CN-122024471-A - Vehicle queue risk fault tolerance assessment method based on zero trust architecture

CN122024471ACN 122024471 ACN122024471 ACN 122024471ACN-122024471-A

Abstract

The invention relates to the technical field of intelligent traffic systems, in particular to a vehicle queue risk fault tolerance assessment method based on a zero trust architecture, which comprises the steps of establishing zero trust triggering conditions for reflecting the safety situation between any vehicle pair in the running process of a vehicle group, acquiring vehicle multi-source running data after triggering, and outputting cleaned window data after data preprocessing; the risk assessment mechanism based on the data reliability carries out data consistency check on window data of any vehicle node in the vehicle group to obtain comprehensive consistency scores of the vehicle nodes, scoring sub-items of the vehicle nodes are constructed, the comprehensive consistency scores are used as evidence quality to carry out consistency adjustment on the scoring sub-items, and dynamic trust values of all the vehicle nodes are output. According to the intelligent network vehicle-connected node dynamic trust evaluation method, dynamic trust evaluation can be carried out on the intelligent network vehicle-connected node under the condition that information is incomplete and data is inconsistent, driving risk evaluation with fault tolerance is achieved, and safety and robustness of a vehicle group system in a complex traffic environment are improved.

Inventors

HUANG DARONG
LI BOXI
ZHANG LEI
XIA QIN
LIU HAITAO

Assignees

安徽大学
中国汽车工程研究院股份有限公司
润建股份有限公司

Dates

Publication Date: 20260512
Application Date: 20260121

Claims (10)

1. A vehicle queue risk fault tolerance assessment method based on a zero trust architecture is characterized by comprising the following steps: S1, establishing a system for reflecting any vehicle pair in the running process of a vehicle group Zero trust triggering condition of safety situation and collecting vehicle multisource operation data after triggering After data preprocessing, outputting cleaned window data ; S2, a risk assessment mechanism based on data credibility is used for any vehicle node in a vehicle group Window data of (2) Data consistency test is carried out to obtain vehicle nodes Is a comprehensive consistency score of (1) ; S3, constructing a vehicle node Scoring sub-items of (1) using comprehensive consistency scoring Consistency adjustment of scoring sub-items as evidence quality and output of each vehicle node Dynamic trust value of (2) Realizing vehicle node Continuous verification and dynamic update of trust status; S4, modeling node vulnerability based on subsystem vulnerability and performing smooth update to generate a model for measuring vehicle nodes At the moment of time Comprehensive vulnerability of risk fault tolerance capability ; S5, constructing a basic risk field in the road space, and according to the dynamic trust value And integrated vulnerability The risk field intensity of the basic risk field is regulated, a coupling risk field for comprehensively describing the multidimensional coupling risk of the communication, perception and control system is generated, and the bicycle risk and the vehicle group risk are calculated based on the coupling risk field; s6, identifying abnormal vehicle nodes in vehicle group And establishing a fault-tolerant fusion strategy for reflecting driving risks, generating a fault-tolerant scheme based on the fault-tolerant fusion strategy, and outputting a vehicle group total risk including a single vehicle risk set, an abnormal vehicle node list and a triggering reason.
2. The vehicle queue risk fault tolerance assessment method according to claim 1, wherein in step S1, the vehicle is caused to With vehicles Relative distance of (2) Difference in relative speed Wherein Respectively vehicles With vehicles Is provided in the position of (a), Respectively vehicles With vehicles The time to collision TTC satisfies the following equation: ; The zero trust triggering condition is that when meeting And is also provided with Triggering zero trust evaluation/information exchange; After triggering, the vehicle To its neighbor set Broadcasting an assessment request, adjacent vehicles Returning evidence data which it can provide, while the vehicle Locally collecting own sensor/controller data to form vehicle multi-source operation data 。
3. The vehicle fleet risk fault tolerance assessment method according to claim 1 or 2, wherein the vehicle multi-source operational data Comprising the following steps: operational status data position Speed of Acceleration of ; Communication state data, namely time delay, packet loss rate, bandwidth/occupation; interactive behavior data, namely information sharing times/frequency, collaborative execution success rate and emergency collaborative events; historical statistical data, namely historical trust average, historical anomaly times and historical successful interaction rate.
4. The vehicle fleet risk tolerance assessment method according to claim 1, wherein the risk assessment mechanism comprises consistency of location Speed consistency Consistency of acceleration Communication consistency Prediction consistency Wherein: By data in windows Internal calculation of position variance Then obtain the position consistency The calculation formula is as follows: ; Wherein, the A sensitive factor which is the evidence of the position consistency and is used for adjusting the punishment intensity of the degree of the position inconsistency in the consistency score; similarly, the velocity variance is calculated Speed uniformity The calculation formula of (2) is as follows: ; Wherein, the A sensitivity factor which is speed consistency evidence and is used for adjusting the punishment intensity of the speed inconsistency degree in consistency scores; Defining vehicle nodes At length of Time window of (2) In, the fluctuation range of the acceleration is The method comprises the following steps: ; Wherein, the For vehicle nodes At the moment of time Longitudinal or synthetic acceleration of (a); to reflect time window The overall severity of the internal acceleration change; Is provided with Is the tolerance threshold value of acceleration consistency, and the acceleration consistency Using segmentation rules, if Give a high score if Then the linearity reduces the consistency, the lowest clipping to 0, the expression is: ; ; ; Wherein, the Representing vehicle nodes At the moment of time Acceleration consistency of (2); representative acceleration fluctuation levels statistically derived during normal operation/low risk periods; the acceleration abnormal tolerance margin is used for avoiding the extreme value from generating infinite influence on the score; Communication consistency The stability of the delay and packet loss in the window can be determined, namely: Defining window data Standard deviation of inter-communication delay as fluctuation index The more stable the delay, the more the fluctuation index The smaller the expression: ; ; Wherein, the Representing a time window All delay data in Average value of (2); Expressed in a time window Inner time of day Delay data on the window, wherein N is the number of data points in the window; Defining window data Internal packet loss rate The method comprises the following steps: ; Wherein, the Indicating variable for communication packet loss for representing vehicle node At the moment of time If packet loss continues to occur, the packet loss rate Larger; the expression of the communication consistency score is: ; Wherein, the And (3) with Penalty coefficients of communication delay stability and packet loss persistence are respectively used for describing risk sensitivity degree of the system to communication instability; Defining a predicted deviation For the combination of position prediction error and velocity prediction error, namely: ; Wherein, the Respectively vehicle nodes At the moment of time Is a predicted position and an actual position of the vehicle; Respectively vehicle nodes At the moment of time Actual speed and predicted speed of (a); the scale balance coefficient is the position and speed prediction error; Then, predict consistency The expression of (2) is: ; Wherein, the For predicting the consistency penalty coefficient, the method is used for adjusting the influence intensity of the prediction deviation on the consistency score.
5. The vehicle fleet risk fault tolerance assessment method according to claim 4, wherein the integrated consistency score The expression is: ; Wherein, the And the weight coefficients correspond to the consistency respectively.
6. The vehicle queue risk fault tolerance assessment method according to claim 1, characterized in that in step S3, it comprises trust sub-item construction, consistency adjustment, annealed trust update and output, namely: Trust sub-item construction: construction of vehicle nodes Scoring sub-items and normalized to Including historical behavioral scoring Real-time status scoring Communication quality scoring Scoring of interaction behavior ; Consistency adjustment: will integrate consistency scores As evidence quality, comprehensive consistency scoring The lower the trust value update, the more cautious is, namely: When comprehensive consistency scores Below a threshold value When the real-time state score is reduced Communication quality scoring And interaction behavior scoring Or directly with conservative updates; When comprehensive consistency scores Above a threshold value When the weight is updated according to the normal weight; Updated trust value ; Annealing trust update: let the trust value at the last time be The time is comprehensively scored as And adopting annealing type update, namely: If comprehensive score Quickly regulating downwards, otherwise, slowly regulating upwards; And (3) outputting: outputting each vehicle node Dynamic trust value of (2) And a low trusted node set The expression is: ; Wherein, the Is a trust decision threshold for distinguishing between normally trusted nodes and low trusted nodes.
7. The vehicle fleet risk fault tolerance assessment method according to claim 1, wherein the integrated vulnerability Including communication subsystem vulnerability Perceived subsystem vulnerability And execution and control subsystem vulnerability The expression is: ; Wherein, the The self-adaptive weights of the communication subsystem, the perception subsystem and the control subsystem in the comprehensive fusion are respectively represented; to comprehensive vulnerability Make smooth update and output to adopt The method comprises the following steps: ; Wherein, the Is a smoothing coefficient; is a smoothed steady-state composite vulnerability.
8. The vehicle queue risk fault tolerance assessment method according to claim 1, characterized in that in step S5, it comprises: Uniformly modeling a road boundary, static barriers and adjacent vehicles as basic risk sources, and constructing a basic risk field in a road space, namely: ; Wherein, the Representing spatial position Is a comprehensive risk field; Representing spatial position Is a static risk field of (1); Representing spatial position Dynamic risk fields of (a); Representing spatial position Is a collision risk field of (a); According to vehicle nodes Is the combined vulnerability of (2) Constructing vulnerability-scaling functions The method comprises the following steps: ; Wherein, the For adjusting the coefficient; adjusting functions by vulnerability Amplifying the intensity of the basic risk field to obtain a risk field with regulated vulnerability The method comprises the following steps: ; According to vehicle nodes Dynamic trust value of (2) Constructing a trust adjustment function for adjusting the spatial attenuation characteristics of a risk field The method comprises the following steps: ; Wherein, the Is a range sensitivity coefficient; By trust adjusting functions For risk fields Is adjusted to obtain a coupling risk field when the dynamic trust value is At lower levels, the risk field is increased To reflect information uncertainty and uncertainty increase when dynamic trust values Higher, then risk field Is relatively convergent; In the coupling risk field, according to the vehicle node Current location of Calculating the bicycle risk index The method comprises the following steps: ; Then, corresponding bicycle risk indexes are carried out on all the vehicles And carrying out weighted average to form an overall index, namely obtaining the vehicle group risk.
9. The vehicle queue risk fault tolerance assessment method according to claim 8, wherein in step S6, the abnormal vehicle nodes in the cluster of vehicles Comprising the following steps: If the vehicle is at a node Dynamic trust value of (2) A low trusted node if the integrated consistency score If the continuous anomaly exceeds the length of the anomaly window Then it is a serious abnormal node; Wherein, the A trust decision threshold; Is a consistency threshold.
10. The vehicle fleet risk fault tolerance assessment method according to claim 9, wherein the fault tolerance fusion strategy is: Risk index for bicycle Setting fusion weights The method meets the requirements that the normal node weight is high, the low-reliability/low-consistency node weight is reduced, and the serious abnormal nodes can be set to be extremely low weight or are directly removed, so that the overall risk of few abnormal non-leading nodes is realized; Final output bicycle risk set Total risk of vehicle group Abnormal node list and trigger reason, to facilitate subsequent control or early warning.

Description

Vehicle queue risk fault tolerance assessment method based on zero trust architecture Technical Field The invention relates to the technical field of intelligent traffic systems, internet of vehicles safety and automatic driving risk assessment, in particular to a vehicle queue risk fault tolerance assessment method based on a zero trust architecture. The intelligent network vehicle linkage risk assessment method is suitable for intelligent network vehicle linkage group formation, collaborative driving and automatic driving safety control scenes. Background Along with the development of the internet of vehicles and the automatic driving technology, the intelligent network vehicle gradually evolves from single vehicle intelligence to group collaborative intelligence. Vehicles are highly dependent on vehicle-to-vehicle (V2V) communication with vehicle-to-road (V2I) and multi-source sensor data fusion in a fleet driving, collaborative decision-making process. However, in a real traffic environment, due to communication delay, packet loss, sensor faults, measurement noise and potential network attacks, the shared data among vehicles often presents characteristics of incomplete information and inconsistent data, and the safety and reliability of vehicle group decision are seriously affected. The existing car networking security scheme is mostly based on boundary trust or priori trusted assumption, and the internal nodes of the default car group are trusted and only defend external attacks. Once an internal node is abnormal or attacked, the error information may spread rapidly, amplifying the overall risk. In addition, the traditional risk assessment method is mainly based on a physical collision model, is difficult to reflect the comprehensive influence of vulnerability of a communication, perception and control system on driving risks, and lacks effective fault tolerance capability when part of nodes fail. The zero trust architecture provides a security concept of 'never trust and always verify', emphasizes continuous trust evaluation on the nodes based on verifiable evidence, and provides a new theoretical basis for Internet of vehicles security. However, the existing zero-trust research is concentrated on identity authentication and access control, and a systematic method for integrating data consistency and risk assessment for intelligent network coupling groups has not been formed yet. Therefore, it is needed to propose a driving risk assessment method with fault tolerance capable of performing dynamic trust assessment on intelligent network coupling nodes under the condition that information is not completely consistent with data in a zero trust architecture, so as to improve the safety and robustness of a vehicle group system in a complex traffic environment. Disclosure of Invention Aiming at the defects of the prior art, the invention provides a vehicle queue risk fault tolerance assessment method based on a zero trust architecture, which is used for solving the defects of the prior art pointed out in the background art, and specifically comprises the following steps: Firstly, the security architecture is limited, the existing car networking security scheme is mostly based on boundary trust or priori trusted assumption, the internal nodes of the default car group are trusted, only external attacks are defended, once the internal nodes are abnormal or attacked, error information can be rapidly diffused, and the overall risk is amplified, so that the traditional boundary protection mode can not be suitable for the complex security requirement of a modern intelligent network car networking system; Secondly, the risk assessment method is single, the traditional risk assessment method is mainly based on a physical collision model, the comprehensive influence of vulnerability of a communication, perception and control system on driving risk is difficult to reflect, and the methods lack effective assessment on key factors such as communication delay, data consistency and the like, and cannot comprehensively reflect the real risk condition of an intelligent network vehicle connection system; Thirdly, the fault tolerance is insufficient, the prior art lacks effective fault tolerance when part of nodes fail, the safe operation of the system under abnormal conditions is difficult to ensure, and when a fault node or an attacked node occurs in a vehicle group, the system cannot be timely identified and isolated, so that risk diffusion is caused; The fourth is that the zero trust architecture is not applied enough, the zero trust architecture provides a security concept of 'never trust and always verification', the continuous trust evaluation is emphasized on the nodes based on verifiable evidence, a new theoretical basis is provided for the safety of the internet of vehicles, however, the existing zero trust research is concentrated on identity authentication and access control, and a systematic method for intelligen