A vehicle platoon control and resilience assessment method based on zero trust architecture

Abstract

The present invention proposes a vehicle queue control and resilience assessment method based on a zero-trust architecture, which is used to accurately estimate the system state of a nonlinear network related to a vehicle queue, provide a reasonable control measurement sharing mechanism, and evaluate the ability to resist network attacks and quickly recover; the method is mainly divided into three parts: state estimation, two-layer controller design, and vehicle queue resilience assessment, specifically including: building a nonlinear network system for a vehicle queue; estimating the system state based on an extended Kalman filter; designing a two-layer controller based on state deviation and introducing node trust values ​​to achieve control of the vehicle queue; and evaluating the resilience of the vehicle queue based on node trust values; the method proposed in the present invention has the advantages of high flexibility and robustness of the two-layer controller; strong adaptability of the system state estimation; and a comprehensive system resilience assessment framework with high practical applicability, which can be applied to actual road driving conditions.

Description

Technical Field

[0001] The present invention relates to the field of intelligent transportation systems and resilience engineering technology, and in particular to a vehicle queue control and resilience assessment method based on a zero-trust architecture. Background Art

[0002] With the acceleration of urbanization and the continuous growth of traffic demand, traffic congestion is becoming increasingly serious. Intelligent Transportation System (ITS) has emerged. It aims to effectively integrate and apply advanced information technology, data communication technology, sensor technology, control technology, etc. to the entire transportation management system, so as to establish a real-time, accurate and efficient intelligent transportation system. As an important part of the intelligent transportation system, vehicle platoons need to formulate reasonable control strategies to complete the coordinated control between vehicles to improve road capacity and safety. However, the intelligent connected vehicle (ICV) system is in an open traffic environment, and each node in the system may be subject to malicious attacks, identity impersonation and unauthorized services. Therefore, for network systems such as vehicle platoons, it is necessary to evaluate the ability of network systems that apply control strategies to resist interference events and quickly recover from interference events.

[0003] In common control strategy research, the first vehicle in a vehicle queue is regarded as the leader, and other vehicles are regarded as followers to track the speed changes of the leading vehicle; and the expected spacing of the vehicle queue is set to a fixed value. In this case, the behavior of all following vehicles is highly dependent on the speed setting of the first vehicle. If the trust value of the first vehicle is low, it is susceptible to external interference events or network attacks, which can lead to improper speed settings and affect the driving stability of the entire vehicle queue; and the setting of fixed spacing cannot adapt to changes in factors such as traffic flow and vehicle behavior. The control strategy of the present invention obtains the target speed and expected spacing driven by the trust value according to the trust value, state information, predicted behavior, physical parameters, etc. of the vehicle node, allowing all nodes to make joint decisions, thereby utilizing global information to improve driving safety.

[0004] The resilience metric is mainly a function of network structure, network performance, interference type, interference duration and recovery strategy. Different interference types, interference intensities and recovery strategies may lead to differences in the resilience assessment results. For example, for the same network system, low-intensity interference shows higher resilience than high-intensity interference; the random link recovery strategy shows stronger resilience than the maximum link importance recovery strategy. The resilience assessment of the present invention combines the two factors of network structure and network performance to achieve a comprehensive resilience assessment for the recovery process without being affected by the interference type and interference intensity. Therefore, how to comprehensively and effectively evaluate the resilience of vehicle queues is a technical problem that needs to be solved urgently. Summary of the invention

[0005] In response to the problems existing in the above-mentioned prior art, the present invention proposes a vehicle queue control and resilience assessment method based on a zero-trust architecture, which realizes precise control of the vehicle queue through Kalman filtering and designing a two-layer controller. At the same time, it combines the two factors of network structure and network performance to reflect the actual operating status of the nodes and links inside the system, thereby solving the problem of defects in resilience assessment in the prior art.

[0006] In order to achieve the above technical objectives, the present invention provides the following technical solutions:

[0007] A vehicle platoon control and resilience assessment method based on zero-trust architecture includes three parts: state estimation, two-layer controller design, and vehicle platoon resilience assessment. The specific process is as follows:

[0008] S1. Construct a nonlinear network system about vehicle platoons, describe a single vehicle with a node, describe the communication relationship and distance relationship between nodes with a link, and describe a link as an edge connecting a pair of nodes in the network;

[0009] The network system also includes sensors configured on the vehicle; the sensors are used to observe system status data, including: vehicle position information, surrounding vehicle distance, and vehicle speed information;

[0010] S2, estimate the system state based on the extended Kalman filter; continuously adjust the system state estimate according to the latest observation data of the sensor in each time step until the minimum mean square error is met and the optimal system state estimate is obtained;

[0011] S3. Design a two-layer controller based on state deviation and introduce node trust value to realize the control of vehicle queue, including: selecting target speed; determining expected spacing; designing upper-layer PI controller to adjust spacing; designing lower-layer PID controller to control the queue speed to reach target speed;

[0012] S4. Evaluate the resilience of the vehicle queue based on the node trust value. When the network attack affects the node and link performance, analyze the resilience of the vehicle queue based on the recovery time of the vehicle queue facing the attack event in combination with the network performance and network structure. The network performance and recovery time are both determined based on the node trust value.

[0013] Based on the above technical solution, the method proposed by the present invention has the following beneficial effects:

[0014] 1. The controller is highly flexible and robust; the introduction of zero-trust architecture improves the robustness of information transmission and interaction between vehicles. In the control system of the vehicle queue, the vehicle queue faces various uncertainties or external interference. If some nodes fail due to network attacks, the weights of the corresponding error signals are adjusted to quickly redistribute control tasks, improve the system's fault tolerance, and enable the controller to better maintain the stability and performance of the system. Among them, the node weights can be dynamically adjusted according to real-time data and trust values, so that the controller can adapt to environmental changes and improve the adaptability of the system.

[0015] 2. The system state estimation is highly adaptable; due to the limited capacity of the communication channel, different sensors will generate data loss with different probabilities. This method uses a series of Bernoulli sequences to describe the packet loss rate of multiple sensors, which enables the recursive network extended Kalman filter to adapt to more complex network environments. In addition, the network extended Kalman filter has a recursive nature, and the state estimation of each step is based on the estimation of the previous step and the current observation to update the state estimation, without recalculating the estimation of the entire time series.

[0016] 3. The system resilience assessment framework is highly comprehensive and practically applicable. Compared with the traditional resilience assessment framework, the resilience assessment method based on the trust value of vehicle nodes and the actual operating status uses the node's response to network attacks to evaluate network performance, rather than relying solely on the network's communication topology, making the resilience assessment more comprehensive. Among them, network performance is based on the performance of nodes and links. The performance of nodes in a vehicle network consists of the node trust value and the degree of proximity between the vehicle node's speed and the target speed, which is quantified using speed deviation; while the performance of a link consists of the probability that the link transmits reliable information and the degree of proximity between the distance between two adjacent nodes and the expected spacing. The resilience of a vehicle platoon is quantified by restoring damaged nodes and links and the system's performance recovery rate during the recovery period. This resilience assessment framework has high practical applicability and can reflect the changes in the trust value and operating status of a vehicle platoon during the recovery process, focusing on performance curves.

[0017] 4. The selection of the target speed and expected spacing of the vehicle queue is more in line with the actual road driving conditions; each vehicle node gives a vehicle queue speed that is biased towards the speed of its own vehicle from the perspective of its own vehicle speed and the speed at the next moment, and jointly decides on a consistent queue target speed based on the proposals of all vehicle nodes for the vehicle queue speed. This target speed selection method abandons an idealized control target, that is, the first vehicle in the vehicle queue must be the leading vehicle that determines the target speed, making the control target more in line with the actual road driving conditions. Since vehicle queues are often composed of various types of vehicles, the expected spacing between adjacent vehicles is not necessarily consistent. Therefore, the selection of the expected spacing of the vehicle queue in this method is based on the communication trust value between adjacent vehicles, vehicle size, braking distance, etc., so as to formulate a safer and more reasonable control target. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] Figure 1 This is an overall flow chart of a vehicle platoon control and resilience assessment method based on a zero-trust architecture proposed in the present invention;

[0019] Figure 2 It is a flow chart of the system state estimation method based on extended Kalman filtering in the method proposed by the present invention;

[0020] Figure 3 A flowchart of a two-layer controller based on state deviation in the method proposed in the present invention;

[0021] Figure 4 It is a flow chart of the comprehensive vehicle platoon resilience evaluation method based on node trust value in the method proposed by the present invention;

[0022] Figure 5 This is a schematic diagram of the working principle of the extended Kalman filter;

[0023] Figure 6 It is a schematic diagram of the network performance recovery curve in the present invention. DETAILED DESCRIPTION

[0024] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0025] Although the steps in the present invention are arranged with numbers, they are not used to limit the order of the steps. Unless the order of the steps is clearly stated or the execution of a certain step requires other steps as a basis, the relative order of the steps can be adjusted. It is understood that the term "and / or" used in this article involves and covers any and all possible combinations of one or more of the associated listed items.

[0026] Please refer to Figure 1 , shows a vehicle queue control and resilience assessment method based on zero-trust architecture proposed in the present invention, including three parts: state estimation, two-layer controller design, and vehicle queue resilience assessment. The specific process is as follows:

[0027] S1. Construct a nonlinear network system about vehicle platoons, describe individual vehicles with nodes, and describe the communication relationship and distance relationship between nodes with links;

[0028] The network system also includes sensors configured on the vehicle; the sensors are used to observe system status data, including: vehicle position information, surrounding vehicle distance, and vehicle speed information;

[0029] It should be noted that in the present invention, each node is used to describe each vehicle, and an edge formed by a pair of nodes is used to describe a link in the network; for example, node i represents the current vehicle, i.e., the self-vehicle, node i-1 represents the front vehicle, and node 1 represents the leading vehicle; edge e ij That is, it represents the link between vehicle i and vehicle j, that is, the communication relationship and distance relationship between the two vehicles; these two expressions are used in this application from the perspectives of real physical meaning (vehicle, link) and network model design (node, link). All expressions about nodes and vehicles, edges and links in the article refer to the description here.

[0030] As a preferred implementation of step S1, in this embodiment, the sensor used is specifically:

[0031] Radar sensors to measure the distance and speed between vehicles;

[0032] Global Positioning System (GPS), which provides the vehicle’s real-time geographic location;

[0033] S2, estimate the system state based on the extended Kalman filter; continuously adjust the system state estimate according to the latest observation data of the sensor in each time step until the minimum mean square error is met and the optimal system state estimate is obtained;

[0034] As a preferred implementation of step S2, Figure 2 As shown, S2 specifically includes the following steps:

[0035] S21, sensor data collection: before data transmission, all sensors on the vehicle k The state information at the moment is observed, that is, the system state is obtained ; At the same time obtain k The observation value of the system state by each sensor at the moment, which is related to the system state The relationship can be expressed as:

[0036] ;

[0037] in, is the observation value of the system state by sensor r before data transmission, represents a real number; is the system state before data transfer, represents a real vector, and n represents the dimension; is the sensor measurement noise (scalar), which is independent of the system state; It means that sensor r measures the system state of a certain dimension;

[0038] S22, data transmission; use Bernoulli sequence to describe the packet loss rate of each sensor during data transmission. The packet loss rates of different sensors are independent of each other, and the packet loss rates of the same sensor at different times are also independent of each other. Considering the packet loss rate, after data transmission, the observation value of sensor r on the system state is recorded as , the formula is:

[0039] ;

[0040] in, Indicated in k Bernoulli random variables with known distribution at each moment; , indicating that sensor r has no data loss at time k; , indicating that sensor r loses data at time k, and can only receive data from sensor r at time k. k -1 Observation value at time ;

[0041] The packet loss rate of each sensor can be expressed as:

[0042] ;

[0043] in, express The probability of is the probability value; express The probability of represents the packet loss rate of sensor r;

[0044] Integrate the observation values ​​of all sensors after data transmission to obtain the system state after data transmission , the formula is:

[0045] ;

[0046] in, m is the number of sensors, Indicates that the rth sensor measures the system state, represents the measurement noise of the rth sensor; Indicates that the rth sensor is k -1 The observed value at time instant;

[0047] S23, prediction process; based on extended Kalman filter and sensor observations k The system status at the moment k +1 time to predict and estimate the system status; specifically including:

[0048] S231, perform a one-step prediction estimate of the system state:

[0049] k The system state at time +1 can be expressed as:

[0050] ;

[0051] in, is the parameter vector, is the system noise; is the control input, is the system noise gain matrix;

[0052] The state transfer function is as follows:

[0053] ;

[0054] in, for k- 1 moment k The system state estimate at time for k- 1 moment k- System state estimation at time 1;

[0055] S232, setting the covariance matrix of the system state estimation, and evaluating the consistency between the true value and the observed value;

[0056] The prediction covariance matrix formula of system state estimation is expressed as:

[0057] ;

[0058] in, for k -1 moment to k The predicted covariance matrix of the system state estimate at time , Var(.) is the covariance function; introducing partial derivatives, the above formula can be transformed into:

[0059] ;

[0060] in, is the partial derivative of the state transfer function with respect to the state, is the system noise The covariance matrix of

[0061] In this embodiment, the partial derivative of the state transfer function with respect to the state , which is the Jacobian matrix of the state transfer equation to the state, that is, the matrix calculated when the state transfer equation is linearized in the extended Kalman filter (EKF), and its formula can be expressed as:

[0062] ;

[0063] S24, the updating process includes: updating the system state estimation according to the observed value and the estimated value of the system state at the same time; updating the covariance matrix of the system state estimation according to the observed value and the predicted covariance matrix of the system state at the same time, specifically:

[0064] S241, update system state estimation; according to k The sensor measurement value at time is updated with a one-step prediction estimate, expressed as:

[0065] ;

[0066] in, for k Always k System state estimation at time is the residual sequence, which represents the difference between the actual observation value and the estimated value predicted based on the current estimation model; is the Kalman gain matrix, and the residual sequence Mapping to state space;

[0067] , The formulas are respectively expressed as:

[0068] ;

[0069] ;

[0070] in, yes The inverse matrix of is the residual sequence The covariance matrix of

[0071] S242, update the covariance matrix, the formula is expressed as:

[0072] ;

[0073] Introducing the observation value and prediction covariance matrix at time k, the above formula can be transformed into:

[0074] ;

[0075] in, represents the identity matrix;

[0076] The system state estimate at each moment is evaluated according to the updated covariance matrix, and the mean square error is minimized (that is, the trace of the covariance matrix is ​​minimized) to obtain the optimal system state estimate.

[0077] In this embodiment, the specific working principle of the extended Kalman filter is as follows: Figure 5 Shown.

[0078] S3. Design a two-layer controller based on state deviation and introduce node trust value to realize the control of vehicle queue, including: selecting target speed; determining expected spacing; designing upper-layer PI controller to adjust spacing; designing lower-layer PID controller to control the queue speed to reach target speed;

[0079] As a preferred implementation of step S3, Figure 3 As shown, step S3 specifically includes the following steps:

[0080] S31, establish the communication topology of the vehicle queue, use indicates; among them represents a set of nodes, i.e., each vehicle, represents an edge set; edge , ; Define the adjacency matrix of the vehicle platoon communication topology , if there is information flow from node i to node j in the communication network, that is, there is a communication link, then the edge , and let ,otherwise ;

[0081] S32, each node calculates the proposed queue speed based on its own status information , and vote for the target speed of the vehicle queue ; Specifically includes the following sub-steps:

[0082] S321, each node calculates a proposed queue speed based on the status information received from other nodes and combined with its own status information , its formula is expressed as:

[0083] ;

[0084] in, , Respectively represent the weights of the vehicle speeds predicted by node i and node j during speed update; , Respectively represent the speed of node i and node j at time t; , Respectively represent the predicted time step at time t The speed after There is information flow from node j to node i, It means that there is no information flow from node j to node i; represents the communication trust value of node j; represents the weight of the velocity update of node i relative to the velocity updates of other nodes;

[0085] S322, each node sends its calculated proposed queue speed to the central control system of the vehicle queue, and the central control system collects all the proposed speeds and feeds them back to each vehicle; the proposed queue speeds calculated by each node are different, recorded as , ;

[0086] S323, all nodes vote on the proposed queue speed, and each node i casts its only vote for the proposed queue speed of other nodes that is closest to the proposed queue speed of the vehicle itself. ;

[0087] in, ;

[0088] The trust value of each node is used as the weight of the ticket, and the proposed queue speed with the highest product of the number of votes and the weight is selected as the target speed of the vehicle queue. , the formula is:

[0089] ;

[0090] ;

[0091] ;

[0092] in, Indicates the proposed queue speed The total weighted votes of is the communication trust value of node i;

[0093] S324: Optimize and adjust the selection of the target speed based on the difference between the proposed queue speed of the leading vehicle (ie, the first node) in the vehicle queue and the target speed.

[0094] More specifically, the difference between the proposed queue speed of the leading vehicle and the target speed in step S324 has three cases:

[0095] Set the threshold for the difference between the target speed of the vehicle queue and the proposed speed of the leading vehicle. ; According to the difference, it can be divided into the following three cases:

[0096] Case 1: If the target speed of the vehicle queue is the same as the proposed queue speed of the leading vehicle Close, that is , then all vehicles in the vehicle queue adjust their control signals so that their speed gradually reaches the target speed ;

[0097] Case 2: If the target speed of the vehicle queue is much smaller than the proposed queue speed of the leading vehicle ,Right now ; Since the leading vehicle will not significantly reduce its speed for the sake of the following vehicles, all vehicle nodes re-elect a new target speed based on the latest data and trust value (appropriately increase or decrease the trust value of the leading vehicle); if the new target speed is always less than the queue speed proposed by the leading vehicle, it means that the trust value of the leading vehicle is low to a large extent, and if the trust value is about to fall below the trust threshold , a reward mechanism needs to be triggered, that is, if the leading vehicle gradually slows down, its trust value will be appropriately increased, otherwise the leading vehicle will be removed from the vehicle queue;

[0098] Case 3: If the target speed of the vehicle queue is much greater than the proposed queue speed of the leading vehicle ,Right now , then through the communication mechanism within the fleet, coordinate with the leading vehicle to gradually increase the vehicle speed without violating traffic rules and ensuring safe driving. This can smoothly transition to a higher target speed, avoid the impact of sudden acceleration on the fleet, and ensure the consistency of the vehicle queue speed; if the leading vehicle is unwilling to increase the driving speed, all vehicle nodes will re-elect a new target speed based on the latest data and trust value (appropriately increase or decrease the trust value of the leading vehicle); when the new target speed is always greater than the queue speed proposed by the leading vehicle, the reward mechanism is triggered: if the leading vehicle gradually accelerates, its trust value is appropriately increased, otherwise the leading vehicle is removed from the vehicle queue.

[0099] In this embodiment, reducing the trust value of the leading vehicle indicates that the leading vehicle is attacked by the network, resulting in unreliable data transmission, or the leading vehicle has selfish behavior that maliciously affects the speed decision of the vehicle queue, and the influence of the leading vehicle needs to be reduced; increasing the trust value of the leading vehicle is due to the fact that the leading vehicle gradually changes its speed and can continuously send accurate and reliable messages to the following vehicles. The advantage of setting up this decision-making mechanism is that when the leading vehicle makes an inappropriate speed decision, the subsequent following vehicles may not agree with the decision information of the first vehicle, thereby removing the first vehicle from the vehicle queue or incentivizing the first vehicle to make a correct decision. In addition, since the target speed of the vehicle queue is the result of a joint decision of all vehicle nodes, the vehicle queue can reach consistency faster than the decision to track the speed of the leading vehicle.

[0100] S33, determine the expected distance between vehicles; based on the physical trust value and communication trust value between node i-1 and node i, the expected distance between adjacent nodes is calculated taking into account the vehicle size, braking distance and safety distance between vehicles , the formula is:

[0101] ;

[0102] ;

[0103] ;

[0104] ;

[0105] in, , Represent the physical trust values ​​of node i and node i-1 respectively; represents the communication trust value of node i-1; and is an adjustment factor used to convert the communication trust value and the physical trust value into an actual spacing adjustment amount; represents the safe distance between vehicles calculated by node i-1; represents the vehicle size of node i, Represents the braking distance of node i, which is determined by the actual braking distance After normalization, we get q is the weight of vehicle size in physical trust; the physical trust value is determined by vehicle size and system distance, that is, the larger the vehicle size or the longer the braking distance, the lower its physical trust value; represents the maximum deceleration of node i; represents the speed of node i; T represents the driver's reaction time.

[0106] In this embodiment, for heterogeneous vehicle queues, the expected distance between two adjacent vehicles is affected by the front and rear vehicles. Vehicle node i calculates the expected distance from the front vehicle i-1. According to the trust values ​​of the front and rear vehicles (including communication trust and physical trust), the credibility of the information transmitted by the adjacent vehicles (i.e., communication trust value), vehicle size, braking distance, etc. are comprehensively considered. If node i or node i-1 is a truck, the vehicle size is large and the physical trust is low, and the expected distance between node i and node i-1 should be appropriately increased; if the communication trust of the front vehicle i-1 is low, the distance between the vehicle i and the front vehicle i-1 should be appropriately increased to ensure the safety of the vehicle. Therefore, the expected distance is decided by the adjacent vehicles together. ,In addition, it is necessary to ensure that the expected distance is greater than the safe distance between vehicles.

[0107] S34, design the upper PI controller to adjust the vehicle spacing, control the comprehensive error, and combine the target speed Calculate the expected speed of each car The comprehensive error includes the speed error and the distance error between vehicles. In this embodiment, the upper PI controller is a vehicle distance regulator. According to the distance error and speed error between the vehicle and the neighboring nodes in the communication topology, the PI controller is used to control the deviation to obtain the expected speed of vehicle i at time t. For the PLF (leader-follower) communication topology, except for the leading vehicle, each vehicle receives status information of the leading vehicle and the leading vehicle.

[0108] Therefore, step S34 specifically also includes the following sub-steps:

[0109] S341, input node status data; the status information of each node obtained based on the extended Kalman filter is used as the input of the upper PI controller; for the ego vehicle, the position and speed information of the preceding vehicle and the leading vehicle need to be obtained;

[0110] S342, calculate the distance error and speed error; record the distance error and speed error of the vehicle relative to the front vehicle as and ; The spacing error and speed error relative to the leading vehicle are and ; The formula is:

[0111] ;

[0112] ;

[0113] ;

[0114] ;

[0115] Then adjust it through the PI controller to get the comprehensive error of the vehicle relative to the front vehicle and the leading vehicle. and ; The formula is:

[0116] ;

[0117] ;

[0118] in, is the position of the vehicle, the subscript represents a specific vehicle, i represents the vehicle itself, i-1 represents the preceding vehicle, and 1 represents the leading vehicle; , is the PI controller gain of the ego vehicle relative to the preceding vehicle, , is the PI controller gain of the ego vehicle relative to the leading vehicle;

[0119] S343, based on the target speed selected in step S31 The expected speed of vehicle i at time t is calculated by the comprehensive error obtained in S342 , the formula is:

[0120] ;

[0121] in, represents the communication trust value of the preceding vehicle i-1, represents the communication trust value of the leading vehicle; it should be noted that in this application, t Both represent the moment; the weight can be dynamically adjusted according to the real-time node data interaction and trust value, so that the controller can adapt to environmental changes. The higher the node trust value, the more trust the message sent by the front vehicle or the leading vehicle, and the more accurate the deviation information. Whether the message sent by the node is reliable has nothing to do with the physical trust, so only the communication trust is considered here.

[0122] S35. Design the lower PID controller to control the speed of each vehicle to reach the expected speed of vehicle i given by the upper PI controller. , eliminate the speed and spacing errors between vehicles, and achieve the target speed of the vehicle queue ; Control signal The formula is:

[0123] ;

[0124] in, , , is the PID controller gain; Indicates the deviation between the desired speed and the actual speed.

[0125] In this embodiment, by designing a control strategy of a double-layer controller as described above, the vehicle node can efficiently and accurately reach the desired speed, the distance and speed error of the vehicle relative to the preceding vehicle and the leading vehicle can be continuously reduced, and the vehicle queue gradually reaches a consistent target speed, thereby realizing collaborative adaptive cruise control.

[0126] S4, evaluate the resilience of vehicle platoons based on node trust values;

[0127] In the actual driving process of vehicle queues, due to the open traffic environment, the vehicle queues may be affected by interference events during driving, including denial of service attacks, delay attacks, data spoofing attacks and other network attacks, resulting in:

[0128] ① The trust value of the node decreases, and it needs to be constantly re-verified and authorized to improve the trust value of the node;

[0129] ② The data received by the controller is tampered with, causing the status information of the vehicle node to change, etc.

[0130] Therefore, the ability of the vehicle queue to which the controller is applied to resist interference and quickly recover from interference is analyzed according to the response of the vehicle queue to the interference event. In this embodiment, during the recovery process, since the trust value of the node changes, the recovery process does not redetermine the target speed and the expected spacing, but restores the formation of the vehicle queue to the state before the network attack, including the vehicle speed and spacing.

[0131] In addition, since network attacks affect the performance of nodes and links, this embodiment comprehensively analyzes the resilience of the vehicle queue by combining network performance and network structure; therefore, as a preferred implementation of step S4, Figure 4 As shown, S4 specifically includes:

[0132] S41. Determine the node performance of the vehicle queue based on the communication status and driving status and link performance ;

[0133] Node performance It is composed of the node trust value and the degree of proximity between the vehicle node's speed and the target speed; the formula is expressed as:

[0134] ;

[0135] ;

[0136] ;

[0137] ;

[0138] in, represents the performance of node i; yes An increasing function of Indicates the communication trust value of node i, that is, the credibility of the message sent by node i; speed correlation index Reflects the actual speed of node i at any time With target speed the degree of proximity; It represents the difference between the actual speed of node i and the target speed at any time during the recovery period after the vehicle queue is attacked by the network; the target speed Indicates the speed of the vehicle queue before the network attack (the speed of the vehicle queue before the network attack has been agreed upon);

[0139] Link performance It is composed of the link trust value and the degree of proximity between the actual spacing and the expected spacing of adjacent vehicle nodes; the formula is expressed as:

[0140] ;

[0141] ;

[0142] ;

[0143] ;

[0144] in, Represents edge performance; yes An increasing function of For edge The link trust value is determined by the communication trust value of the nodes at both ends of the link; represents the actual distance between node i and node j, represents the expected distance between nodes i and j;

[0145] The closer it is to 1, the closer the actual speed of node i is to the target speed at time t, and the stronger the node's ability to resist network attacks.

[0146] S42. Considering the node performance and link performance in the time dimension, the network performance of the nonlinear network about the vehicle queue is obtained. , the formula is:

[0147] ;

[0148] S43, determine the recovery time of nodes and links; for a given topology network , there are N nodes and L links in total, the recovery time of node i is Determined by the recovery time of the trust value, ; Link recovery time It is determined by the maximum time it takes for the trust values ​​of the nodes at both ends of the link to recover. ;

[0149] S44. Performing a vehicle platoon resilience assessment based on the network topology and network performance; specifically comprising the following steps:

[0150] S441. Determine the recovery time of each node and link from network attacks, and select the maximum recovery time ;

[0151] S442. Calculate the performance recovery rate of network performance within the recovery time , its formula is expressed as:

[0152] ;

[0153] S443, remember At time t, the vehicle queue begins to be affected by the network attack, and the initial network performance before the network attack is calculated. , Network performance when a network attack occurs , Network performance after recovery ;

[0154] S444, Reference Resilience Centrality Index Then determine the resilience of the vehicle platoon , the formula is:

[0155] ;

[0156] in, Indicates in unit time The increments of the network recovery curve give each time point a weight reflecting its level of functionality; Indicates the duration of recovery; A time-weighted average representing the performance of the network in recovering its performance;

[0157] The maximum recovery time , Initial network performance and network performance during cyber attacks Introducing the above formula, the toughness of the vehicle queue is calculated .

[0158] In this embodiment, the resilience index is based on the network topology and network performance, and not only reflects the degree of recovery of network performance, but also reflects the changes in the network performance recovery curve, such as Figure 6shown.

[0159] More specifically, in step S444, the toughness of the vehicle queue is calculated There are two specific situations:

[0160] Case 1: The performance of the vehicle queue is fully restored during the recovery period, that is, ;

[0161] At this time there are:

[0162]

[0163] in ， Indicates the minimum time period for node trust value recovery, Indicates the minimum time period for link trust value recovery. and They represent the network performance recovery rate in the node and link recovery time periods respectively;

[0164] Case 2: The performance of the vehicle queue is not fully restored during the recovery period, that is, ;

[0165] Then we have:

[0166] ;

[0167] In both cases above, The larger the value, the stronger the resilience of the vehicle queue, that is, the stronger the ability of the vehicle queue to resist network attacks and recover performance. It can be seen that the method proposed in the present invention can make an accurate quantitative assessment of the resilience of the vehicle queue based on the trust value and driving state changes of the vehicle queue facing network attacks, and can also describe the performance recovery curve, which is more impressive.

[0168] In summary, the method proposed in the present invention has the advantages of high flexibility and strong robustness of the two-layer controller; strong adaptability of system state estimation; strong comprehensiveness and high practical applicability of the system resilience evaluation framework, and can be applied to actual road driving conditions.

[0169] In this specification, the terms "one embodiment", "some embodiments", "examples", "specific examples" or "some examples" refer to at least one embodiment or example described in combination with specific features, structures, materials or characteristics. These described specific features, structures, materials or characteristics can be combined in an appropriate manner in one or more embodiments or examples. In addition, the technician can combine and combine different embodiments or examples and their features described in this specification without contradiction.

[0170] The logic and / or steps shown in the flowchart or described in other ways can be regarded as a sequence of executable instructions for implementing logical functions. These instructions can be implemented in any computer-readable medium for use by instruction execution systems, devices or equipment. These systems, devices or equipment include processor systems or other systems capable of receiving and executing instructions.

[0171] The above embodiments have described the principles and implementations of the present invention in detail, and have used specific examples to illustrate the working principles thereof. These examples are only used to help understand the method of the present invention and its core concept. Meanwhile, according to the concept of the present invention, the actual implementation methods and application scope may vary. Therefore, the contents of this specification should not be construed as limiting the present invention.

Claims

1. A vehicle platoon control and resilience assessment method based on zero trust architecture, characterized in that: It includes three parts: state estimation, two-layer controller design, and vehicle platoon resilience assessment. The specific process is as follows: S1. Construct a nonlinear network system about vehicle platoons, describe a single vehicle with a node, describe the communication relationship and distance relationship between nodes with a link, and describe a link as an edge connecting a pair of nodes in the network; The network system also has sensors configured on the vehicle; The sensor is used to observe system status data, including: vehicle position information, surrounding vehicle distance, and vehicle speed information; S2, estimate the system state based on the extended Kalman filter; continuously adjust the system state estimate according to the latest observation data of the sensor in each time step until the minimum mean square error is met and the optimal system state estimate is obtained; S3. Design a two-layer controller based on state deviation and introduce node trust value to realize the control of vehicle queue, including: selecting target speed; determining expected spacing; designing upper-layer PI controller to adjust spacing; designing lower-layer PID controller to control the queue speed to reach target speed; S4. Evaluate the resilience of the vehicle queue based on the node trust value. When the network attack affects the node and link performance, analyze the resilience of the vehicle queue based on the recovery time of the vehicle queue facing the attack event in combination with the network performance and network structure. The network performance and recovery time are both determined based on the node trust value.

2. A vehicle platoon control and resilience assessment method based on zero trust architecture according to claim 1, characterized in that: Step S2 specifically includes: S21, sensor data collection: before data transmission, all sensors on the vehicle k The state information at the moment is observed, that is, the system state is obtained ; At the same time obtain k The observation value of the system state by each sensor at the moment, which is related to the system state The relationship can be expressed as: ； in, is the observation value of the system state by sensor r before data transmission, represents a real number; is the system state before data transfer, represents a real vector, and n represents the dimension; is the sensor measurement noise, which is independent of the system state; It means that sensor r measures the system state of a certain dimension; S22, data transmission; use Bernoulli sequence to describe the packet loss rate of each sensor during data transmission. The packet loss rates of different sensors are independent of each other, and the packet loss rates of the same sensor at different times are also independent of each other. Considering the packet loss rate, after data transmission, the observation value of sensor r on the system state is recorded as , the formula is: ； in, Indicated in k Bernoulli random variables with known distribution at each moment; , indicating that the sensor r is k No data loss at any time; , indicating that the sensor r is k Data loss occurs at all times, and only the sensor r can receive k -1 Observation value at time ; The packet loss rate of each sensor can be expressed as: ； in, express The probability of is the probability value; express The probability of represents the packet loss rate of sensor r; Integrate the observation values ​​of all sensors after data transmission to obtain the system state after data transmission , the formula is: ； in, m is the number of sensors, Indicates that the rth sensor measures the system state, represents the measurement noise of the rth sensor; Indicates that the rth sensor is k -1 The observed value at time instant; S23, prediction process; based on extended Kalman filter and sensor observations k The system state at time k+1 is predicted and estimated, including: S231, perform a one-step prediction estimate of the system state: The system state at time k+1 can be expressed as: ； in, is the parameter vector, is the system noise; is the control input, is the system noise gain matrix; The state transfer function is as follows: ； in, for k- 1 moment k The system state estimate at time for k- 1 moment k- System state estimation at time 1; S232, setting the covariance matrix of the system state estimation, and evaluating the consistency between the true value and the observed value; The prediction covariance matrix formula of system state estimation is expressed as: ； in, for k -1 moment to k The predicted covariance matrix of the system state estimate at time , Var(.) is the covariance function; introducing partial derivatives, the above formula can be transformed into: ； in, is the partial derivative of the state transfer function with respect to the state, is the system noise The covariance matrix of S24, the updating process includes: updating the system state estimation according to the observed value and the estimated value of the system state at the same time; updating the covariance matrix of the system state estimation according to the observed value and the predicted covariance matrix of the system state at the same time; and evaluating the system state estimation according to the updated covariance matrix.

3. The vehicle platoon control and resilience assessment method based on zero trust architecture according to claim 2 is characterized in that: Step S24 specifically includes: S241, update the system state estimate; update it with a one-step prediction estimate based on the sensor measurement value at time k, the formula is expressed as: ； in, for k Always k System state estimation at time is the residual sequence, which represents the difference between the actual observation value and the estimated value predicted based on the current estimation model; is the Kalman gain matrix, and the residual sequence Mapping to state space; , The formulas are respectively expressed as: ； ； in, yes The inverse matrix of is the residual sequence The covariance matrix of S242, update the covariance matrix, the formula is expressed as: ； Introducing the observation value and prediction covariance matrix at time k, the above formula can be transformed into: ； in, represents the identity matrix; The system state estimate at each moment is evaluated according to the updated covariance matrix, and the mean square error is minimized, that is, the trace of the covariance matrix is ​​minimized, and the optimal system state estimate is obtained.

4. The vehicle platoon control and resilience assessment method based on zero trust architecture according to claim 1, characterized in that: Step S3 specifically includes: S31, establish the communication topology of the vehicle queue, use indicates; among them represents a set of nodes, i.e., each vehicle, represents an edge set; edge , ; Define the adjacency matrix of the vehicle platoon communication topology , if there is information flow from node i to node j in the communication network, that is, there is a communication link, then the edge , and let ,otherwise ; S32, each node calculates the proposed queue speed based on its own status information , and vote for the target speed of the vehicle queue ; S33, determine the expected distance between vehicles; based on the physical trust value and communication trust value between node i-1 and node i, the expected distance between adjacent nodes is calculated taking into account the vehicle size, braking distance and safety distance between vehicles , the formula is: ； ； ； ； in, , Represent the physical trust values ​​of node i and node i-1 respectively; represents the communication trust value of node i-1; and is an adjustment factor used to convert the communication trust value and the physical trust value into an actual spacing adjustment amount; represents the safe distance between vehicles calculated by node i-1; represents the vehicle size of node i, Represents the braking distance of node i, which is determined by the actual braking distance After normalization, we get q is the weight of vehicle size in physical trust; the physical trust value is determined by vehicle size and braking distance; represents the maximum deceleration of node i; represents the speed of node i; T represents the driver's reaction time; S34, design the upper PI controller to adjust the vehicle spacing, control the comprehensive error, and combine the target speed Calculate the expected speed of each car ; The comprehensive error includes the speed error and the spacing error between vehicles; S35. Design the lower PID controller to control the speed of each vehicle to reach the expected speed of vehicle i given by the upper PI controller. , eliminate the speed and spacing errors between vehicles, and achieve the target speed of the vehicle queue ; Control signal The formula is: ； in, , , is the PID controller gain; Indicates the deviation between the desired speed and the actual speed.

5. The vehicle platoon control and resilience assessment method based on zero trust architecture according to claim 4 is characterized in that: Step S32 specifically includes: S321, each node calculates a proposed queue speed based on the status information received from other nodes and combined with its own status information , its formula is expressed as: ； in, , Respectively represent the weights of the vehicle speeds predicted by node i and node j during speed update; , Respectively represent the speed of node i and node j at time t; , Respectively represent the predicted time step at time t The speed after There is an information flow from node j to node i, It means that there is no information flow from node j to node i; represents the communication trust value of node j; represents the weight of the velocity update of node i relative to the velocity updates of other nodes; S322, each node sends its calculated proposed queue speed to the central control system of the vehicle queue, and the central control system collects all the proposed speeds and feeds them back to each vehicle; the proposed queue speeds calculated by each node are different, which are recorded as , ; S323, all nodes vote on the proposed queue speed, and each node i casts its only vote for the proposed queue speed of other nodes that is closest to the proposed queue speed of the vehicle itself. ; in, ; The trust value of each node is used as the weight of the ticket, and the proposed queue speed with the highest product of the number of votes and the weight is selected as the target speed of the vehicle queue. , the formula is: ； ； ； in, Indicates the proposed queue speed The total weighted votes of is the communication trust value of node i; S324: Optimize and adjust the selection of the target speed based on the difference between the proposed queue speed of the leading vehicle in the vehicle queue, that is, the first node, and the target speed.

6. A vehicle platoon control and resilience assessment method based on zero trust architecture according to claim 4, characterized in that: Step S34 specifically includes: S341, input node status data; the status information of each node obtained based on the extended Kalman filter is used as the input of the upper PI controller; for the ego vehicle, the position and speed information of the preceding vehicle and the leading vehicle need to be obtained; S342, calculate the distance error and speed error; record the distance error and speed error of the vehicle relative to the front vehicle as and ; The spacing error and speed error relative to the leading vehicle are and ; The formula is: ； ； ； ； Then adjust it through the PI controller to get the comprehensive error of the vehicle relative to the front vehicle and the leading vehicle. and ; The formula is: ； ； in, is the position of the vehicle, the subscript represents a specific vehicle, i represents the vehicle itself, i-1 represents the preceding vehicle, and 1 represents the leading vehicle; , is the PI controller gain of the ego vehicle relative to the preceding vehicle, , is the PI controller gain of the ego vehicle relative to the leading vehicle; S343, based on the target speed selected in step S31 The expected speed of vehicle i at time t is calculated by the comprehensive error obtained in S342 , the formula is: ； in, represents the communication trust value of the preceding vehicle i-1, represents the communication trust value of the leading vehicle, t For the moment.

7. The vehicle platoon control and resilience assessment method based on zero trust architecture according to claim 5, characterized in that: There are three cases for the difference between the proposed queue speed of the leading vehicle and the target speed in step S324: Set the threshold for the difference between the target speed of the vehicle queue and the proposed speed of the leading vehicle. ; According to the difference, it can be divided into the following three cases: Case 1: If the target speed of the vehicle queue is the same as the proposed queue speed of the leading vehicle Close, that is , then all vehicles in the vehicle queue adjust their control signals so that their speed gradually reaches the target speed ; Case 2: If the target speed of the vehicle queue is much smaller than the proposed queue speed of the leading vehicle ,Right now ; A new target speed is re-elected. If the new target speed is always lower than the queue speed proposed by the leading vehicle, the reward mechanism is triggered: if the leading vehicle gradually slows down, its trust value is increased; otherwise, the leading vehicle is removed from the vehicle queue; Case 3: If the target speed of the vehicle queue is much greater than the proposed queue speed of the leading vehicle ,Right now , the leading vehicle is ordered to gradually increase its speed under safe conditions; if the leading vehicle is unwilling to increase its speed, a new target speed is re-elected. When the new target speed is always greater than the queue speed proposed by the leading vehicle, the reward mechanism is triggered: if the leading vehicle gradually accelerates, its trust value is increased, otherwise the leading vehicle is removed from the vehicle queue.

8. The vehicle platoon control and resilience assessment method based on zero trust architecture according to claim 1, characterized in that: Step S4 specifically includes: S41. Determine the node performance of the vehicle queue based on the communication status and driving status and link performance ; Node performance It is composed of the node trust value and the proximity between the vehicle node's speed and the target speed; the formula is expressed as: ； ； ； ； in, Represents a set of nodes, i.e., individual vehicles; represents the performance of node i; yes An increasing function of Indicates the communication trust value of node i; speed correlation index Reflects the actual speed of node i at any time With target speed the degree of proximity; It represents the difference between the actual speed of node i and the target speed at any time during the recovery period after the vehicle queue is attacked by the network; the target speed Indicates the speed of the vehicle queue before the network attack, and the speed of the vehicle queue before the network attack has been agreed upon; Link performance It is composed of the link trust value and the degree of proximity between the actual spacing and the expected spacing of adjacent vehicle nodes; the formula is expressed as: ； ； ； ； in, Represents edge performance, represents an edge set; yes An increasing function of For edge The link trust value is determined by the communication trust value of the nodes at both ends of the link; represents the actual distance between node i and node j, represents the expected distance between nodes i and j; S42. Considering node performance and link performance in the time dimension, the network performance of the nonlinear network of the vehicle queue is obtained. , the formula is: ； S43, determine the recovery time of nodes and links; for a given topology network , there are N nodes and L links in total, the recovery time of node i is Determined by the recovery time of the trust value, ; Link recovery time It is determined by the maximum time it takes for the trust values ​​of the nodes at both ends of the link to recover. ; S44. Evaluate vehicle platoon resilience based on network topology and network performance.

9. The vehicle platoon control and resilience assessment method based on zero trust architecture according to claim 8, characterized in that: Step S44 specifically includes: S441. Determine the recovery time of each node and link from network attacks, and select the maximum recovery time ; S442. Calculate the performance recovery rate of network performance within the recovery time , its formula is expressed as: ； S443, remember At time t, the vehicle queue begins to be affected by the network attack, and the initial network performance before the network attack is calculated. , Network performance when a network attack occurs , Network performance after recovery ; S444, Reference Resilience Centrality Index Then determine the resilience of the vehicle platoon , the formula is: ； in, Indicates in unit time The increments of the network recovery curve give each time point a weight reflecting its level of functionality; Indicates the duration of recovery; A time-weighted average representing the performance of the network in recovering its performance; The maximum recovery time , Initial network performance and network performance during cyber attacks Introducing the above formula, we can calculate the toughness of the vehicle queue: .

10. The vehicle platoon control and resilience assessment method based on zero trust architecture according to claim 9, characterized in that: In step S444, the toughness of the vehicle queue is calculated There are two specific situations: Case 1: The performance of the vehicle queue is fully restored during the recovery period, that is, ; At this time there are: ， in ， Indicates the minimum time period for node trust value recovery. Indicates the minimum time period for link trust value recovery. and They represent the network performance recovery rate in the node and link recovery time periods respectively; Case 2: The performance of the vehicle queue is not fully restored during the recovery period, that is, ; Then we have: ； In both cases above, The larger the value, the stronger the resilience of the vehicle platoon, that is, the stronger the ability of the vehicle platoon to resist network attacks and recover performance.

Images