Search

CN-121973804-A - Vehicle fault-tolerant control method under abnormal condition of multivariable measurement sensor

CN121973804ACN 121973804 ACN121973804 ACN 121973804ACN-121973804-A

Abstract

A vehicle fault-tolerant control method under the abnormal condition of a multivariable measurement sensor belongs to the crossing field of electric digital data processing and intelligent driving vehicle control technology. The method comprises the steps of firstly conducting redundancy modeling and reliability assessment on a vehicle key state based on space-time correlation and physical consistency constraint of multi-source sensor data, then realizing real-time compensation and correction on failure observation by using a state reconstruction model driven by redundant data when sensor abnormality is detected, further combining a rapid fault identification and virtual sensor switching mechanism, constructing a fault-tolerant control strategy facing to minimum functional requirements of the vehicle, and ensuring that the vehicle can still maintain basic running and safety control capability under the condition that partial sensor functions are damaged. The intelligent vehicle control system and the intelligent vehicle control method thereof have the advantages that the real-time performance of the system is ensured, the safety and the robustness of the intelligent driving vehicle under complex environments and abnormal conditions are obviously improved, and the intelligent vehicle control system is suitable for the intelligent vehicle control system with multi-sensor fusion.

Inventors

  • CHEN YANFENG
  • YANG PEIRU
  • WANG YAN
  • LIU HAOQIAN
  • LIU ZIHAN

Assignees

  • 辽宁大学

Dates

Publication Date
20260505
Application Date
20260408

Claims (8)

  1. 1. A vehicle fault-tolerant control method under the abnormal condition of a multivariable measurement sensor is characterized in that: Step 1) unified modeling and physical consistency constraint construction of a multivariable measurement sensor, namely acquiring multivariable measurement data of a vehicle multisource sensor, performing time alignment processing on the measurement data of different sensors to construct a synchronous multivariable measurement set; Step 2) multivariate measurement consistency analysis and anomaly determination based on space-time redundancy, namely respectively establishing an observation model between multisource sensor measurement and vehicle states based on a unified state space model, and introducing sensor anomaly deviation items; Step 3) reconstructing the state based on the credibility weighting and generating a virtual sensor, namely carrying out space-time consistency analysis on the measurement results of each sensor based on the time redundancy characteristic and the space redundancy characteristic measured by the multi-source sensor in a preset time window, calculating the space-time consistency measurement corresponding to each sensor, and combining a physical consistency constraint set to obtain the credibility index of each sensor; Step 4) constructing a real-time fault-tolerant control strategy oriented to the minimum functional constraint of the vehicle, namely carrying out reliability weighting processing on the measurement of the multi-source sensor according to the reliability index of each sensor to construct a reconstruction estimation result of the key state of the vehicle; Step 5) a fault-tolerant mode switching and normal control recovery mechanism, namely constructing a state constraint set meeting the running requirement of the minimum function of the vehicle based on the reconstruction state and the effective observation set, and solving a fault-tolerant control instruction of the vehicle under the state constraint set so as to enable the vehicle to maintain the basic running and safety control function under the condition of abnormal or invalid sensor; Step 6) a system real-time operation and abnormal closed loop updating mechanism, namely switching between a normal control mode and a fault-tolerant control mode according to the change condition of the sensor reliability index and the vehicle operation state, reconfiguring control parameters in the fault-tolerant control mode, continuously monitoring an abnormal sensor, and smoothly switching back to the normal control mode when the sensor reliability is restored to a preset restoration threshold value; And 7) carrying out overall execution flow management and termination condition judgment of the method, namely circularly executing the processing flow formed in the steps 1) to 6) with a preset sampling period, carrying out on-line judgment on the stability of the vehicle state and the abnormality persistence of the sensor, and outputting the running state of the vehicle, the control instruction and the reliability information of the sensor when the preset termination condition is met, so as to finish the execution of the method.
  2. 2. The method for fault-tolerant control of a vehicle in the event of an abnormality of a multivariable measurement sensor according to claim 1, wherein the specific method in step 1) is as follows: under the condition that the sensor has abnormal or failure probability, a unified state description model irrelevant to the specific sensor type is constructed, and vehicle physical consistency constraint is introduced in advance, so that a basic model and a judgment basis are provided for subsequent abnormal identification, state reconstruction and fault tolerance control; during operation of the vehicle, acquisition of data from a plurality of onboard sensors at discrete moments Form a multivariate measurement set: Wherein, the Indicating the number of sensors to be used, Represent the first The observation vectors obtained by the individual sensors at the sampling moments of the individual sensors; Because of the difference between sampling frequency and time stamp of different sensors, the measurement data of each sensor is subjected to uniform time axis alignment processing and mapped to uniform system sampling time : Wherein, the The sampling period is uniformed for the system, Is the first Mapping relation from each sensor original timestamp to a unified time grid; After time alignment is completed, a synchronous measurement set for subsequent modeling is obtained: defining a vehicle at a moment based on vehicle kinematics and control requirements The key physical state vectors of (a) are: Wherein, the Representing the coordinates of the position of the vehicle in three dimensions, The vehicle speed component is represented as such, Representing a vehicle acceleration component; The key physical state vector is used as a unified expression form of the running state of the vehicle, is irrelevant to the type and the quantity of specific sensors and is used for describing the core physical characteristics of the vehicle in terms of space position, motion state and dynamic change; For the first The sensors establish a unified observation model between measurement data and vehicle states: Wherein the method comprises the steps of , Is the first The measurement mapping function of the individual sensors, In order to measure the noise vector, The measurement noise covariance matrix is corresponding; to describe the measurement characteristics of the sensor in abnormal, drifting or disturbed conditions, abnormal terms are introduced into the observation model: Wherein, the Represent the first The individual sensors are at the moment The abnormal deviation term of (2) is used for describing systematic errors or abnormal disturbance which are possibly existing in the output of the sensor; based on vehicle kinematics or dynamics, a discrete time transfer model of the vehicle state is constructed: in which the process noise , A vehicle state transfer function is represented and, For the vehicle to control the input vector, In order to process the noise vector of the process, A process noise covariance matrix; according to the physical accessibility and the safe operation requirement of the vehicle, constructing a physical consistency constraint set which needs to be met by the vehicle state: defining the speed and acceleration constraints as: Wherein, the And (3) with Respectively representing the maximum speed and the maximum acceleration threshold value allowed by the vehicle under the current working condition; After the processing is finished, a unified modeling result is output, wherein the unified modeling result comprises a vehicle state space model, synchronous measurement data, observation model parameters, a state transition model and a physical consistency constraint set and is used for subsequent space-time redundancy analysis and sensor abnormality judgment; In the running process of the system, measurement data of a plurality of sensors on the same physical state are synchronously collected to form a multi-source measurement set corresponding to time Wherein, the A set of all sensor measurements acquired at time tn; j Representing the measured value of the jth sensor at the time tn, wherein M is the number of sensors; Each sensor corresponds to a unique number j, j=1, 2, & gt, M, for subsequent error analysis and consistency determination during acquisition; according to the historical measurement sample, the experimental calibration result or the sensor technical parameter, respectively determining the random noise amplitude range and the communication delay range of each sensor; The noise amplitude range is defined as Wherein ε + represents the maximum deviation of the sensor measurement noise and the communication delay range is defined as Where τ - and τ + represent minimum and maximum delays, respectively; and determining the error distribution characteristics of the measurement output of each sensor through the parameters, establishing an error boundary model capable of comprehensively reflecting noise and delay uncertainty, and providing a quantitative basis for subsequent boundary correction and fusion analysis.
  3. 3. The method for fault-tolerant control of a vehicle in the event of an abnormality of a multivariable measurement sensor according to claim 1, wherein the specific method in step 2) is as follows: Based on the unified state model, the observation model and the physical consistency constraint constructed in the step 1), consistency analysis is carried out by utilizing the space-time redundancy relation measured by the multi-source sensor, so that the abnormal state of the sensor is rapidly judged, and the method aims at the first step of Each sensor builds its measurement sequence over a continuous time window: Wherein, the Is the time window length; based on the predicted state Calculating a corresponding predictive observation: Constructing a time residual sequence: And calculating a time consistency metric: Wherein, the Is the first The measurement noise covariance matrix of the individual sensors, For measuring stability and consistency of the sensor measurement in the time dimension; Sensor set with observation capability for same vehicle state variable Constructing a cross-sensor space consistency analysis model for any two sensors The observed differences are defined as: constructing a spatial consistency index based on observed noise characteristics: further to the first The consistency of the individual sensors in the spatial dimension is comprehensively evaluated: Wherein, the Represent the first The degree of spatial consistency of the individual sensors relative to the other redundant sensors; combining the time consistency index and the space consistency index to construct the first Spatiotemporal consistency metric for individual sensors: Wherein, the The consistency weight coefficient of the time dimension and the space dimension is used for adjusting the importance of two types of redundant information according to the operation condition of the system; Introducing the physical coherence constraint in step 1), the method comprising State component of individual sensor dominant estimation And (3) performing constraint detection: when the constraint condition is violated, a physical violation index is constructed: Wherein, the For reflecting the degree of deviation between the sensor measurement and the physical feasible region of the vehicle; Combining space-time consistency measurement and physical consistency index to construct the first Abnormality determination function of individual sensors: Wherein, the Punishment coefficients for physical consistency; calculating the sensor reliability according to the anomaly determination function: Wherein, the When the above condition is satisfied, determining the first step as the reliability threshold The individual sensors are at the moment Is in an abnormal state; finally, the abnormality judgment result of each sensor and the corresponding credibility index are output As direct inputs for sensor weight adjustment, state reconstruction and virtual sensor generation in step 3).
  4. 4. The method for fault-tolerant control of a vehicle in the event of an abnormality of a multivariable measurement sensor according to claim 1, wherein the specific method in step 3) is as follows: Obtaining the time of each sensor in step 2) Reliability index of (2) And then, carrying out weight normalization processing on the multisource sensors participating in state reconstruction, and constructing a credibility weighting coefficient: Wherein, the For the number of sensors currently participating in the estimation, Represent the first The relative contribution weights of the individual sensors to the vehicle state estimation at the current moment are as follows: Based on the unified state space model and the observation model constructed in the step 1), reconstructing and estimating the key state of the vehicle by utilizing the multisource measurement weighted by the credibility: Wherein, the At the moment for the vehicle Is used to reconstruct the state estimate vector of (c), Is the first time aligned The individual sensors measure a vector of values, For the corresponding observation mapping matrix, the calculation form is as follows: Wherein, the For a linearization matrix of the observation function at the current estimated point, Is the first A measurement noise covariance matrix of each sensor; when a certain sensor is determined to be abnormal in step 2), that is, the reliability thereof satisfies: , wherein, For the preset credibility threshold value, the vehicle state obtained based on reconstruction Generating a virtual measurement output of the sensor by a corresponding observation function: Wherein, the The sensor is used for replacing measured data of the sensor to participate in subsequent control and judgment during abnormal or invalid sensor; According to the reliability state of the sensor, the real measurement and the virtual measurement are adaptively switched and fused to form a final available observed quantity: Wherein, the As a switching coefficient based on reliability, it is defined as: Through the processing, the real data is directly adopted when the sensor is in a normal state, and the virtual sensor is output to replace when the sensor is abnormal or invalid, thereby ensuring the continuity and consistency of the vehicle state information under the abnormal working condition, and finally outputting the vehicle state reconstruction estimation result Fused set of effective observations As input for the vehicle minimum function constraint determination and fault tolerant control decision in the subsequent steps.
  5. 5. The method for fault-tolerant control of a vehicle in the event of an abnormality of a multivariable measurement sensor according to claim 1, wherein the specific method in step 4) is as follows: Based on the vehicle reconstruction state estimation and the effective observation set output in the step 3), constructing a constraint system and a control target around the minimum function operation requirement of the vehicle, solving an optimal fault-tolerant control instruction to ensure safe running under abnormal working conditions, and determining a minimum function operation state constraint set of the vehicle based on the basic running and the safety requirement which are still required to be maintained by the vehicle under the abnormal sensor condition, wherein the constraint system and the control target are used for limiting the feasible range of the vehicle state in the fault-tolerant control process: Wherein, the Further given to the set of physical coherence constraint functions constructed in step 1), the typical form of minimum function constraints: Wherein, the As an upper limit for the safe speed, For the yaw rate of the vehicle, As a threshold value for the yaw stability, The threshold is preset according to the dynamics characteristics of the vehicle and the safety strategy; Vehicle reconstruction state estimation output at step 3) On the basis, constructing a fault-tolerant control objective function oriented to the minimum functional constraint: Wherein, the Is a reference state vector in a minimum function mode of operation, The state error weighting matrix is used for adjusting the importance degree of different state variables in a control target; reference state According to the current running condition, the longitudinal stability, the transverse controllability and the gesture safety of the vehicle are preferentially ensured, and high-precision track tracking is not pursued; under the constraint of a vehicle dynamics model, the control input is optimized and solved: Wherein, the For the vehicle to control the input vector, A set of control inputs allowed in the minimum function mode for limiting the magnitude of change in steering angle, driving force or braking force, The method comprises the steps of obtaining an optimal fault-tolerant control instruction under a constraint condition; optimal fault-tolerant control instruction to be obtained Acts on the vehicle actuating mechanism and monitors the running state of the vehicle in real time: when the vehicle state continues to satisfy the minimum set of functional constraints When the reliability of the sensor is recovered to be above a threshold value, providing conditions for the follow-up exiting of the fault-tolerant mode and the recovery of normal control; The step is output as a real-time fault-tolerant control instruction sequence meeting the minimum function constraint As a direct control result of the vehicle maintaining safe operation in the event of sensor anomalies or failures.
  6. 6. The method for fault-tolerant control of a vehicle in the event of an abnormality of a multivariable measurement sensor according to claim 1, wherein the specific method in step 5) is as follows: Based on the sensor reliability index output in the step 2) and the effective observation set in the step 3), judging the running mode of the system, and triggering the fault-tolerant control mode when one of the following conditions is met: Wherein, the Is the first The individual sensors are at the moment Is used for the reliability of the (a) and (b), As a threshold value of the degree of certainty, In order to indicate the function, When the condition is met, the system is switched from the normal control mode to the fault-tolerant control mode; in the fault-tolerant control mode, maintaining the minimum function constraint control strategy constructed in the step 4), and reconfiguring control parameters: Wherein, the For a state error weight matrix in fault tolerant mode, For a set of control input constraints in fault tolerant mode, And (3) with For adjusting the coefficient, the system is used for reducing the dependence of the system on high-precision tracking performance, and preferentially ensuring the stability and safety of the vehicle; during fault-tolerant control mode operation, the system continuously monitors the status of the anomaly sensor and reevaluates its confidence level over a time window: Wherein, the The window length is sliding estimated for confidence, Is the first The average confidence of each sensor over a time window is considered to have recovered when the following conditions are met: Wherein, the To restore the decision threshold and satisfy ; When all key sensors participating in control meet the recovery condition, the system is smoothly switched from a fault-tolerant control mode to a normal control mode, and in order to avoid abrupt change of control instructions, control inputs are gradually fused: Wherein, the Is a control instruction in the normal control mode, For a control instruction in the fault-tolerant control mode, For a smooth switching coefficient that increases with time, the following are satisfied: When (when) And when the switching from the fault-tolerant control mode to the normal control mode is completed, outputting a system running mode state and a corresponding control instruction switching result, and ensuring the safe and continuous running of the vehicle after the sensor abnormality is eliminated.
  7. 7. The method for fault-tolerant control of a vehicle in the event of an abnormality of a multivariable measurement sensor according to claim 1, wherein the specific method in step 6) is as follows: In the running process of the vehicle, the system collects the measurement data of the multisource sensors in real time by the unified sampling period T s and updates the operator F by the state according to the processing results of the steps 1) to 3) ) Integrating the current observation and the historical state information to form a vehicle state estimation updated in real time: Wherein, the To fuse the actual and virtual measured active observation sets, The representation state updating operator is used for integrating current observation and historical state information and realizing continuous estimation of the key state of the vehicle; based on the sensor credibility and the mode state information obtained in the step 2) and the step 5), the abnormality judgment threshold value and the weight parameter are updated in a closed loop manner In order to prevent the history information of the abnormal sensor from generating accumulated influence on the subsequent estimation and control, carrying out dynamic attenuation processing on the data channel corresponding to the abnormal sensor, wherein the delta c (k) and the delta alpha (k) are self-adaptive adjustment amounts obtained by calculation according to the current running stability and the abnormal duration time respectively: Wherein, the The reliability attenuation coefficient is used for reducing the influence of the abnormal sensor on state estimation and control decision during the duration of the abnormality; Reducing the influence of the abnormal sensor on state estimation and control decision during the duration of the abnormality, and judging the stability of the state and control result of the vehicle in the real-time running process: Wherein, the And (3) with The control system is characterized in that the control system is respectively a state change threshold value and a control change threshold value, when the system continuously meets the conditions, the current running state is judged to be stable, and the vehicle state, a control instruction and sensor reliability information are output and used as the input of a next control period and an upper layer decision module.
  8. 8. The method for fault-tolerant control of a vehicle in the event of an abnormality of a multivariable measurement sensor according to claim 1, wherein the specific method in step 7) is as follows: during the running process of the vehicle, the processing flows formed in the steps 1) to 6) are used as a complete control period to be circularly scheduled and executed, and the system performs the operation of the vehicle in each sampling period And sequentially completing multivariable measurement acquisition, space-time redundancy consistency analysis, state reconstruction and virtual sensor generation, minimum function fault tolerance control, mode switching and closed loop updating processing and forming a periodic execution sequence: Wherein, the Is shown at the moment The executed complete method flow ensures that each functional module orderly cooperates in time and logically runs in a closed loop; In the cyclic execution process of the method, key operation indexes are monitored on line and used for evaluating the effectiveness of the current method execution: Wherein, the An index indicating the magnitude of change in the vehicle state estimation, When the index is continuously within a preset threshold value range, the method is considered to be stable and effective in executing process; Comprehensively judging the continuity of the sensor abnormality and the system running state, and constructing a safety state judgment condition: Wherein, the In order to continuously determine the window length, When the vehicle state always meets the physical consistency constraint in the continuous window and the number of the abnormal sensors does not exceed the threshold value, judging that the system is in a safe running state; Ending the execution cycle of the current method or exiting the fault tolerant control flow when any one of the following conditions is satisfied Wherein For the termination of the execution of the method, or When the method is terminated or the period is completed, the system outputs the running state of the vehicle, a control instruction sequence, a sensor credibility evolution result and a fault-tolerant control execution mark, and the fault-tolerant control execution mark is used for recording, analyzing or triggering subsequent processing such as manual taking over and the like by a vehicle upper layer decision system.

Description

Vehicle fault-tolerant control method under abnormal condition of multivariable measurement sensor Technical Field The invention belongs to the crossing field of electric digital data processing and intelligent driving vehicle control technology, in particular to a vehicle fault tolerance control method under the abnormal condition of a multivariable measurement sensor, which relates to the technologies of multisource sensor data synchronization, unified state space modeling, observation model construction, physical consistency constraint design, space-time redundancy analysis, state reconstruction, virtual sensor generation, fault tolerance control strategy optimization and the like, and is suitable for safety control of the intelligent driving vehicle fused by multiple sensors under the scenes of sensor failure, abnormal drift or physical attack and the like. Background With the rapid development of intelligent driving technology, vehicle sensing systems often integrate various types of sensors such as laser Radar (LiDAR), camera (Camera), millimeter wave Radar (Radar), and Inertial Measurement Unit (IMU). The sensors can sense the surrounding environment and the state of the vehicle from different modes and dimensions, provide high-precision measurement data for vehicle decision and control, and directly determine driving safety by the reliability of the sensors. However, in a complex road environment, the sensor is vulnerable to external interference, physical damage or malicious attack, and has problems of failure, abnormal drift and the like. The existing vehicle control system has obvious defects that (1) sensor abnormal scenes are not fully considered, systematic abnormality recognition and coping mechanisms are not available, the system is easy to be out of control due to single sensor faults, (2) space-time redundancy characteristics of multi-source sensor data are not fully mined, accurate reconstruction of vehicle states is difficult to achieve when part of sensors are abnormal, (3) a control strategy is not optimized for abnormal working conditions, basic running safety and controllability of a vehicle are difficult to maintain when sensor functions are damaged, and (4) a smooth transition mechanism is not available for mode switching, control command mutation is easy to be caused, and vehicle running stability is affected. Therefore, how to ensure the safe operation capability of the vehicle through the utilization of redundant data, state reconstruction and control strategy adjustment under the abnormal condition of the sensor becomes a key problem to be solved in an intelligent driving vehicle control system. Aiming at the problems, the invention provides a real-time fault-tolerant control method for fusing the sensor space-time redundant data and the vehicle dynamics characteristics, which improves the safety and the robustness of the intelligent driving vehicle under abnormal conditions. Disclosure of Invention The invention aims to solve the problem that the existing intelligent driving vehicle control system is difficult to maintain safe operation when a sensor is abnormal, and provides a vehicle fault-tolerant control method under the condition of abnormal multivariable measurement sensor. The technical scheme of the invention is as follows: In order to achieve the above object, the present invention provides a vehicle fault-tolerant control method under abnormal conditions of a multivariable measurement sensor, which is characterized by comprising the steps of: Step 1) unified modeling and physical consistency constraint construction of multivariable measurement sensor And acquiring the multivariable measurement data of the vehicle multisource sensor, performing time alignment processing on the measurement data of different sensors, and constructing a synchronous multivariable measurement set. And meanwhile, defining a vehicle key physical state vector irrelevant to the specific sensor type, and establishing a unified state space model of the vehicle state. Step 2) multivariate measurement consistency analysis and anomaly determination based on spatio-temporal redundancy And respectively establishing an observation model between the multi-source sensor measurement and the vehicle state based on the unified state space model, and introducing a sensor abnormal deviation term. Meanwhile, a vehicle dynamics state transition model is built, and a physical consistency constraint set which needs to be met by the vehicle state is built. And in a preset time window, based on the time redundancy characteristic and the space redundancy characteristic measured by the multi-source sensor, carrying out space-time consistency analysis on the measurement results of each sensor, calculating the space-time consistency measurement corresponding to each sensor, and combining the physical consistency constraint to obtain the reliability index of each sensor. Step 3) confidence weighting based state reconst