Commercial vehicle Beidou ETC device anti-cheating system and method based on multi-source information fusion

Abstract

The invention discloses a commercial vehicle Beidou ETC device anti-cheating system and method based on multi-source information fusion, and relates to the technical field of intelligent traffic and vehicle-mounted information safety crossing, and the system comprises a data collection module which is used for synchronous satellite positioning information, transaction information, vehicle attitude and motion information and vehicle bottom layer state information; the multi-source information fusion and cheat judgment module is used for executing motion state consistency and integrity monitoring based on adaptive extended Kalman filtering and performing multi-dimensional cheat behavior judgment in combination with ETC transaction information; and the data processing and response module is used for generating an evidence chain with mathematical traceability when the cheating behavior is judged to occur, and carrying out local storage and remote real-time reporting.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent transportation and vehicle information security intersection, and particularly relates to a commercial vehicle Beidou ETC device anti-fraud system and method based on multi-source information fusion. BACKGROUND

[0002] With the rapid development of highway network, the electronic toll collection (ETC) system has become the core infrastructure for improving traffic efficiency and realizing intelligent transportation. At the same time, according to relevant policies and regulations, heavy trucks, long-distance buses, and dangerous goods transport vehicles (referred to as "two passengers and one dangerous goods") must be installed with and use Beidou satellite navigation system vehicle terminals to realize real-time dynamic monitoring and management of vehicles. Therefore, intelligent vehicle terminals integrated with ETC functions and Beidou positioning functions have become a standard configuration in the commercial vehicle field. The toll fee of commercial vehicles is usually calculated in stages according to factors such as vehicle type, axle number, total weight, and driving mileage, and its rate is much higher than that of ordinary small passenger cars.

[0003] Currently, the main ETC fraud behaviors can be divided into two categories, namely identity fraud and path fraud.

[0004] Currently, the main techniques for dealing with vehicle terminal fraud signals can be divided into two categories: single information source-based detection and simple information association-based detection. Single information source-based detection mainly relies on the data source of a single sensor to identify abnormal behavior by analyzing its characteristics. The most commonly used method is GNSS signal integrity monitoring and CAN bus data anomaly analysis. Simple information association-based detection is to simply combine multiple information sources for preliminary comparison before or during the incident: such as simple comparison of GNSS and CAN bus data, ETC card and vehicle model background binding.

[0005] Simple comparison of GNSS and CAN bus data is a common approach that can provide basic prevention capabilities and can be used for post-inspection. However, this association is static and rough, lacking a dynamic and continuous verification mechanism. If the fraudster can simultaneously fake CAN bus data, this comparison will fail. At the same time, this method mainly relies on post-data analysis and lacks real-time warning capabilities, and cannot immediately alert or prevent fraud when it occurs.

[0006] The ETC card and vehicle model background binding method can effectively prevent identity fraud behaviors such as "fake license plates" or "card swapping". However, it relies on the accurate identification of gantry camera systems and can only be verified when ETC transactions occur. This method cannot handle illegal card swapping behaviors that are not recognized by the camera system or fraud that occurs during highway travel.

[0007] CN116794966A discloses a rail transit Beidou safety timing system, method, electronic device and storage medium, time synchronization is realized by communication between the control center and each station timing system, and anti-interference and anti-fraud reinforcement is carried out by using the timing safety protection unit. The clock consistency and timing safety of the rail transit are improved by the star-like timing system, and protection at the link level of timing is realized. However, the protection object is mainly the fixed nodes of the rail transit, and the core concern is "time synchronization" rather than "motion behavior verification". Essentially, it still belongs to the safety reinforcement category based on GNSS signal integrity monitoring, that is, whether it is interfered or cheated is judged by detecting characteristic parameters such as signal power, code phase and Doppler shift. Since this method cannot effectively distinguish between real satellite signals and highly simulated cheating signals, the protection ability mainly depends on the external signal protection unit rather than the dynamic consistency analysis inside the terminal. In addition, the rail transit scene is a closed system, and there is a lack of real-time coupling mechanism of multi-source data in the vehicle-mounted environment, so it is impossible to jointly identify the running track, path consistency and identity fraud of the vehicle.

[0008] CN116609797A discloses a GNSS cheating identification method, device, electronic equipment and storage medium, which matches and compares the GNSS positioning data obtained by the positioning device with the target trusted data provided by the safety terminal, so as to calculate the trustworthiness of the positioning data. The safety terminal signature verification mechanism is ingeniously used to improve the accuracy of the authenticity judgment of GNSS data, which can effectively reduce the false rejection rate and the false rejection rate. However, this invention essentially belongs to a static identification mechanism of single information source comparison, and the cheating identification process relies on external trusted data cache verification, lacking dynamic fusion of multi-source information such as CAN bus, speed sensor and ETC communication in the vehicle-mounted system. Its working logic is similar to the simple comparison of GNSS and CAN bus data, which can only provide prevention ability in post-event or periodic inspection, lacking continuous verification and real-time warning. When the cheater can simultaneously fake CAN bus data, the trustworthiness comparison of this method will fail. Moreover, no detection strategy is designed for identity fraud or card fraud, and no multi-modal consistency verification of vehicle dynamic behavior is realized, so there are obvious deficiencies in real-time, protection depth and adaptation to complex cheating scenes.

[0009] In summary, the existing technologies have more or less limitations. The detection method of single information source is easy to be bypassed by high-level cheaters, and the simple information correlation method lacks real-time and deep fusion ability, which is difficult to cope with complex and variable cheating scenes. This is the core problem to be solved by the present invention, that is, by deeply fusing multi-source information, a dynamic and high-precision motion state trust model is constructed, so as to realize active and real-time identification of cheating behaviors. SUMMARY

[0010] This section is intended to summarize some aspects of the embodiments of the present application and briefly introduce some preferred embodiments, and some simplifications or omissions may be made in this section and the abstract and title of the specification of the present application to avoid obscuring the purpose of this section, the abstract and the title, and such simplifications or omissions cannot be used to limit the scope of the present application.

[0011] In view of the above existing problems, the present application is proposed.

[0012] To solve the above technical problems, the present application provides the following technical solutions: a commercial vehicle Beidou ETC device anti-fraud system based on multi-source information fusion, characterized by comprising: a data acquisition module for synchronizing satellite positioning information, transaction information, vehicle attitude and motion information, and vehicle bottom layer state information; a multi-source information fusion and fraud judgment module for performing motion state consistency and integrity monitoring based on adaptive extended Kalman filtering, and combining ETC transaction information to perform multi-dimensional fraud behavior discrimination; a data processing and response module for generating a mathematically traceable evidence chain and performing local storage and remote real-time reporting when fraud behavior is determined to occur.

[0013] As a preferred scheme of the present application, the data acquisition module comprises: a Beidou GNSS module for calculating the longitude, latitude, elevation, speed, heading and time information of the vehicle according to satellite and other compatible GNSS signals, and outputting quality evaluation parameters; an ETC communication module using DSRC technology to communicate with a roadside unit, for reading the vehicle identity information in the ETC card and obtaining this transaction information when the vehicle passes through the ETC gantry; an IMU sensor module with a three-axis accelerometer and a three-axis gyroscope built-in, outputting three-axis acceleration and angular velocity data of the vehicle, and sensing the dynamic attitude and motion changes of the vehicle; a vehicle CAN bus interface module connected to the vehicle controller area network through an OBD interface or a direct connection, reading the state data of the vehicle bottom layer sensors and ECUs.

[0014] The present application also provides the following technical solutions: a commercial vehicle Beidou ETC device anti-fraud method based on multi-source information fusion, further comprising: synchronously collecting Beidou GNSS data, IMU data and CAN bus data with a unified time reference; using IMU data and CAN bus wheel speed data to predict the theoretical motion state of the vehicle at high frequency through a dead reckoning algorithm; using Beidou GNSS data as an observation value, and fusing it with the theoretical motion state, in the fusion process, dynamically adjusting the noise parameters of the Kalman filter, and performing Chi-Squared (χ 2) statistical hypothesis test of distribution; when the vehicle triggers transaction through the ETC portal, the multi-layer verification is immediately performed; the comprehensive verification result is adopted to finally judge the cheating behavior by using multi-level alarm logic, and the corresponding data storage, evidence generation and real-time alarm reporting process is triggered.

[0015] As a preferred scheme of the application, wherein: the high-frequency prediction of the vehicle's theoretical motion state by the dead reckoning algorithm comprises: using the acceleration, angular velocity provided by the IMU sensor module and the wheel speed information provided by the vehicle CAN bus interface module to construct a vehicle motion state trust model by an adaptive local Kalman filtering algorithm; using least squares or Kalman filtering to solve the measurement value of the Beidou GNSS module; taking the optimal estimation of the current time k in the vehicle motion state trust model as the prediction state of this Kalman filtering, and taking the solution result of the Beidou GNSS module as the observation value, to perform fusion and consistency test.

[0016] As a preferred scheme of the application, wherein: the Kalman filter core parameter is defined as follows: state vector , wherein, represents position, represents velocity, represents north, represents east, represents time; process noise covariance matrix Q and measurement noise covariance matrix R for wheel speed sensor; the initial value of the process noise covariance matrix Q is determined according to the data manual of the IMU sensor and the real vehicle test experience value under stable road conditions; the initial value of the measurement noise covariance matrix R for the wheel speed sensor is determined according to the accuracy specification of the wheel speed sensor and the real vehicle measurement value under the non-slip state.

[0017] As a preferred scheme of the application, wherein: the dynamic joint adaptive correction of the process noise covariance matrix Q comprises: γ=1+

[0018] wherein, γ is an amplification factor, >0 is a proportional coefficient, is a longitudinal acceleration, which is obtained in real time by the accelerometer of the vehicle-mounted IMU; is a yaw rate, which is obtained in real time by the vehicle-mounted IMU, is a motion intensity function, which is defined as:

[0019] wherein, and are the normalized reference values of the longitudinal acceleration and the yaw rate respectively; the corrected process noise covariance matrix is: ; the dynamic joint adaptive correction of the measurement noise covariance matrix R comprises:

[0020]

[0021] wherein, is a penalty factor, >0 is a proportional coefficient, represents the deviation of the inertial navigation interpretation speed from the wheel speed sensor observation value , is an ABS state indication quantity, wherein 1 represents ABS triggering, and 0 represents non-triggering, is a comprehensive weight function, and is defined as:

[0022] wherein, is speed normalization, is a weight coefficient; the corrected measurement noise covariance matrix for the wheel speed sensor is: .

[0023] As a preferred scheme of the present application, wherein: the Kalman filter performs prediction and update once every sampling period: The prediction is divided into attitude update and inertial navigation prediction; The attitude update utilizes the angular velocity measured by the IMU three-axis gyroscope to update the attitude of the vehicle in real time through integration; The quaternion method is used for attitude solution; The attitude is obtained based on the angular velocity increment of the IMU gyroscope within each sampling period, and the attitude at the current time is obtained by iterative integration from the attitude at the last time.

[0024] As a preferred scheme of the present application, wherein: the inertial navigation prediction comprises: The acceleration measured by the IMU three-axis accelerometer is projected from the carrier coordinate system to the local plane coordinate system through the attitude matrix , and the gravity acceleration g is deducted to obtain the acceleration of the vehicle in the local coordinate system ; The acceleration is used for dead reckoning to predict the state at the next time, and the predicted covariance matrix is calculated:​​

[0025]

[0026] wherein, is a state transition matrix, is a control input matrix, is an acceleration measurement provided by the IMU accelerometer, and T is a transpose; using wheel speed observations from the vehicle CAN bus updating the predicted state, including: first calculating the Kalman gain , the calculation formula is as follows: ; correcting the state vector and the covariance matrix :

[0027]

[0028] wherein I is an identity matrix, is the obtained wheel speed observation, is an observation matrix, which is mapped to the total forward speed through the heading angle ψ: .

[0029] As a preferred scheme of the present application, wherein: the fusion with the theoretical motion state includes: taking the state vector and the covariance matrix as the prior state and the prior covariance of the GNSS fusion at the current time respectively; obtaining the wheel speed observation from the Beidou GNSS module at the current time, and performing the update step of the Kalman filter to calculate the innovation vector :

[0030] calculating the normalized innovation square value NIS in the multi-source information fusion and fraud judgment module:

[0031]

[0032]

[0033] wherein, is the innovation covariance matrix, which has a dimension of 4x4; NIS is a dimensionless scalar, is a unit observation matrix of 4x4, is a GNSS measurement noise covariance matrix,

[0034] is a dynamically adjusted GNSS measurement noise covariance matrix, is a scaling factor related to observation quality.

[0035] As a preferred scheme of the present application, wherein: the multi-layer verification includes consistency verification of vehicle solidification identity and ETC card identity, and space-time consistency verification of gantry geographic position and current filtered position; the final discrimination of cheating behavior by using multi-level alarm logic includes: if the identity consistency verification fails or Chi-Squared (χ 2 ) test determines GNSS cheating, and space-time inconsistency occurs at the same time, it is determined as a cheating signal confirmation; if Chi-Squared (χ 2 ) test determines GNSS cheating, it is determined as a highly suspicious cheating signal; if single point result or short time space-time inconsistency occurs, but Chi-Squared (χ 2 ) test is normal, it is determined as transient anomaly.

[0036] The present application has the following beneficial effects: the present application not only accurately identifies identity cheating of "card switching" through identity consistency verification, but also effectively identifies "trajectory forgery" based on GNSS cheating through AEK and Chi-Squared test mechanism, thereby realizing comprehensive defense against mainstream ETC cheating methods; the adaptive filtering mechanism can closely track the real dynamics of the vehicle, and the discrimination method based on statistical hypothesis testing can effectively distinguish random noise and maliciously injected systematic bias, thereby greatly reducing the false positive rate in complex environments such as tunnel entrances and urban canyons compared with the fixed threshold method; the present application is completely based on mature filtering theory and statistical methods, and the computational complexity is controllable, without the need for special AI hardware acceleration units, and is very suitable for deployment on cost-sensitive and resource-limited vehicle embedded platforms; the system can quickly detect and respond within a few seconds of the occurrence of cheating attacks. The generated "mathematical traceable" evidence chain contains algorithm-level judgment basis (such as NIS value, test statistic, etc.), and has high evidence strength, thereby providing solid technical support for subsequent recovery. BRIEF DESCRIPTION OF DRAWINGS

[0037] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings needed in the embodiment description will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present application, and other drawings can be obtained by those skilled in the art without creative labor. Among them: Figure 1A function module structure schematic diagram of the commercial vehicle Beidou ETC device anti-spoofing system based on multi-source information fusion shown in the present application.

[0038] Figure 2 A whole flow chart of the commercial vehicle Beidou ETC device anti-spoofing method based on multi-source information fusion shown in the present application.

[0039] Figure 3 A self-adaptive Kalman filtering and chi-square test flow chart shown in the present application. DETAILED DESCRIPTION

[0040] In order to make the above objectives, characteristics and advantages of the present application more obvious and easy to understand, the specific embodiments of the present application will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are part of the embodiments of the present application, rather than all the embodiments.

[0041] Based on the embodiments in the present application, all other embodiments obtained by the ordinary person in the art without creative labor should belong to the protection scope of the present application.

[0042] In the following description, many specific details are set forth in order to provide a thorough understanding of the present application, but the present application can also be implemented in other ways different from those described herein, and those skilled in the art can make similar generalizations without departing from the connotation of the present application, therefore the present application is not limited to the specific embodiments disclosed below.

[0043] According to the embodiments of the present application, combined with Figure 1 The module structure schematic diagram shown in the present application, a commercial vehicle Beidou ETC device anti-spoofing system based on multi-source information fusion, comprising: A data acquisition module for synchronously acquiring satellite positioning information output by a Beidou GNSS module, transaction information acquired by an ETC communication module, vehicle attitude and motion information output by an IMU sensor module, and vehicle bottom layer state information read by a vehicle CAN bus interface module. A multi-source information fusion and spoofing judgment module, as the core of the system, is used for performing motion state consistency and integrity monitoring based on adaptive extended Kalman filtering (AEKF), and multi-dimensional spoofing behavior discrimination combined with ETC transaction information. A data processing and response module for generating an evidence chain with digital traceability when spoofing behavior is determined to occur, and performing local storage and remote real-time reporting.

[0044] Specifically, in the embodiments of the present application, the data acquisition module comprises: Beidou GNSS module: used to calculate the longitude, latitude, altitude, speed, heading and time information of the vehicle according to satellite and other compatible GNSS (such as GPS) signals, and output satellite quantity, signal quality, geometric precision factor (HDOP) and other quality evaluation parameters.

[0045] ETC communication module: using DSRC (Dedicated Short Range Communication) technology, communicating with Road Side Unit (RSU). When passing through the ETC gantry, it is responsible for reading the vehicle identity information (vehicle type, license plate number, etc.) in the ETC card, and obtaining the timestamp of this transaction, the precise geographic coordinates of the gantry, etc.

[0046] IMU sensor module: built-in three-axis accelerometer and three-axis gyroscope, outputting three-axis acceleration and angular velocity data of the vehicle at high frequency (for example, ≥100Hz), used to perceive the dynamic attitude and motion changes of the vehicle.

[0047] Vehicle CAN bus interface module: accessing the vehicle's controller area network through OBD interface or direct connection, reading the status data of the vehicle's underlying sensors and ECU (Electronic Control Unit), including at least the vehicle speed calculated by the wheel speed sensor, steering wheel angle, etc.

[0048] Multi-source information fusion and fraud judgment module internally solidifies the anti-fraud algorithm of the application, receives multi-source data from the data acquisition module, and performs real-time analysis according to the system preset significance level (P ), and judges whether it is cheated by chi-square test.

[0049] Data processing and response module is responsible for the action after fraud event judgment, including storing the fraud evidence chain to local storage, and reporting alarm information and evidence package to the background monitoring and inspection platform in real time through the vehicle communication module (such as 4G / 5G).

[0050] Under the guidance of the above embodiments, as shown in Figure 2 and Figure 3 , the application embodiment discloses another aspect of a commercial vehicle Beidou ETC device anti-fraud method based on multi-source information fusion, comprising: S1: synchronously collecting Beidou GNSS data, IMU data and CAN bus data with a unified time reference.

[0051] After the system is powered on, it is initialized, and each sensor module starts working. The system uses the second pulse (PPS) or high-precision time information provided by the Beidou GNSS module as the unified time reference of the entire system, ensuring that the data from different modules are strictly aligned in time.

[0052] S2: Using IMU data and CAN bus wheel speed data, the theoretical motion state of the vehicle is predicted by the dead reckoning algorithm.

[0053] S2.1: Using the acceleration, angular velocity provided by the IMU sensor module and the wheel speed information provided by the vehicle CAN bus interface module, a high-frequency and high-precision vehicle motion state trust model, i.e. the trusted core, is constructed by the adaptive local Kalman filter algorithm, which can provide continuous and high-precision vehicle position and speed information when GNSS signal is abnormal or lost, and provide reliable prediction reference for subsequent fraud detection.

[0054] S2.1.1: Define the parameters of the local Kalman filter, where the Kalman filter is used to fuse the IMU inertial navigation prediction value and the wheel speed measurement value, and the core parameters are defined as follows: State vector , used to describe the position and speed state of the vehicle in the local plane coordinate system, where, represents the position, represents the speed, represents the north direction, represents the east direction, represents the time.

[0055] State covariance matrix , representing the uncertainty of the state vector , with a dimension of 4x4. Its initial value is usually set to a larger diagonal matrix, and here the diagonal elements are set to 3600 to reflect the high uncertainty of the initial state of the system. With the running of the Kalman filter, the state covariance matrix will be updated continuously, and its diagonal elements will gradually decrease and tend to be stable.

[0056] Process noise covariance matrix Q, used to describe the uncertainty of the dead reckoning model (based on IMU and vehicle kinematics model).

[0057] Measurement noise covariance matrix R for wheel speed sensor, used to describe the uncertainty of the wheel speed measurement value.

[0058] It should be noted that in order to simplify the design of Kalman filter, improve the calculation efficiency and facilitate multi-source data fusion, the local plane coordinate system is selected here.

[0059] S2.1.2: The initial value of the process noise covariance matrix Q is determined according to the data manual of the IMU sensor (zero bias instability, random walk, etc.) and the real vehicle test experience value in stable road conditions.

[0060] The initial value of the measurement noise covariance matrix R of the wheel speed sensor is determined according to the accuracy specification of the wheel speed sensor and the real vehicle measurement value in the non-slip state.

[0061] To cope with the situation of the vehicle in different driving states, the application proposes a navigation fusion method based on adaptive adjustment of the covariance matrix. According to the dynamic state of the vehicle and the abnormal situation of the sensor, the process noise covariance matrix Q and the measurement noise covariance matrix R of the wheel speed sensor are dynamically and adaptively corrected: When it is detected from the vehicle CAN bus interface module that the vehicle is in a sharp turn (steering wheel angle is large), sharp acceleration or sharp deceleration state, or from the IMU sensor module that it is in a severe vibration, an amplification factor γ is introduced to the diagonal elements of the process noise covariance matrix Q, which is defined as follows: γ = 1 +

[0062] wherein, > 0 is a proportional coefficient, is the longitudinal acceleration, which is obtained in real time by the accelerometer of the vehicle-mounted IMU; is the yaw rate, which is obtained in real time by the vehicle-mounted IMU, is the motion intensity function, which is defined as:

[0063] wherein, and are the normalized reference values of the longitudinal acceleration and the yaw rate, respectively, which are used to eliminate the dimensional influence.

[0064] Therefore, the corrected process noise covariance matrix is: .

[0065] When the CAN bus detects that the wheel is slipping or the anti-lock braking system (ABS) is started, in order to reduce the trust degree of the wheel speed observation data, a penalty factor β is set to the diagonal elements of the measurement noise covariance matrix R, which is defined as follows:

[0066] =

[0067] wherein, > 0 is a proportional coefficient, represents the deviation of the inertial navigation interpretation speed from the wheel speed sensor observation value , ABS state indicator (1 represents ABS triggering, 0 represents non-triggering), is the ABS state indicator (1 represents ABS triggering, 0 represents non-triggering), The comprehensive weight function is defined as:

[0068] where, is the speed normalization, is the weight coefficient. Thus, the modified measurement noise covariance matrix is: .

[0069] S2.1.3: The Kalman filter performs prediction and update once per sampling period.

[0070] The prediction is divided into attitude update and inertial navigation prediction. The attitude update is to update the attitude of the vehicle in real time by integrating the angular velocity measured by the three-axis gyroscope of the IMU.

[0071] In order to avoid the gimbal lock problem inherent in Euler angles, quaternion method is used for attitude solution:

[0072] where, is the first derivative of the quaternion, representing the update rate of the attitude; is the attitude quaternion; is the quaternion representation of the IMU angular velocity in the body coordinate system; is the quaternion multiplication operator.

[0073] The attitude is obtained based on the angular velocity increment of the IMU gyroscope in each sampling period, and the attitude at the current time is obtained by iterative integration from the attitude at the last time.

[0074] The inertial navigation prediction uses the acceleration measured by the three-axis accelerometer of the IMU, projects it from the body coordinate system to the north / east (n / e) local plane coordinate system through the attitude matrix (converted from the quaternion q), and deducts the gravity acceleration g to obtain the acceleration of the vehicle in the local coordinate system:

[0075] Then, the acceleration is used for dead reckoning to predict the state at the next time and calculate the predicted covariance matrix :

[0076]

[0077] where, is the state transition matrix, is the control input matrix, is the acceleration measurement provided by the IMU accelerometer, and T is the transpose.

[0078] It should be noted that the state transition matrix is obtained by linearizing the system motion model; the control input matrix represents the IMU measurement on the position, velocity, and attitude change amount per unit time.

[0079] Update step: first calculate the Kalman gain , and the calculation formula is as follows:

[0080] In the above formula, the dynamically adjusted measurement noise covariance matrix R is used to calculate the Kalman gain When increases, decreases, thereby reducing the influence of the measurement value on the final state correction, and vice versa.

[0081] Further, the state vector and the covariance matrix are corrected as follows:

[0082]

[0083] where I is the unit matrix, is the wheel speed observation value obtained from the CAN bus, is the observation matrix, which is used to map the northward and eastward velocity components in the state vector to the total forward velocity through the heading angle ψ: .

[0084] where the heading angle is obtained from the attitude update process.

[0085] It should be noted that the Kalman filter uses the wheel speed sensor data to correct the velocity prediction obtained by the inertial navigation, effectively suppresses the accumulation of IMU errors, and outputs the optimal fused vehicle state and the covariance matrix .

[0086] It should be noted that the above fused prediction result is reliable for a short time. Considering that the IMU errors (such as zero offset and random noise) are integrated through acceleration and angle, leading to linear or quadratic growth of velocity error with time, and quadratic or cubic growth of position error with time, the short time is generally not more than 1 min.

[0087] S2.2: Solve the measurement value of Beidou GNSS module by least squares or Kalman filter, get the longitude, latitude, height and three-dimensional velocity of the vehicle in WGS84 coordinate system and the corresponding covariance matrix, and perform coordinate system transformation to local horizontal coordinate as the measurement value of multi-source information fusion and fraud judgment .

[0088] S2.3: Dynamically adjust the GNSS measurement noise covariance matrix , the initial value needs to be calibrated in advance by static and dynamic test or assigned according to the specification of Beidou GNSS module.

[0089] When receiving a lower number of satellites, a higher HDOP value (geometric precision factor), a lower signal carrier-to-noise ratio or detecting multipath effect from the Beidou GNSS module, the measurement noise covariance matrix is dynamically expanded to reduce the trust of the observation information. The specific method is:

[0090] wherein, is the observation quality related scaling factor, defined as:

[0091] wherein, is the number of visible satellites; is the geometric precision factor; is the average signal carrier-to-noise ratio; is the multipath effect index; ~ is the weighting coefficient; is the normalization function, which is used to map each index to the interval [0, 1].

[0092] S3: Take the Beidou GNSS data as the observation value, and fuse it with the predicted theoretical motion state. In the fusion process, dynamically adjust the noise parameters of the Kalman filter, and perform statistical hypothesis testing based on Chi-Squared (χ 2 ) distribution on the difference between the observation value and the predicted value, and finally update the state after fusion.

[0093] Take the and output by S2.1 as the prior state and prior covariance of this GNSS update.

[0094] First, the multi-source information fusion and fraud judgment module obtains the measurement value from the Beidou GNSS module at the current time, and performs the update step of Kalman filter to calculate the innovation vector :

[0095] in, It is the innovation vector, with a dimension of 4 x 1; It is a 4 x 4 identity matrix I.

[0096] Secondly, the normalized innovation square value (NIS) is calculated in the multi-source information fusion and deception judgment module.

[0097]

[0098]

[0099] in, Let NIS be the innovation covariance matrix with a dimension of 4×4. NIS is a dimensionless scalar that quantifies the magnitude of the innovation and takes into account uncertainty.

[0100] Under ideal conditions where the system is operating normally without being affected by deceptive signals and the measurement noise follows a Gaussian distribution, a single NIS value follows a chi-square distribution with degrees of freedom equal to the dimension of the observation vector. In this invention, since the observation vector contains position and velocity information and has a dimension of 4, the degrees of freedom of a single NIS value are 4.

[0101] Furthermore, to effectively distinguish between occasional signal anomalies and persistent malicious spoofing, this invention does not rely on a single NIS value for judgment, but instead employs a composite chi-square test based on a sliding window. The composite test statistic is:

[0102] Based on the properties of the chi-square distribution, the statistic... It follows a chi-square distribution with 4N degrees of freedom (N takes the value of 60).

[0103] Null hypothesis H0: GNSS observations and predictions are consistent, and the deviation comes only from Gaussian noise; Alternative hypothesis H1: GNSS observations are subject to systematic bias, i.e., they may be deceived.

[0104] The system presets a significance level of α = 0.001 (i.e., a confidence level of 99.9%). When the statistic... When the corresponding threshold is exceeded, the null hypothesis is rejected, the integrity of the GNSS data is determined to be compromised, and a spoofing alarm is triggered.

[0105] It is important to note that there is a trade-off between the sliding window size N and the detection latency and false alarm rate. The smaller the sliding window size N, the lower the detection latency, but the higher the false alarm rate; conversely, the larger the sliding window size N, the lower the false alarm rate, but the higher the detection latency.

[0106] Finally, if the above checks pass, the fused state update is performed, and the Kalman filter gain matrix is ​​calculated. and the updated optimal state estimate Otherwise, no state update will be performed.

[0107] S4: When a vehicle triggers a transaction by passing through an ETC gantry, immediately perform consistency checks on the vehicle's fixed identity, the ETC card identity, and the spatiotemporal consistency check between the gantry's geographical location and the current filtered location.

[0108] Identity consistency verification: Compare the vehicle information (license plate, vehicle model) read from the ETC card with the vehicle information pre-installed in the ETC system to see if they are completely consistent. If they are inconsistent, it is directly judged as "identity fraud".

[0109] Spatiotemporal consistency verification: Compare the distance between the precise geographic coordinates issued by the ETC gantry and the optimal position output by the current Kalman filter of the system, i.e., the output result of S3. If the distance exceeds a reasonable threshold, this invention sets the threshold to 100 meters, and then determines it as a "spatiotemporal anomaly".

[0110] It should be noted that this invention, by triggering identity and spatiotemporal consistency verification when an ETC transaction occurs, can complete dual verification against "identity fraud" and "path fraud" at the instant of the actual vehicle transaction, significantly improving the real-time performance and targeting of anti-fraud detection. The event-driven mechanism can proactively complete identity comparison and spatial consistency judgment without relying on continuous monitoring, avoiding the computational redundancy and response delay caused by traditional periodic detection. Simultaneously, based on real-time difference analysis of gantry coordinates and filter positions, anomalies can be identified in time before fraudulent behavior causes economic losses, achieving simultaneous identification of GNSS path anomalies and identity forgery, thereby improving the protection accuracy and dynamic response capability of the vehicle-mounted system.

[0111] S5: Based on the comprehensive integrity monitoring results, identity verification results, and spatiotemporal verification results, a multi-level alarm logic is used to make the final judgment on the deceptive behavior and trigger the corresponding data storage, evidence generation, and real-time alarm reporting process.

[0112] Spoofing signal confirmation (highest level): Identity consistency check failed or Chi-Squared (χ²) 2 The test results indicated GNSS spoofing, and at the same time, spatiotemporal inconsistency occurred.

[0113] Highly suspicious deception signal (second highest level): Chi-Squared (χ 2 ) Verify and determine GNSS deception.

[0114] Transient anomalies (low level): Inconsistencies in spatiotemporal data at a single location or over short periods, but Chi-Squared (χ²) 2The test results were normal.

[0115] Based on the discrimination level, the data processing and response module performs the corresponding operation.

[0116] For the highest and second-highest level of deception, a mathematically traceable chain of evidence will be immediately generated, including the original sensor data, Kalman filter intermediate variables (NIS value sequence, Q, R, conditioning records), and test statistics, and then reported in real time.

[0117] Furthermore, after completing one cycle, the system returns to step S2.2 to continue real-time monitoring.

[0118] The system also includes one or more processors and memory.

[0119] The memory is used to store operable instructions that, when executed by the one or more processors, cause the one or more processors to perform operations, including the processes of the anti-fraud system for commercial vehicle Beidou ETC devices based on multi-source information fusion as described in the foregoing embodiments, especially... Figure 1 The flowchart of the system is shown.

[0120] Other aspects disclosed in the embodiments of the present invention also propose a computer-readable medium for storing software including instructions executable by one or more computers, which, upon execution, cause the one or more computers to perform operations including the flow of the anti-fraud system for commercial vehicle Beidou ETC devices based on multi-source information fusion as described in the foregoing embodiments, particularly... Figure 1 The flowchart of the system is shown.

[0121] It should be recognized that embodiments of the present invention may be implemented or carried out by computer hardware, a combination of hardware and software, or by computer instructions stored in a non-transitory computer-readable storage medium.

[0122] The method can be implemented using standard programming techniques, including a non-transitory computer-readable storage medium configured with a computer program in the computer program, wherein the storage medium is configured such that the computer operates in a specific and predefined manner.

[0123] Each program can be implemented in a high-level procedural or object-oriented programming language to communicate with the computer system; however, if required, the program can be implemented in assembly or machine language.

[0124] In any case, the language can be either compiled or interpreted.

[0125] Furthermore, for this purpose, the program can run on programmed application-specific integrated circuits.

[0126] The processes described herein (or variations and / or combinations thereof) can be executed under the control of one or more computer systems configured with executable instructions, and can be implemented by hardware or a combination thereof as code (e.g., executable instructions, one or more computer programs, or one or more applications) that commonly executes on one or more processors. The computer program includes a plurality of instructions executable by one or more processors.

[0127] Furthermore, the method can be implemented in any suitable computing platform, including but not limited to personal computers, minicomputers, mainframes, workstations, networked or distributed computing environments, standalone or integrated computer platforms, or in communication with charged particle tools or other imaging devices.

[0128] Various aspects of the present invention can be implemented in machine-readable code stored on a non-transitory storage medium or device, whether portable or integrated into a computing platform, such as a hard disk, optical read and / or write storage medium, RAM, ROM, etc., such that it can be read by a programmable computer, and when the storage medium or device is read by the computer, it can be used to configure and operate the computer to perform the processes described herein.

[0129] Furthermore, machine-readable code, or parts thereof, can be transmitted via wired or wireless networks.

[0130] When such media includes instructions or programs that combine with a microprocessor or other data processor to implement the steps described above, the invention described herein includes these and other different types of non-transitory computer-readable storage media.

[0131] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A Beidou ETC device anti-fraud system for commercial vehicles based on multi-source information fusion, characterized in that, include: The data acquisition module is used to synchronize satellite positioning information, transaction information, vehicle attitude and motion information, and vehicle underlying status information. The multi-source information fusion and deception judgment module is used to perform motion state consistency and integrity monitoring based on adaptive extended Kalman filtering, and to judge multi-dimensional deception behavior in combination with ETC transaction information; The data processing and response module is used to generate a digitally traceable chain of evidence when fraudulent behavior is determined to have occurred, and to store it locally and report it remotely in real time.

2. The multi-source information fusion based anti-spoofing system for Beidou ETC device of commercial vehicle according to claim 1, characterized in that: The data acquisition module includes: The BeiDou GNSS module is used to calculate the vehicle's longitude, latitude, altitude, speed, heading, and time information based on satellite and other compatible GNSS signals, and output quality assessment parameters. The ETC communication module uses DSRC technology to communicate with the roadside unit. It is used to read the vehicle identity information in the ETC card and obtain the transaction information when the vehicle passes through the ETC gantry. The IMU sensor module, which integrates a three-axis accelerometer and a three-axis gyroscope, outputs three-axis acceleration and angular velocity data of the vehicle to sense the vehicle's dynamic attitude and motion changes. The vehicle CAN bus interface module connects to the vehicle's controller area network via the OBD interface or direct connection to read the status data of the vehicle's underlying sensors and ECU.

3. A method for preventing fraud of a Beidou ETC device for commercial vehicles based on multi-source information fusion, based on the system for preventing fraud of the Beidou ETC device for commercial vehicles based on multi-source information fusion according to any one of claims 1-2, characterized in that: Also includes: Simultaneously collect BeiDou GNSS data, IMU data, and CAN bus data using a unified time reference; Using IMU data and CAN bus wheel speed data, the theoretical motion state of the vehicle is predicted at high frequency through dead reckoning algorithm; Fusing the Beidou GNSS data as observation values with the theoretical motion state, in the fusion process, dynamically adjusting the noise parameters of Kalman filter, and performing statistical hypothesis testing based on Chi-Squared (χ 2 ) distribution on the difference between the observation values and the theoretical motion state; When a vehicle triggers a transaction by passing through an ETC gantry, multi-level verification is immediately performed; Based on the comprehensive verification results, a multi-level alarm logic is used to make a final judgment on the deceptive behavior and trigger the corresponding data storage, evidence generation and real-time alarm reporting process.

4. The anti-spoofing method for a commercial vehicle Beidou ETC device based on multi-source information fusion according to claim 3, characterized in that: The high-frequency prediction of the vehicle's theoretical motion state using dead reckoning algorithms includes: Using acceleration and angular velocity information provided by the IMU sensor module and wheel speed information provided by the vehicle CAN bus interface module, an adaptive local Kalman filter algorithm is used to construct a vehicle motion state trust model. The measurement values ​​of the BeiDou GNSS module are solved using least squares or Kalman filtering; In the vehicle motion state trust model, the optimal estimate of the current time k is used as the predicted state of this Kalman filter, and the solution result of the Beidou GNSS module is used as the observation value for fusion and consistency verification.

5. The anti-spoofing method for a commercial vehicle Beidou ETC device based on multi-source information fusion according to claim 4, characterized in that: The core parameters of the Kalman filter are defined as follows: state vector wherein, denotes position, denotes velocity, denotes north, denotes east, denotes time; The process noise covariance matrix Q and the measurement noise covariance matrix R for the wheel speed sensor; The initial value of the process noise covariance matrix Q is determined based on the IMU sensor datasheet and empirical values ​​from real vehicle tests under stable road conditions. The initial value of the measurement noise covariance matrix R for the wheel speed sensor is determined based on the accuracy specifications of the wheel speed sensor and the actual vehicle measurement value under non-slip conditions.

6. The anti-spoofing method for a commercial vehicle Beidou ETC device based on multi-source information fusion according to claim 5, characterized in that: The dynamic joint adaptive correction of the process noise covariance matrix Q includes: γ=1+ ; where γ is a magnification factor, >0 is a proportionality coefficient, is the longitudinal acceleration, which is obtained in real time by the accelerometer of the vehicle-mounted IMU; is the yaw rate, which is obtained in real time by the vehicle-mounted IMU, is the motion intensity function, which is defined as: ； wherein and are normalized reference values for the longitudinal acceleration and the yaw rate, respectively. Revised process noise covariance matrix is: ; The dynamic joint adaptive correction of the measurement noise covariance matrix R includes: ； = ； wherein is a penalty factor, > 0 is a proportionality factor, denotes the inertial navigation interpreted velocity deviation from the wheel speed sensor observation , is an ABS state indicator, wherein 1 indicates ABS triggered and 0 indicates not triggered, is a combined weight function defined as: ； wherein, is speed normalized, is a weight coefficient; Corrected measurement noise covariance matrix for wheel speed sensors are: .

7. The anti-spoofing method for a commercial vehicle Beidou ETC device based on multi-source information fusion according to claim 6, characterized in that: The Kalman filter performs a prediction and update once per sampling period: The prediction comprises a pose update and an inertial navigation prediction; The pose update utilizes the angular velocity measured by the IMU three-axis gyroscope to update the pose of the vehicle in real time through integration; A quaternion method is adopted for pose solution; The pose is obtained based on the angular velocity increment of the IMU gyroscope in each sampling period, and the pose at the current time is obtained by recursively integrating the pose at the last time through iteration.

8. The anti-spoofing method for a commercial vehicle Beidou ETC device based on multi-source information fusion according to claim 7, characterized in that: The inertial navigation prediction comprises: Acceleration measured using an IMU triaxial accelerometer Through the attitude matrix By projecting from the vehicle coordinate system to the local plane coordinate system and subtracting the gravitational acceleration g, we obtain the vehicle's acceleration in the local coordinate system. ; Utilizing acceleration dead reckoning, predicting a state vector at a next time and computing a predicted covariance matrix : ； ； wherein is a state transition matrix, is a control input matrix, is an acceleration measurement provided by the IM accelerometer, and T is the transpose. Utilizing wheel speed observations from a vehicle can bus updating the predicted state, including: calculating the kalman gain The formula is as follows: ； The state vector and the covariance matrix are corrected: ； ； where I is the identity matrix, is the acquired wheel speed observation, is the observation matrix, mapping to the total forward velocity: via the heading angle .

9. The anti-spoofing method for a commercial vehicle Beidou ETC device based on multi-source information fusion according to claim 8, characterized in that: The fusion with the theoretical motion state comprises: state vector and covariance matrix , respectively, as the current time GNSS fusion prior state and prior covariance ; Obtain wheel speed observation value from Beidou GNSS module at current time and perform the update step of Kalman filtering to calculate innovation vector : ； A normalized innovation square value NIS is calculated in the multi-source information fusion and fraud judgment module: ； ； ； wherein, is the innovation covariance matrix, dimension 4x4; NIS is a dimensionless scalar, is the 4x4 identity observation matrix, is the GNSS measurement noise covariance matrix, is the dynamically adjusted GNSS measurement noise covariance matrix, is the observation quality dependent scaling factor.

10. The anti-spoofing method for a commercial vehicle Beidou ETC device based on multi-source information fusion according to claim 9, characterized in that: The multi-layer verification comprises consistency verification of the vehicle fixed identity and the ETC card identity, and space-time consistency verification of the gantry geographic position and the current filtered position; Final discrimination of the fraud behavior is performed by adopting a multi-level alarm logic, which comprises: If the identity consistency check fails or the Chi-Squared (χ 2 ) test determines that GNSS spoofing has occurred, and a spatiotemporal inconsistency has occurred at the same time, the spoofing signal is confirmed. If the Chi-Squared (χ 2 ) test determines that the GNSS spoofing, then it is determined to be a highly suspicious spoofing signal; If the single point result or short time space is inconsistent, but the Chi-Squared (χ 2 ) test is normal, it is determined as transient anomaly.

Images