0.21＝5.37 3-4 0.49+0.31.0+0.3*0.5＝3.6+0.3+0.15＝4.05 4-5 0.413+0.31.3+0.3*0.8＝5.2+0.39+0.24＝5.83

[0339] Step 3: Construct and determine the tolerance range.

[0340] System tolerance range setting,

[0341] Meets the criteria: ≤3 → Normal driving.

[0342] Slight deviation: 3–5 → Warning indicator

[0343] Severe deviation: >5 → Strong warning

[0344] The judgment results are shown in Table 4 below.

[0345] Table 4:

[0346]

[0347]

[0348] Step 4: System response logic,

[0349] For minor deviations, the system continues monitoring, displaying dashboard alerts and providing voice prompts. The data is then entered into an "observation queue" for subsequent trend analysis.

[0350] If a serious deviation occurs, the system will immediately trigger a strong warning (visual, auditory, and tactile) and record the deviation behavior in the driving behavior database, which can be used as a reference for optimizing the driver's individualized baseline model.

[0351] In summary, the examples above demonstrate that the weighted summation of deviation scores takes into account the importance of each indicator and the impact of the additional time period environment on driving behavior, making the judgment closer to the actual risk, achieving continuous dynamic monitoring every minute, improving the timeliness of early warning, and providing layered response based on tolerance range to reduce false alarm rate. It also supports personalized and environmentally adaptive early warning, making the system more reliable in complex driving environments.

[0352] In this embodiment, before step S1 of detecting the vehicle's driving parameters during driving based on the driver's identity information pre-recorded by the vehicle's central control system, the method further includes:

[0353] S101: Based on the identity identifier entered by the user in the vehicle, read the user's identity data from the vehicle's central control unit;

[0354] S102: Determine whether the identity data is included in the vehicle's central control system;

[0355] S103: If not, activate the preset driving text of the vehicle central control, perform driving verification on the user, and construct the driving profile of the user driving the vehicle based on the verification result of the driving verification. The driving profile retains the user's driving event habits, wherein the driving event habits specifically include rapid acceleration, emergency braking, emergency lane change, continuous speeding and lane departure.

[0356] In this embodiment, the system reads the user's identity data from the vehicle's central control unit (NCU) based on the user's registered identity identifier. The system then determines whether this identity data is recorded in the NCU and executes corresponding steps accordingly. For example, if the system determines that the user's identity data is recorded in the NCU, it assumes that the NCU has identified and stored the user's identity information, and the user is a known driver or registered user. The system directly calls the user's individualized driving model, including driving habit baselines, preset driving indicators, and deviation thresholds, for real-time driving behavior monitoring. Simultaneously, it automatically loads user preferences based on the identity information, such as seat position, rearview mirror angle, air conditioning settings, and personalized warning strategies. Furthermore, during driving behavior analysis and abnormal warnings, the system matches the user's identity data with real-time driving behavior data to achieve personalized and accurate risk assessment and alerts. Conversely, if the system determines that the user's identity data is not recorded in the NCU, it assumes that the NCU has not stored the user's identity information, and the user is a new driver. The system activates the pre-set driving text in the NCU and performs a simple driving verification on the user. Based on the verification result... The system constructs a driving profile for each user's vehicle, retaining their driving habits, including rapid acceleration, sudden braking, emergency lane changes, continuous speeding, and lane departure. When the system detects that a user's identity data is not included in the vehicle's central control system, it automatically identifies them as a new driver and verifies them using a pre-set driving text. This mechanism can quickly establish a driving profile for a user's first drive, avoiding system malfunctions due to a lack of historical data, thus improving the system's versatility and intelligent adaptability. During driving verification and subsequent driving, the system records the new user's key driving habits (such as rapid acceleration, sudden braking, emergency lane changes, continuous speeding, and lane departure) in the driving profile. As data accumulates, the system gradually grasps the user's personalized driving characteristics, avoiding a "one-size-fits-all" approach to risk assessment and improving the accuracy of anomaly detection. By constructing driving profiles for new users and continuously retaining their driving habits, the system can conduct long-term tracking and dynamic updates during subsequent driving, achieving comprehensive management of driving behavior from scratch. The establishment of driving profiles also provides a traceable basis for subsequent driving assessments, anomaly warnings, and personalized safety interventions.

[0357] It should be noted that the system activates the preset driving text in the vehicle's central control unit to perform driving verification on the user. Based on the verification result, a driving profile of the user driving the vehicle is constructed, and the user's driving event habits are retained in the driving profile. A specific example is as follows:

[0358] Scenario assumption: User B is driving the vehicle for the first time. The vehicle's central control system detects that the user's identity data has not been recorded, so it activates the "driving verification text". The verification text is designed to be a 15-minute driving test in mixed road conditions, covering urban roads and short expressways, to ensure that it can cover common driving behavior scenarios.

[0359] Step 1: Driving verification begins. The system prompts: Please drive normally for 15 minutes. The system will automatically collect data for driving verification. Sensor data collection content:

[0360] Speed: Vehicle speed sensor (unit: km / h)

[0361] Acceleration: Longitudinal acceleration sensor (unit: m / s²) 2 ),

[0362] Braking: Brake pedal pressure and deceleration,

[0363] Steering wheel angle: Steering wheel angular rate sensor (° / s),

[0364] Lane keeping assist: Lane departure monitoring camera;

[0365] Step 2: Data collection process. During the 15-minute test drive, the system collected the following key events:

[0366] 2nd minute (city road, speed limit 50km / h),

[0367] The user suddenly accelerated, with an instantaneous acceleration of 2.9 m / s². 2 (The system determined this to be a rapid acceleration).

[0368] Continuous speeding for 12 seconds, but not exceeding the system's 15-second threshold → not counted as "continuous speeding".

[0369] 5 minutes (at the intersection of city roads),

[0370] A user brakes suddenly at a red light, achieving a braking depth of 85% and a deceleration of 3.6 m / s². 2 (The system determined this to be an emergency braking incident).

[0371] 7th minute (expressway, speed limit 80km / h),

[0372] When the user changed lanes, the steering wheel speed reached 155° / s, accompanied by lane departure (the system judged this as an emergency lane change + lane departure).

[0373] 10 minutes (continuous section of expressway),

[0374] The user accelerated rapidly again, reaching a peak acceleration of 3.1 m / s². 2 ,

[0375] Immediately afterwards, braked, decelerating at 3.4 m / s². 2 (The system determined this to be a second instance of rapid acceleration followed by a second instance of rapid braking).

[0376] 13th minute (return trip on city roads),

[0377] The user followed the vehicle normally, without speeding or lane departure;

[0378] Step 3: Event statistics and scoring.

[0379] The system compiles statistics on the aforementioned driving behaviors.

[0380] Rapid acceleration: 2 times (average peak speed 3.0 m / s) 2 ),

[0381] Emergency braking: 2 times (average deceleration 3.5 m / s) 2 ),

[0382] Emergency lane change: 1 time (steering wheel speed 155° / s),

[0383] Lane departure: 1 time

[0384] Continuous speeding: 0 times

[0385] The system calculates a comprehensive score (out of 100) based on event weights.

[0386] Rapid acceleration: Weight 25%, Deduction 10 points.

[0387] Emergency braking: Weighting 25%, Deduction 10 points.

[0388] Emergency lane change: Weighting 20%, Deduction 6 points.

[0389] Lane departure: Weighting 20%, Deduction 6 points.

[0390] Speeding: Weight 10%, not triggered, 0 points deducted.

[0391] Overall score = 100 - (10 + 10 + 6 + 6) = 68 points;

[0392] Step 4: Driving profile generation. Because the score is below the verification passing score (70 points), the system determines that user B's driving habits pose a certain risk, but an initial driving profile can still be generated and marked as "to be observed".

[0393] Driver's record contents:

[0394] Rapid acceleration habit: average 0.13 times / minute, peak 3.0 m / s 2 ,

[0395] Emergency braking habit: average 0.13 times / minute, peak deceleration 3.5 m / s² 2 ,

[0396] Lane-changing habit: Emergency lane change every 15 minutes, which is high.

[0397] Lane keeping: Lane departure incidents have occurred, requiring close monitoring.

[0398] Speeding habits: None

[0399] The file is stored in the vehicle's central control unit and will be dynamically corrected during the subsequent 30 days of driving data collection. For example, if the user gradually reduces the frequency of rapid acceleration and hard braking, the system will automatically adjust the file baseline.

[0400] In summary, as illustrated in the examples above, even new drivers can generate an individual driving profile after their first drive. Subsequent driving behaviors will continuously supplement and verify this profile, forming a personalized driving baseline. Furthermore, through a scoring mechanism and event habit recording, the system can detect potential dangerous driving tendencies at an early stage.

[0401] Reference Appendix Figure 2 An abnormal warning system based on driving behavior, as described in one embodiment of the present invention, includes:

[0402] The detection module 10 is used to detect the driving parameters of the vehicle during driving based on the driver identity information pre-recorded in the vehicle's central control system. The driving parameters specifically include the current acceleration, the rate of change of vehicle speed, and the brake pedal depth.

[0403] The judgment module 20 is used to determine whether the driving parameters match the driving habits corresponding to the driving identity information;

[0404] The execution module 30 is configured to, if not, obtain traffic control information of the vehicle within a preset range based on the driving road type pre-identified by the vehicle central control system, input the traffic control information into the prediction model pre-trained by the vehicle central control system in real time, and generate driving behavior data of the vehicle through the prediction model. The driving road type specifically includes fast-passage roads, urban roads, and complex environment roads, and the traffic control information specifically includes road traffic restriction, speed time control, and emergency control.

[0405] The second judgment module 40 is used to determine whether the driving behavior data conforms to the driving indicators of the driving identity information;

[0406] The second execution module 50 is used to, if the conditions are not met, compare the group driving behavior reference data with the driving behavior data based on the group driving behavior reference data pre-collected from the driving platform by the vehicle central control, calculate the difference value between the vehicle and the group baseline, identify the corresponding abnormal driving behavior in the driving behavior data, and generate an abnormal warning prompt for the driver through the vehicle based on the abnormal driving behavior. The group driving behavior reference data specifically includes vehicle speed distribution, acceleration range, and lane change frequency.

[0407] In this embodiment, the detection module 10 detects the vehicle's driving parameters during operation using built-in sensors based on the driver's identity information pre-recorded in the vehicle's central control system. These parameters include current acceleration, rate of change of vehicle speed, and brake pedal depth. The judgment module 20 then determines whether these driving parameters match the driving habits corresponding to the driver's identity information, and executes the corresponding steps accordingly. For example, if the system determines that the driving parameters match the driving habits corresponding to the driver's identity information, the system considers the driver's current operation consistent with their long-established driving characteristics, belonging to their normal driving mode. The system sets the current state to "conforming to driving habits," does not trigger an abnormal warning process, and uses the real-time driving parameters as the normal sample size. The data is stored for long-term optimization of the prediction model. Simultaneously, the matching driving parameters are bound and stored with the driver's identity information. If such behavior is recorded multiple times in different environments, the stability of the prediction model can be further enhanced. Furthermore, subsequent computationally expensive prediction model calculations and group baseline comparisons are no longer performed; only low-power real-time monitoring is maintained, saving system resources. For example, when the system determines that the vehicle's driving parameters do not match the driving habits corresponding to the driver's identity information, the execution module 30 will consider that the driver's current operation is inconsistent with their long-established driving characteristics, detecting some rare driving behaviors. The system will activate the camera devices in the vehicle's central control unit, using these cameras to identify the driving road type. Specifically, this includes roads categorized as expressways, urban roads, and roads in complex environments. The system acquires traffic control information for the vehicle within a pre-defined range on the driving platform. This traffic control information includes road traffic restrictions, speed limits during specific times, and emergency controls. This information is then input in real-time into a pre-trained prediction model in the vehicle's central control system. The prediction model generates driving behavior data. When driving parameters do not match the driving habits corresponding to the driver's identity information, the system can promptly identify rare or abnormal driver behaviors. This deviation detection method, based on individual characteristics, avoids the "one-size-fits-all" problem caused by applying a uniform threshold to all drivers in traditional methods. This allows for more accurate detection of potential dangerous behaviors and effectively improves driving behavior recognition. In addition to improving sensitivity and personalization, the system incorporates pre-identified road types and traffic control information within a preset range, enabling secondary verification of deviation behavior within the environmental context. This avoids misjudging behavioral changes caused by external environmental factors as abnormalities, making the warning results more consistent with actual driving scenarios and improving the rationality and practicality of the system's judgment. Furthermore, by inputting traffic control information into the prediction model and generating driving behavior data, the system can not only dynamically reconstruct the driver's behavioral characteristics in specific environments but also provide a reliable input basis for subsequent anomaly judgment and warnings. Through intelligent analysis of the model, the system can optimize the warning effect under the dual dimensions of individual differences and environmental constraints, thereby effectively reducing the false alarm rate and improving the overall driving safety assurance capability.The second judgment module 40 then determines whether these driving behavior data conform to the driving indicators of the driver's identity information, and executes the corresponding steps accordingly. For example, when the system determines that the vehicle's driving behavior data conforms to the driving indicators of the driver's identity information, the system will consider that the driver's current behavior is not only normal at the driving parameter level, but also still conforms to the individual driving indicators after correction by road type, traffic control information, and prediction model. This indicates that the driving state is safe and stable, and the system will not trigger an abnormal warning, but will continue to monitor in real time. At the same time, the driving behavior data that conforms to the driving indicators will be stored in the individual driving file as a basis for subsequent model optimization and personalized benchmarking. The system can gradually increase the weight of a behavior pattern if it is repeatedly verified as reasonable in different environments, making the model closer to the driver's real habits. This eliminates the need for complex subsequent steps such as group baseline comparison, reducing computational resource consumption and improving system efficiency. For example, when the system determines that the vehicle's driving behavior data does not meet the driving indicators of the driver's identity information, the second execution module 50 will consider the driver's current behavior to be abnormal at the driving parameter level and the driving state to be unstable. The system will then compare this group driving behavior reference data with the driving behavior data pre-collected from the driving platform by the vehicle's central control system. The system compares the vehicle's driving behavior data with the group baseline, calculates the difference between the vehicle and the baseline, and identifies corresponding abnormal driving behaviors. Based on different abnormal driving behaviors, it generates abnormal warning prompts for the driver. When the driving behavior data does not meet the driving indicators of the driver's identity information, the system can determine that the driving state is unstable. At this time, the system introduces the group driving behavior reference data as a baseline to further verify whether there is a real dangerous behavior in the case of individual abnormality, thereby avoiding misjudgment caused by relying solely on individual models and improving the accuracy of abnormal behavior identification. At the same time, by calculating the difference between the vehicle's driving behavior data and the group baseline, the system can quantify the degree of deviation of the driver's behavior from the normal range of the group and classify different types of abnormal driving behaviors. This method avoids classifying all deviations as abnormal indiscriminately and helps to distinguish between minor deviations and serious dangerous behaviors, making the abnormal judgment more hierarchical and reasonable. After completing the abnormal behavior classification, the system can generate graded warning prompts according to different abnormality types, reminding the driver through visual, audible, or vibration means. This graded warning avoids the interference caused by excessive reminders and can intervene in time when there is real danger, improving the driver's perception and response to risks, thereby effectively ensuring vehicle driving safety.

[0408] In this embodiment, it also includes:

[0409] The acquisition module is used to acquire the vehicle's location information through the driving platform based on the communication data pre-sent by the vehicle's central control system to the driving platform;

[0410] The third judgment module is used to determine whether the location information is within the geographical range pre-recorded by the vehicle's central control system;

[0411] The third execution module is used to, if so, construct driving behavior rules for the vehicle based on the preset traffic control requirements of the geographical area, mark the corresponding violations during the vehicle's operation according to the driving behavior rules, and dynamically present the violations to the vehicle through the vehicle's central control system. The dynamic presentation specifically includes dashboard image reminders, voice prompts, and steering wheel and seat vibrations.

[0412] In this embodiment, the system obtains the vehicle's location information through the driving platform based on communication data pre-sent by the vehicle's central control unit. The system then determines whether the location information is within the geographical area pre-recorded by the vehicle's central control unit and executes corresponding steps accordingly. For example, if the system determines that the vehicle's location information is not within the geographical area pre-recorded by the vehicle's central control unit, the system considers the vehicle to be in an unfamiliar or uncovered area. Its driving environment may be inconsistent with the driver's long-term driving habits and lacks corresponding group baseline or control rule data support. The system will increase the trigger threshold of the warning algorithm to avoid excessive warnings in unknown environments, only issuing alerts for "strong rule-based risks" (such as speeding, running red lights, and driving against traffic), while not immediately warning for minor deviations. Simultaneously, the system calls real-time traffic control information, road type characteristics, and group data for the new area through the driving platform or cloud. If the platform lacks readily available driving reference data, it temporarily uses nationwide / regional baseline data as a substitute and records the driving data in the new region, marking it as a "new region behavior sample." If the vehicle repeatedly operates in the region, a corresponding extended dataset of individual driving habits is gradually built and stored in the regional database of the vehicle's central control system. For example, when the system determines that the vehicle's location information is within the pre-collected regional range of the vehicle's central control system, the system will consider the vehicle to be in a known area. The driving environment is consistent with the driver's long-term driving habit data. The system will construct driving behavior rules for the vehicle based on the pre-set traffic control requirements of the regional range. Based on these driving behavior rules, the system will mark the corresponding violations during the vehicle's operation and dynamically present different violations to the vehicle through the vehicle's central control system. The dynamic presentation includes dashboard image reminders, voice prompts, and steering wheel and seat vibrations.When the vehicle's location is within the area recorded by the central control system, it indicates that the system has grasped the road characteristics, traffic rules, and driving environment of that area. Based on this premise, the system establishes driving behavior rules that fully reflect the driver's long-term habits in familiar areas, making the judgment of violations or abnormal behaviors more accurate. This avoids false alarms or unnecessary warnings due to environmental differences, improving the practicality and reliability of the warning system. Simultaneously, relying on traffic control requirements within the area, the system can automatically compare and verify driving behavior during vehicle operation. Once behaviors such as speeding, illegal lane changes, or illegal overtaking occur, the system can mark and output violation information in real time, displayed on the dashboard. Multimodal presentation methods, including image alerts, voice prompts, and seat / steering wheel vibrations, ensure that drivers receive timely warning signals regardless of their attention level, reducing risks caused by driver negligence. Furthermore, within a known geographical area, the system can gradually optimize driving behavior rules by combining individual drivers' historical behavior data with traffic control requirements. For example, for common driving habits such as minor speeding or sudden acceleration, the system can distinguish between "reasonable deviations" and "genuine violations" through long-term data recording and comparison. This dynamic adjustment not only enhances the personalization of warnings but also provides stable training samples and data support for subsequent cross-regional expansion of driving models.

[0413] In this embodiment, the execution module further includes:

[0414] The generation unit is used to generate the control content of the vehicle within the preset range based on the control type of the traffic control information through the vehicle central control system, wherein the control type specifically includes dynamic control and static control;

[0415] The judgment unit is used to determine whether the vehicle has driven out of the preset range;

[0416] The execution unit is used to dynamically map the control content if no, calculate the duration of the impact of each control rule on the vehicle within a preset driving time window based on the real-time driving path of the vehicle, and adaptively filter invalid control information of the control content on the vehicle based on the duration of the impact.

[0417] In this embodiment, the system generates control measures based on traffic control information, specifically dynamic and static control. Through the vehicle's central control system, it generates control measures for the vehicle within a pre-defined range. The system then determines whether the vehicle has left the pre-defined range and executes corresponding steps accordingly. For example, if the system determines that the vehicle has not left the pre-defined range, it considers the vehicle's behavior still subject to established traffic restrictions. The system-generated control measures (whether dynamic controls such as temporary road closures or time-based speed limits, or static controls such as fixed restricted areas) remain valid, and the system maintains current control requirements, such as speed limits, no-entry restrictions, or detour prompts. During vehicle movement, the system continuously compares the vehicle's driving parameters with the control measures. The system requires that potential violations be flagged. If the control is dynamic (e.g., road closures due to accidents, temporary speed limits), the system continuously updates rules from the traffic platform to avoid outdated alerts due to changes in control conditions. If the control is static (e.g., fixed restricted areas), the existing rules remain unchanged, only monitoring for vehicles entering at risk of violation. Furthermore, if a vehicle's behavior violates control measures (e.g., exceeding dynamic speed limits, approaching a restricted area), the system immediately issues a warning via dashboard image, voice prompts, or steering wheel / seat vibration. For example, if the system determines that a vehicle has left a pre-defined area, it considers the traffic restrictions on the vehicle's behavior needing to be updated. The system dynamically maps the control measures to the vehicle's behavior based on the vehicle's position. The system calculates the duration of each traffic control rule's impact on the vehicle within a pre-set driving time window based on the vehicle's real-time driving path. Based on different impact durations, it adaptively filters out invalid control information. When a vehicle leaves the preset area, the system automatically refreshes traffic restrictions and dynamically maps the control content. This mechanism prevents vehicles from continuing to be restricted by outdated control information that is no longer applicable (e.g., triggering a speeding warning even after leaving a speed-limited area), thereby reducing redundant warnings and error messages, ensuring the accuracy and real-time nature of warning information, and increasing driver trust in the system. Furthermore, the system not only simply removes controls from old areas but also calculates the impact of each control rule within the driving time window based on the vehicle's real-time driving path. The system adaptively filters rules based on their duration of impact, ensuring that short-term invalid control information (such as speed limits for completed temporary construction projects) is no longer prompted, while rules still within their effective time periods (such as peak-hour restrictions and long-term no-entry zones) are retained. This achieves refined and differentiated control, improving the system's usability in complex traffic scenarios. Furthermore, by dynamically filtering and refreshing control information, drivers receive prompts highly relevant to their current driving route, rather than being distracted by a large number of irrelevant or invalid rules. This optimization not only helps drivers focus on genuine traffic restrictions and reduce the risk of violations but also reduces unnecessary reminders that distract the driver's attention. Thus, while ensuring safety, it also improves the overall driving experience and the user-friendliness of human-computer interaction.

[0418] In this embodiment, the second execution module further includes:

[0419] The identification unit is used to identify the source channels of the group driving behavior reference data based on the pre-established communication connection between the vehicle central control and the driving platform. Specifically, the source channels include data from official traffic management platforms, data collected by car manufacturers, and data from internet travel platforms.

[0420] The second judgment unit is used to determine whether the data format of the source channel is uniform;

[0421] The second execution unit is configured to, if not, obtain the temporal continuity of the group driving behavior reference data, detect the missing data values ​​of the group driving behavior reference data based on the temporal continuity, generate the corresponding missing ratio, and correct the dynamic timeliness of the group driving behavior reference data in real time based on the missing ratio.

[0422] In this embodiment, the system identifies the source channels of group driving behavior reference data based on the pre-established communication connection between the vehicle's central control system and the driving platform. These source channels specifically include data from official traffic management platforms, data collected by vehicle manufacturers, and data from internet-based mobility platforms. The system then determines whether the data formats of these source channels are consistent, and executes corresponding steps accordingly. For example, when the system determines that the data formats of the source channels for group driving behavior reference data are consistent, it considers that the data can be directly interfaced and processed without additional data cleaning or conversion, resulting in high data compatibility and integrity. The system directly merges and aggregates data from different channels to build a baseline database covering a wider range of drivers. Simultaneously, it directly calls data in a consistent format into the prediction model for difference calculation and group baseline verification, omitting the format conversion step, reducing processing latency, and improving real-time performance. This allows for rapid comparison of individual driving behavior with group reference data, improving the accuracy and response efficiency of warnings. Conversely, when the system determines that the data formats of the source channels for group driving behavior reference data are inconsistent, it considers that the data from these source channels may have data delays or missing data. The system then obtains the temporal continuity of this group driving behavior reference data and, based on different temporal continuity, detects... The system identifies missing values ​​in the group driving behavior reference data and generates corresponding missing proportions. Based on these missing proportions, the dynamic timeliness of the group driving behavior reference data is corrected in real time. When the system detects that the data formats from different sources are inconsistent, it analyzes the temporal continuity of the data and identifies missing values ​​to calculate the missing proportion in real time. This method avoids distortion of the group baseline caused by data delays or missing values, ensuring that the data input to the prediction model always maintains high integrity and reliability, thereby avoiding false warnings or misjudgments caused by data gaps. Furthermore, the system not only detects missing values ​​but also dynamically corrects the timeliness of the group driving behavior reference data based on the missing proportion. This dynamic correction mechanism ensures that the group baseline can be adaptively updated even when the data source fluctuates, improving the system's compatibility with multi-source heterogeneous data and making it more suitable for high-frequency calculations and warning needs in real-time driving scenarios. Through correction based on the missing proportion, the group reference data can more realistically reflect the distribution characteristics of driving behavior, avoiding comparisons between individual driving behavior and "distorted" group data, thus reducing the false alarm rate. This mechanism can significantly improve the accuracy of abnormal warnings, ensuring that the warning information is both consistent with real-time driving scenarios and has practical reference value, enhancing the system's practicality in complex road environments.

[0423] In this embodiment, the determination module further includes:

[0424] The construction unit is used to construct external environment information corresponding to the external driving environment based on the external driving environment pre-detected by the vehicle's central control system. Specifically, the external environment information includes meteorological and natural environment, road traffic conditions, and dynamic traffic participants.

[0425] The third judgment unit is used to determine whether the external environment information belongs to a preset normal driving scenario;

[0426] The third execution unit is used to, if not, continuously collect the vehicle's deviation behavior within a preset driving time window according to the environmental type of the external driving environment, identify the trigger frequency of the deviation behavior, and dynamically mark the deviation behavior as an observation state based on the trigger frequency.

[0427] In this embodiment, the system constructs external environment information corresponding to the external driving environment on a pre-installed display screen in the vehicle, based on the external driving environment pre-detected by the vehicle's central control system. This external environment information specifically includes weather conditions, road traffic conditions, and dynamic traffic participants. The system then determines whether this external environment information belongs to a pre-set normal driving scenario and executes the corresponding steps accordingly. For example, when the system determines that the external environment information corresponding to the external driving environment belongs to a pre-set normal driving scenario, the system considers the current external environment to be within the normal range preset by the vehicle's central control system, without severe weather, special road restrictions, or abnormal traffic behavior. The system will then maintain a normal monitoring mode, monitoring acceleration and vehicle speed. The system routinely compares driving parameters such as rate of change and brake pedal depth without triggering additional warnings or interventions. Simultaneously, it aligns external environmental information with driving habit data corresponding to the driver's identity information, dynamically verifying whether the driver's behavior conforms to an individualized baseline. If there is no significant deviation, it continues to maintain stable monitoring and continuously presents routine environmental information on the vehicle's display screen in a concise and intuitive manner, including normal weather icons, traffic flow indicators, and regular traffic distribution, to help the driver maintain environmental awareness. For example, if the system determines that the external environmental information corresponding to the external driving environment does not belong to the pre-set routine driving scenario, the system will assume that the vehicle's central control system cannot understand the current external environment in real time, potentially leading to adverse conditions. In response to weather conditions, special road restrictions, or abnormal traffic behavior, the system continuously collects vehicle deviation behavior data within a pre-set driving time window, based on the type of external driving environment. It identifies the trigger frequency of these deviations and dynamically marks them as pending observation based on different trigger frequencies. When the external environment deviates from typical driving scenarios, the system proactively identifies potential uncertainties and compensates for the central control system's insufficient perception of the external environment by continuously collecting deviation behavior data. This allows the vehicle to maintain dynamic monitoring of driver behavior even in complex or abnormal environments, enhancing the system's adaptability and robustness. Furthermore, the system does not immediately trigger a warning upon detecting deviation behavior; instead, it first statistically analyzes the deviation data. By marking the trigger frequency as "to be observed," this delayed judgment mechanism effectively avoids false alarms caused by occasional operations (such as sudden braking or lane changing). It ensures that the risk judgment process is only initiated when the deviation behavior shows a high frequency or continuous trend, thereby improving the accuracy and practicality of abnormal warnings. Furthermore, through dynamic marking based on trigger frequency, the system can form a hierarchical risk observation mechanism. It can not only monitor the development trend of deviation behavior in real time, but also dynamically adjust the trigger threshold according to individual driver differences. This method makes the warning system more flexible in the face of uncertain environments, enabling it to detect real risks in a timely manner while avoiding excessive intervention, ultimately improving overall driving safety and user experience.

[0428] In this embodiment, the second determination module further includes:

[0429] The data acquisition unit is used to acquire the real-time deviation value of the driving behavior data based on the preset upper limit of the deviation value of the driving index. The real-time deviation value specifically includes speed deviation rate, acceleration deviation and vehicle distance deviation.

[0430] The fourth judgment unit is used to determine whether the real-time deviation value exceeds the upper limit of the deviation value;

[0431] The fourth execution unit is used to identify the additional time period of the real-time deviation value if the condition is met, perform weighted summation of the deviation values ​​of each indicator of the driving behavior data to generate a comprehensive deviation score, and construct a tolerance range for the driving behavior data based on the comprehensive deviation score. The additional time period specifically includes daytime, nighttime, peak hours and off-peak hours, and the tolerance range specifically includes compliance with the indicator, slight deviation and severe deviation.

[0432] In this embodiment, the system collects real-time deviation values ​​of driving behavior data based on a pre-set upper limit for deviation values ​​in the driving indicators. These real-time deviation values ​​specifically include speed deviation rate, acceleration deviation, and distance deviation. The system then determines whether these real-time deviation values ​​exceed the upper limit and executes corresponding steps accordingly. For example, if the system determines that the real-time deviation values ​​of the driving behavior data do not exceed the upper limit, the system considers the vehicle's current driving behavior to be within the safe range allowed by the driving indicators, and the driver's speed, acceleration, and distance from the vehicle in front all conform to the predetermined individualized baseline or group reference standard. The system will continue to collect and monitor driving behavior data in real time, maintaining the dynamic nature of the deviation values. The system tracks the vehicle's deviation without triggering any abnormal warnings. It compares the real-time deviation value with driving indicators and continuously updates the data on the display screen or in the background log for subsequent trend analysis and predictive model optimization. For deviations within the upper limit, the system presents the data as regular information, such as speed indicators or distance status, through the dashboard or app without interfering with driver operation. However, if the system determines that the real-time deviation of the driving behavior data exceeds the upper limit, it considers the current driving behavior unsafe and identifies additional time periods for the real-time deviation value. These additional time periods include daytime, nighttime, peak hours, and off-peak hours. The deviation values ​​of the indicators are weighted and summed to generate a comprehensive deviation score. Based on different comprehensive deviation scores, a tolerance range for driving behavior data is constructed. The tolerance range specifically includes compliance with the indicators, minor deviation, and severe deviation. When the real-time deviation value exceeds the preset upper limit, the system not only detects a single deviation event but also combines additional time periods to weight and sum the deviation values ​​of each indicator to generate a comprehensive deviation score. This method can comprehensively consider the impact of different environmental conditions on driving behavior, improve the accuracy of judging abnormal driving behavior, and avoid misjudging driving risks due to a single indicator or instantaneous event. Furthermore, based on the comprehensive deviation score, the system constructs a tolerance range including compliance with the indicators, minor deviation, and severe deviation. Through this hierarchical management, driving behavior is no longer a simple binary judgment of "normal / abnormal," but rather a risk level that can be differentiated based on the degree of deviation. This allows the warning system to respond flexibly, taking suggestive measures for minor deviations and immediately triggering strong interventions or warnings for serious deviations. Furthermore, additional time periods are included in the scoring calculation, enabling the system to adjust the deviation judgment criteria according to the driving characteristics of different time periods. For example, when visibility is limited at night, a slightly higher speed deviation is allowed, while in peak-hour congestion, the requirements for vehicle spacing deviation are more stringent. This method achieves personalized and environmentally adaptable driving behavior analysis, improving the system's practicality and safety assurance capabilities under complex and variable road conditions.

[0433] In this embodiment, it also includes:

[0434] The reading module is used to read the user's identity data from the vehicle's central control system based on the identity identifier that the user has entered into the vehicle.

[0435] The fourth judgment module is used to determine whether the identity data is included in the vehicle central control system;

[0436] The fourth execution module is used to activate the preset driving text of the vehicle central control if not, to perform driving verification on the user, and to construct a driving profile of the user driving the vehicle based on the verification result of the driving verification. The driving profile retains the user's driving event habits, wherein the driving event habits specifically include rapid acceleration, emergency braking, emergency lane change, continuous speeding and lane departure.

[0437] In this embodiment, the system reads the user's identity data from the vehicle's central control unit (NCU) based on the user's registered identity identifier. The system then determines whether this identity data is recorded in the NCU and executes corresponding steps accordingly. For example, if the system determines that the user's identity data is recorded in the NCU, it assumes that the NCU has identified and stored the user's identity information, and the user is a known driver or registered user. The system directly calls the user's individualized driving model, including driving habit baselines, preset driving indicators, and deviation thresholds, for real-time driving behavior monitoring. Simultaneously, it automatically loads user preferences based on the identity information, such as seat position, rearview mirror angle, air conditioning settings, and personalized warning strategies. Furthermore, during driving behavior analysis and abnormal warnings, the system matches the user's identity data with real-time driving behavior data to achieve personalized and accurate risk assessment and alerts. Conversely, if the system determines that the user's identity data is not recorded in the NCU, it assumes that the NCU has not stored the user's identity information, and the user is a new driver. The system activates the pre-set driving text in the NCU and performs a simple driving verification on the user. Based on the verification result... The system constructs a driving profile for each user's vehicle, retaining their driving habits, including rapid acceleration, sudden braking, emergency lane changes, continuous speeding, and lane departure. When the system detects that a user's identity data is not included in the vehicle's central control system, it automatically identifies them as a new driver and verifies them using a pre-set driving text. This mechanism can quickly establish a driving profile for a user's first drive, avoiding system malfunctions due to a lack of historical data, thus improving the system's versatility and intelligent adaptability. During driving verification and subsequent driving, the system records the new user's key driving habits (such as rapid acceleration, sudden braking, emergency lane changes, continuous speeding, and lane departure) in the driving profile. As data accumulates, the system gradually grasps the user's personalized driving characteristics, avoiding a "one-size-fits-all" approach to risk assessment and improving the accuracy of anomaly detection. By constructing driving profiles for new users and continuously retaining their driving habits, the system can conduct long-term tracking and dynamic updates during subsequent driving, achieving comprehensive management of driving behavior from scratch. The establishment of driving profiles also provides a traceable basis for subsequent driving assessments, anomaly warnings, and personalized safety interventions.

[0438] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A driving behavior based abnormality early warning method, characterized by, Includes the following steps: Based on the driver identity information pre-collected by the vehicle's central control system, the vehicle's driving parameters during the driving process are detected. Specifically, the driving parameters include the current acceleration, the rate of change of vehicle speed, and the brake pedal depth. Determine whether the driving parameters match the driving habits corresponding to the driving identity information; If not, then based on the driving road type pre-identified by the vehicle central control system, the traffic control information of the vehicle within a preset range is obtained, and the traffic control information is input into the prediction model pre-trained by the vehicle central control system in real time. Through the prediction model, the driving behavior data of the vehicle is generated. The driving road type specifically includes fast passage roads, urban passage roads, and complex environment roads. The traffic control information specifically includes road traffic restriction, speed time control, and emergency control. Determine whether the driving behavior data conforms to the driving indicators of the driving identity information; If the conditions are not met, the group driving behavior reference data pre-collected from the driving platform by the vehicle's central control system is compared with the driving behavior data. The difference value between the vehicle and the group baseline is calculated, and the corresponding abnormal driving behavior is identified in the driving behavior data. Based on the abnormal driving behavior, an abnormal warning prompt is generated for the driver through the vehicle. The group driving behavior reference data specifically includes vehicle speed distribution, acceleration range, and lane change frequency.

2. The driving behavior based abnormality early warning method according to claim 1, characterized by, Before the step of generating the driving behavior data of the vehicle, the method further includes: Based on the communication data pre-sent by the vehicle's central control system to the driving platform, the vehicle's location information is obtained through the driving platform; Determine whether the location information is within the geographical range pre-recorded by the vehicle's central control system; If so, then based on the traffic control requirements preset in the geographical area, driving behavior rules for the vehicle are constructed. According to the driving behavior rules, corresponding violations are marked during the vehicle's operation. The violations are dynamically presented to the vehicle through the vehicle's central control system. Specifically, the dynamic presentation includes dashboard image reminders, voice prompts, and steering wheel and seat vibrations.

3. The driving behavior based abnormality early warning method according to claim 1, characterized by, The step of obtaining traffic control information for the vehicle within a preset range based on the driving road type pre-identified by the vehicle's central control system further includes: Based on the control type of the traffic control information, the vehicle central control system generates the control content of the vehicle within the preset range, wherein the control type specifically includes dynamic control and static control; Determine whether the vehicle has left the preset range; If not, the control content is dynamically mapped, and the duration of each control rule's impact on the vehicle within a preset driving time window is calculated based on the vehicle's real-time driving path. Based on the duration of the impact, invalid control information of the control content on the vehicle is adaptively filtered.

4. The driving behavior based abnormality early warning method according to claim 1, characterized by, The step of using the group driving behavior reference data pre-collected from the driving platform by the vehicle's central control system further includes: Based on the pre-established communication connection between the vehicle central control and the driving platform, the source channels of the group driving behavior reference data are identified, wherein the source channels specifically include data from official traffic management platforms, data collected by car manufacturers themselves, and data from Internet travel platforms; Determine whether the data format of the source channels is consistent; If not, the temporal continuity of the group driving behavior reference data is obtained, and based on the temporal continuity, missing data values ​​of the group driving behavior reference data are detected, a corresponding missing ratio is generated, and the dynamic timeliness of the group driving behavior reference data is corrected in real time based on the missing ratio.

5. The driving behavior based abnormality early warning method according to claim 1, characterized by, The step of determining whether the driving parameters match the driving habits corresponding to the driving identity information further includes: Based on the external driving environment pre-detected by the vehicle's central control system, external environment information corresponding to the external driving environment is constructed, wherein the external environment information specifically includes meteorological and natural environment, road traffic conditions, and dynamic traffic participants; Determine whether the external environment information belongs to a preset normal driving scenario; If not, then based on the environmental type of the external driving environment, the vehicle's deviation behavior is continuously collected within a preset driving time window, the trigger frequency of the deviation behavior is identified, and the deviation behavior is dynamically marked as an observation state based on the trigger frequency.

6. The driving behavior based abnormality early warning method according to claim 1, characterized by, The step of determining whether the driving behavior data conforms to the driving indicators of the driving identity information further includes: Based on the preset upper limit of the deviation value of the driving index, the real-time deviation value of the driving behavior data is collected, wherein the real-time deviation value specifically includes speed deviation rate, acceleration deviation and vehicle distance deviation; Determine whether the real-time deviation value exceeds the upper limit of the deviation value; If so, the additional time period of the real-time deviation value is identified, the deviation values ​​of each indicator of the driving behavior data are weighted and summed to generate a comprehensive deviation score, and a tolerance range of the driving behavior data is constructed based on the comprehensive deviation score. The additional time period specifically includes daytime, nighttime, peak hours and off-peak hours, and the tolerance range specifically includes compliance with the indicator, slight deviation and severe deviation.

7. The driving behavior based abnormality early warning method according to claim 1, characterized by, Before the step of detecting the vehicle's driving parameters based on the driver's identity information pre-recorded in the vehicle's central control system, the method further includes: Based on the user's identity identifier entered in the vehicle, the user's identity data is read from the vehicle's central control system; Determine whether the identity data is included in the vehicle's central control system; If not, the preset driving text of the vehicle's central control is activated to verify the user's driving. Based on the verification result, a driving profile of the user driving the vehicle is constructed. The user's driving event habits are retained in the driving profile. The driving event habits specifically include rapid acceleration, sudden braking, emergency lane change, continuous speeding, and lane departure.

8. A driving behavior based abnormality early warning system, characterized by, include: The detection module is used to detect the driving parameters of the vehicle during driving based on the driver identity information pre-recorded in the vehicle's central control system. The driving parameters specifically include the current acceleration, the rate of change of vehicle speed, and the brake pedal depth. The judgment module is used to determine whether the driving parameters match the driving habits corresponding to the driving identity information; The execution module is used to, if not, obtain traffic control information of the vehicle within a preset range based on the driving road type pre-identified by the vehicle central control system, input the traffic control information into the prediction model pre-trained by the vehicle central control system in real time, and generate driving behavior data of the vehicle through the prediction model. The driving road type specifically includes fast passage roads, urban passage roads, and complex environment roads, and the traffic control information specifically includes road traffic restriction, speed time control, and emergency control. The second judgment module is used to determine whether the driving behavior data conforms to the driving indicators of the driving identity information; The second execution module is used to, if the conditions are not met, compare the group driving behavior reference data with the driving behavior data based on the group driving behavior reference data pre-collected from the driving platform by the vehicle central control, calculate the difference value between the vehicle and the group baseline, identify the corresponding abnormal driving behavior in the driving behavior data, and generate an abnormal warning prompt for the driver through the vehicle based on the abnormal driving behavior. The group driving behavior reference data specifically includes vehicle speed distribution, acceleration range, and lane change frequency.

9. The driving behavior based anomaly early warning system of claim 8, wherein, Also includes: The acquisition module is used to acquire the vehicle's location information through the driving platform based on the communication data pre-sent by the vehicle's central control system to the driving platform; The third judgment module is used to determine whether the location information is within the geographical range pre-recorded by the vehicle's central control system; The third execution module is used to, if so, construct driving behavior rules for the vehicle based on the preset traffic control requirements of the geographical area, mark the corresponding violations during the vehicle's operation according to the driving behavior rules, and dynamically present the violations to the vehicle through the vehicle's central control system. The dynamic presentation specifically includes dashboard image reminders, voice prompts, and steering wheel and seat vibrations.

10. The driving behavior based anomaly early warning system of claim 8, wherein, The execution module further includes: The generation unit is used to generate the control content of the vehicle within the preset range based on the control type of the traffic control information through the vehicle central control system, wherein the control type specifically includes dynamic control and static control; The judgment unit is used to determine whether the vehicle has driven out of the preset range; The execution unit is used to dynamically map the control content if no, calculate the duration of the impact of each control rule on the vehicle within a preset driving time window based on the real-time driving path of the vehicle, and adaptively filter invalid control information of the control content on the vehicle based on the duration of the impact.

Images