Search

CN-121973794-A - Driving behavior dynamic monitoring system based on intelligent model

CN121973794ACN 121973794 ACN121973794 ACN 121973794ACN-121973794-A

Abstract

The invention discloses a driving behavior dynamic monitoring system based on an intelligent model, which relates to the technical field of intelligent driving assistance, and comprises a data acquisition module, a characteristic processing module and a driving behavior analysis module; the driving behavior analysis module is connected with the feature processing module and is used for predicting the inertia feature corresponding to the operation to be executed by the driver based on the scene feature, comparing the inertia feature with the inertia feature of the actual execution operation and outputting the judgment result of the driving state by combining the physiological feature analysis result and the inertia feature comparison result.

Inventors

  • ZHAO ZHONGZE

Assignees

  • 上海凌芯体育科技有限公司

Dates

Publication Date
20260505
Application Date
20260316

Claims (10)

  1. 1. A driving behavior dynamic monitoring system based on an intelligent model, comprising: the data acquisition module comprises a physiological data acquisition unit, an inertial data acquisition unit and a scene data acquisition unit; The feature processing module is connected with the data acquisition module and comprises a physiological feature processing unit, an inertial feature extraction unit and a scene feature processing unit; The driving behavior analysis module is connected with the feature processing module and comprises a physiological feature analysis unit, a scene prediction unit, an inertial feature comparison unit, a driving state judgment unit and a reminding unit; The driving behavior analysis module is used for predicting the inertia characteristics corresponding to the operation to be executed by the driver based on the scene characteristics, comparing the inertia characteristics with the inertia characteristics of the actual execution operation, and outputting the judgment result of the driving state by combining the physiological characteristic analysis result and the inertia characteristic comparison result.
  2. 2. The intelligent model-based driving behavior dynamic monitoring system according to claim 1, wherein the physiological data acquisition unit, the inertial data acquisition unit and the scene data acquisition unit comprise: The system comprises a physiological data acquisition unit, an inertia data acquisition unit, a steering wheel grip and brake pedal acceleration sensor, a scene data acquisition unit and a scene data acquisition unit, wherein the physiological data acquisition unit acquires an HRV signal in the driving process of a driver through a wearable HRV physiological sensor, the inertia data acquisition unit acquires the steering wheel grip and the brake pedal acceleration in the driving process of the driver through a steering wheel pressure sensor and a brake pedal acceleration sensor, and the scene data acquisition unit acquires scene data in a front preset distance, including traffic signal lamp states, front target positions and motion parameters, through a front-view camera and a millimeter wave radar.
  3. 3. The driving behavior dynamic monitoring system based on the intelligent model according to claim 1, wherein the physiological characteristic processing unit comprises: Preprocessing the HRV signals acquired by the physiological data acquisition unit, extracting time domain and frequency domain characteristic parameters of the HRV signals, generating physiological characteristic vectors, and transmitting the physiological characteristic vectors to the physiological characteristic analysis unit.
  4. 4. The driving behavior dynamic monitoring system based on the intelligent model according to claim 1, wherein the inertial feature extraction unit comprises: The steering wheel grip strength and the brake pedal acceleration acquired by the inertia data acquisition unit are preprocessed, the steering wheel grip strength change rate, the brake pedal acceleration peak value and the brake pedal acceleration response delay are extracted, an inertia characteristic vector is generated, and the inertia characteristic vector is transmitted to the inertia characteristic comparison unit.
  5. 5. The driving behavior dynamic monitoring system based on the intelligent model according to claim 1, wherein the scene feature processing unit comprises: The scene data in the front preset distance collected by the scene data collection unit is preprocessed, the state of the traffic signal lamp, the front target position and the motion characteristics are extracted, a scene characteristic vector is generated, and the scene characteristic vector is transmitted to the scene prediction unit.
  6. 6. The driving behavior dynamic monitoring system based on the intelligent model according to claim 1, wherein the physiological characteristic analysis unit comprises: Based on the physiological data acquisition unit and the physiological characteristic processing unit, acquiring a physiological characteristic vector of the driver in a normal driving state, and taking the physiological characteristic vector as an HRV normal baseline of the driver; The method comprises the steps of receiving a driver physiological characteristic vector transmitted by a physiological characteristic processing unit in real time, calculating a characteristic deviation value of the driver physiological characteristic vector and an HRV normal baseline, comparing the characteristic deviation value with a preset abnormal threshold value to obtain a physiological characteristic analysis result, specifically, judging that the physiological state is abnormal when the characteristic deviation value exceeds the preset abnormal threshold value, judging that the physiological state is normal when the characteristic deviation value does not exceed the preset abnormal threshold value, and transmitting the physiological characteristic analysis result obtained through judgment to a driving state judging unit.
  7. 7. The driving behavior dynamic monitoring system based on an intelligent model according to claim 1, wherein the scene prediction unit comprises: Based on an inertial data acquisition unit, a scene data acquisition unit, an inertial feature extraction unit and a scene feature processing unit, taking each scene feature vector and a corresponding inertial feature vector under a normal driving state of a driver as training samples, and selecting three scenes of green light straight driving into an intersection, front target speed reduction and vehicle following and red light braking and parking as training scenes; quantizing scene feature vectors, wherein the quantized scene feature vectors comprise traffic signal lamp state codes S, front target relative distances d, front target relative speeds v and preset front monitoring distances L, and form quantized scene feature parameters [ S, d, v, L ] with L and d being larger than 0; when no front target exists in a normal driving state sample and no signal lamp exists, a least square method is adopted to solve a basic grip strength change rate coefficient alpha 0 , a basic steering wheel grip strength change rate k 1 is obtained, a reference value when no scene influence is indicated, and the formula is as follows: ; Based on the basic grip strength change rate k 1 , solving a traffic signal lamp state influence coefficient alpha 1 by combining with a traffic signal lamp state S, and performing first-order correction on k 1 to obtain a corrected grip strength change rate k 2 , wherein the formula is as follows: ; Based on the corrected grip strength change rate k 2 , combining the traffic signal lamp state code S, the front target relative distance d, the front target relative speed v and the preset front monitoring distance L, solving a target relative speed and distance influence coefficient alpha 2 and a signal lamp and target distance influence coefficient alpha 3 , carrying out second-order correction on k 2 to obtain the steering wheel grip strength change rate k, wherein the formula is as follows: ; The formula of the brake pedal acceleration peak value a max is: ; Wherein, beta 0 is basic brake peak value coefficient, beta 1 is red light influence coefficient, beta 2 is front target distance influence coefficient, beta 3 is red light scene and target speed influence coefficient; the brake pedal acceleration response delay t formula is: ; Wherein, gamma 0 is a basic response delay coefficient, gamma 1 is a green light influence coefficient, gamma 2 is a front target distance duty ratio influence coefficient, and gamma 3 is a green light scene and target distance duty ratio influence coefficient; The method comprises the steps of adopting a least square method to solve each group of coefficients to obtain initial coefficients matched with driving habits of a driver, constructing XGBoost regression models based on the initial coefficients, and obtaining optimized prediction models by taking mean square error as a loss function; and receiving the scene feature vectors transmitted by the scene feature processing unit in real time, quantizing the scene feature vectors, inputting a prediction model, outputting a reasonable range of inertia feature values under the corresponding scene by the prediction model, generating a predicted inertia feature range vector, and transmitting the predicted inertia feature range vector to the inertia feature comparison unit.
  8. 8. The driving behavior dynamic monitoring system based on the intelligent model according to claim 1, wherein the inertial feature comparison unit comprises: receiving the predicted inertia characteristic range vector transmitted by the scene prediction unit, extracting a reasonable range of the inertia characteristic value corresponding to the real-time quantized scene characteristic parameter, and calculating the deviation value of the inertia characteristic value in the real-time driving state of the driver and the corresponding predicted inertia characteristic range, wherein the steering wheel grip change rate and the brake pedal acceleration peak value are calculated by adopting a relative deviation formula, and the brake pedal acceleration response delay is calculated by adopting an absolute deviation formula; Comparing the calculated deviation value of the change rate of the grip of the steering wheel, the deviation value of the peak acceleration of the brake pedal and the deviation value of the response delay of the acceleration of the brake pedal with a preset grip threshold value, a preset brake peak value threshold value and a preset brake response delay threshold value of a corresponding scene respectively to obtain an inertial characteristic comparison result, specifically, judging that the inertial operation is abnormal when any deviation value exceeds the corresponding threshold value, judging that the inertial operation is normal when the deviation value does not exceed the corresponding threshold value, and transmitting the inertial characteristic comparison result to a driving state judgment unit.
  9. 9. The driving behavior dynamic monitoring system based on the intelligent model according to claim 1, wherein the driving state determination unit includes: The driving state judgment result is generated by receiving the physiological characteristic analysis result transmitted by the physiological characteristic analysis unit and the inertial characteristic comparison result transmitted by the inertial characteristic comparison unit, specifically, the driving state judgment result is judged to be abnormal when the physiological characteristic analysis result is abnormal in physiological state and the inertial characteristic comparison result is abnormal in inertial operation, the driving state judgment result is judged to be abnormal when the physiological state is abnormal but the inertial operation is normal, the inertial operation judgment result is judged to be abnormal when the physiological state is normal but the inertial operation is abnormal, and the driving state judgment result is judged to be normal when the physiological state is normal and the inertial operation is normal, and the driving state judgment result is transmitted to the reminding unit.
  10. 10. The driving behavior dynamic monitoring system based on an intelligent model according to claim 1, wherein the reminding unit comprises: and receiving a driving state judging result transmitted by the driving state judging unit, and triggering the driving abnormality reminding of the vehicle-mounted terminal when the driving state judging result is abnormal driving state, abnormal physiological state or abnormal inertial operation.

Description

Driving behavior dynamic monitoring system based on intelligent model Technical Field The invention relates to the technical field of intelligent driving assistance, in particular to a driving behavior dynamic monitoring system based on an intelligent model. Background The driving state monitoring is a core technical link of the intelligent driving auxiliary system, the monitoring precision of the intelligent driving auxiliary system directly determines the effectiveness and the practicability of the vehicle-mounted safety early warning, and the intelligent driving auxiliary system has important significance for guaranteeing driving safety. In the prior art, the driving state evaluation scheme has obvious defects, and is difficult to consider monitoring accuracy and anti-interference capability, and the specific defects are as follows: In the prior art, the physiological characteristics of a driver are singly relied on for monitoring, and the physiological state is directly analyzed and mapped into the driving state mainly through collecting physiological data. However, physiological characteristics are easily affected by non-driving related factors such as mood fluctuation, environmental interference, temporary physical state and the like, signal stability is poor, real association of driving behaviors and physiological states cannot be accurately reflected, misjudgment often occurs, and therefore monitoring results have larger deviation and limited practicality. Meanwhile, the prior art lacks association analysis of scene perception and inertial operation, does not combine scene data, does not predict reasonable range of inertial operation which should be executed by a driver under different scenes, only compares actual inertial operation with a fixed threshold value, cannot distinguish normal operation and dangerous abnormal operation of scene adaptation, is separated from actual driving scenes, and further reduces monitoring accuracy. In addition, the prior art does not construct a fusion judgment logic of physiological characteristics and inertial operation characteristics, and each characteristic independently outputs a result, so that interference cannot be eliminated and deviation cannot be corrected through multidimensional complementary verification, driving state evaluation is on the one side, and high-precision and high-reliability driving state monitoring requirements are difficult to meet. Disclosure of Invention The invention aims to provide a driving behavior dynamic monitoring system based on an intelligent model, which aims to solve the problems in the prior art. In order to achieve the above purpose, the present invention provides the following technical solutions: In a first aspect, the present invention provides a driving behavior dynamic monitoring system based on an intelligent model, including: the data acquisition module comprises a physiological data acquisition unit, an inertial data acquisition unit and a scene data acquisition unit; The feature processing module is connected with the data acquisition module and comprises a physiological feature processing unit, an inertial feature extraction unit and a scene feature processing unit; The driving behavior analysis module is connected with the feature processing module and comprises a physiological feature analysis unit, a scene prediction unit, an inertial feature comparison unit, a driving state judgment unit and a reminding unit; The driving behavior analysis module is used for predicting the inertia characteristics corresponding to the operation to be executed by the driver based on the scene characteristics, comparing the inertia characteristics with the inertia characteristics of the actual execution operation, and outputting the judgment result of the driving state by combining the physiological characteristic analysis result and the inertia characteristic comparison result. With reference to the first aspect, in a first implementation manner of the first aspect of the present application, the physiological data acquisition unit, the inertial data acquisition unit, and the scene data acquisition unit include: The system comprises a physiological data acquisition unit, an inertia data acquisition unit, a steering wheel grip and brake pedal acceleration sensor, a scene data acquisition unit and a scene data acquisition unit, wherein the physiological data acquisition unit acquires an HRV signal in the driving process of a driver through a wearable HRV physiological sensor, the inertia data acquisition unit acquires the steering wheel grip and the brake pedal acceleration in the driving process of the driver through a steering wheel pressure sensor and a brake pedal acceleration sensor, and the scene data acquisition unit acquires scene data in a front preset distance, including traffic signal lamp states, front target positions and motion parameters, through a front-view camera and a millimeter wave radar. With reference to th