Vehicle speed estimation method, device and hybrid vehicle

Abstract

The application relates to the technical field of vehicle motion control, and discloses a vehicle speed estimation method, a vehicle speed estimation device and a hybrid vehicle. The method comprises the following steps: acquiring estimation reference data of a target vehicle; wherein the estimation reference data comprises basic measurement data and derived measurement data; the derived measurement data comprises gravity component data and gravity change parameters; performing running condition judgment and wheel speed arbitration on the target vehicle based on the basic measurement data, so as to obtain original reference speed of the target vehicle; and performing vehicle speed estimation based on the estimation reference data and the original reference speed by using a Kalman filtering algorithm, so as to obtain target reference vehicle speed of the target vehicle; wherein an observation vector of the Kalman filtering algorithm comprises the original reference speed and the gravity change parameters. The method solves the problem of estimation failure of vehicle speed in an extreme condition in a traditional scheme, realizes stable and accurate longitudinal reference vehicle speed estimation in all conditions, and improves the robustness of vehicle speed estimation in a slope extreme condition.

Description

Technical Field

[0001] This application relates to the field of vehicle motion control technology, and in particular to a vehicle speed estimation method, device, and hybrid vehicle. Background Technology

[0002] Real-time vehicle speed is a crucial parameter for vehicle control. However, the vehicle's actual longitudinal speed cannot be directly obtained from sensors. It is usually necessary to indirectly estimate it by fusing multi-source sensor signals. The resulting estimate is the longitudinal reference speed, which can serve as the raw input parameter for core functions such as engine and motor torque coordination, shifting strategies, regenerative braking efficiency, and vehicle stability control.

[0003] Hybrid vehicles, equipped with both a combustion engine and an electric drive system, can flexibly switch or couple power outputs under different driving conditions and have been widely used due to their excellent fuel economy and power performance. However, while the multi-power-source architecture improves fuel economy and power performance, it also makes the power system control more complex, placing higher demands on the real-time performance and accuracy of vehicle speed signals. A common approach in related technologies is to fuse wheel speed signals collected by wheel speed sensors with acceleration signals output by an inertial measurement unit (IMU) to estimate the vehicle's longitudinal reference speed. However, the accuracy of this method decreases significantly under extreme conditions such as uphill / downhill roads, wheel slippage, or wheel lockup.

[0004] To address the aforementioned limitations, some related technologies further incorporate operating condition identification, that is, correcting wheel speeds based on different driving conditions before estimating vehicle speed. However, in extreme operating conditions such as vehicles going uphill or downhill, the longitudinal acceleration measured by the inertial measurement unit is coupled with the gravity component corresponding to the slope. This gravity interference cannot be eliminated by correcting wheel speeds based on operating conditions, thus directly introducing acceleration integration error, ultimately leading to a significant deviation in the estimated reference vehicle speed.

[0005] Therefore, a vehicle speed estimation method that can maintain high estimation accuracy under extreme operating conditions is needed. Summary of the Invention

[0006] This application aims to at least partially solve one of the technical problems in related technologies. To this end, this application proposes a vehicle speed estimation method, device, and hybrid vehicle. The main technical solutions adopted in this application include: In a first aspect, this application provides a vehicle speed estimation method, which includes: acquiring estimated reference data of a target vehicle; wherein the estimated reference data includes basic measurement data and derived measurement data; the derived measurement data includes gravity component data and gravity variation parameters; the gravity component data refers to the gravitational acceleration component in the longitudinal direction of the target vehicle; the gravity variation parameters refer to the rate of change of the gravitational acceleration component over time; performing operating condition judgment and wheel speed arbitration on the target vehicle based on the basic measurement data to obtain the original reference speed of the target vehicle; and using a Kalman filter algorithm to estimate the vehicle speed based on the estimated reference data and the original reference speed to obtain the target reference speed of the target vehicle; wherein the observation vector of the Kalman filter algorithm includes the original reference speed and the gravity variation parameters.

[0007] By introducing independently calculated gravity component data and gravity variation parameters, a pitch gravity observation channel completely decoupled from vehicle speed was constructed, solving the estimation failure problems in extreme conditions such as high coupling between vehicle speed and slope and four-wheel slippage on slopes, which are common in traditional schemes. A wheel speed arbitration mechanism for different operating conditions can match the optimal wheel speed observation results under different operating states, improving the observation accuracy under normal conditions. A multi-state Kalman filter framework including jerkiness balances steady-state estimation accuracy and dynamic condition tracking response speed. Ultimately, stable and accurate longitudinal reference vehicle speed estimation is achieved under all operating conditions, improving the robustness of vehicle speed estimation under extreme slope conditions.

[0008] Optionally, the Kalman filter algorithm is used to estimate the vehicle speed based on the estimated reference data and the original reference speed to obtain the target reference vehicle speed. This includes: constructing a state vector and an observation vector based on the estimated reference data and the original reference speed; wherein the state vector includes jerk, gravity component data and gravity change parameters; and using the state vector and the observation vector to perform state prediction and update processing to iteratively calculate the target reference vehicle speed.

[0009] By using Kalman filtering design, on the one hand, the slope gravity interference and the actual motion acceleration are decoupled by using an independent pitch angular velocity calculation channel. On the other hand, second-order advance tracking under rapid acceleration and deceleration conditions is realized by using the jerk state. Compared with traditional Kalman filters, the accuracy, smoothness and response speed of vehicle speed estimation are significantly improved under extreme conditions such as four-wheel slippage, wheel lock-up, rapid acceleration and emergency braking on slopes.

[0010] Optionally, the target vehicle's operating condition is determined and wheel speed is arbitrated based on the basic measurement data to obtain the target vehicle's original reference speed. This includes: determining the target vehicle's operating condition based on the basic measurement data to obtain the operating condition; wherein the operating condition includes normal operating condition, slipping operating condition, lock-up operating condition, and abnormal yaw operating condition; and performing wheel speed arbitration calculation based on the operating condition and the basic measurement data to obtain the target vehicle's original reference speed.

[0011] Through a two-level processing architecture of working condition identification and wheel speed arbitration, the system first classifies the driving state based on multi-source sensor signals, and then outputs reliable wheel speed observations by matching corresponding rules. This can effectively filter wheel speed interference under abnormal working conditions such as slippage, lock-up, and yaw instability, and provide vehicle speed observation input with controllable quality for subsequent Kalman filtering, thereby improving the robustness and accuracy of the final vehicle speed estimation results from the source of observation.

[0012] Optionally, the operating condition of the target vehicle is determined based on the basic measurement data to obtain the operating condition, including: using the basic measurement data and preset conditions to make a state judgment to obtain a state judgment result; wherein, the basic measurement data includes real-time wheel speed data and inertial measurement data; and the operating condition is determined based on the state judgment result.

[0013] The system employs a method of comparing basic measurement data with preset conditions one by one. Its computational logic is simple and highly real-time, enabling rapid identification of operating conditions within each algorithm step. Simultaneously, the judgment threshold is dynamically adjusted using a lookup table, adapting to different vehicle speeds, slopes, and steering angles, effectively improving the accuracy and scenario adaptability of the judgment results.

[0014] Optionally, the preset conditions include a first condition, a second condition, a third condition, and a fourth condition; determining the operating condition based on the state judgment result includes: if the state judgment result shows that the real-time wheel speed data meets the first condition, determining the operating condition as a normal operating condition; if the state judgment result shows that the inertial measurement data meets the second condition, determining the operating condition as a slipping operating condition; if the state judgment result shows that the inertial measurement data meets the third condition, determining the operating condition as a locked-up operating condition; and if the state judgment result shows that the inertial measurement data meets the fourth condition, determining the operating condition as an abnormal yaw operating condition.

[0015] By subdividing the operating conditions into four states—normal, slippage, lock-up, and abnormal yaw—and defining clear judgment conditions for each state based on real-time wheel speed data and inertial measurement data, the system can accurately identify the current adhesion and stability state of the vehicle under different operating conditions, thereby providing a reliable decision-making basis for subsequent wheel speed arbitration strategy selection.

[0016] Optionally, wheel speed arbitration calculation is performed based on operating conditions and basic measurement data to obtain the original reference speed of the target vehicle, including: determining the wheel speed arbitration strategy of the target vehicle based on operating conditions; wherein the wheel speed arbitration strategy includes a weighted strategy and an extreme value strategy; and calculating the original reference speed based on the wheel speed arbitration strategy and basic measurement data.

[0017] By classifying wheel speed arbitration strategies into two main types—weighted strategies and extreme value strategies—the logic becomes simpler, facilitating engineering implementation and parameter tuning. Furthermore, it achieves optimal filtering and fusion of wheel speed signals under different attachment states, providing high-quality speed observation input for subsequent Kalman filtering.

[0018] Optionally, the wheel speed arbitration strategy for the target vehicle is determined based on the operating conditions, including: if the operating conditions are normal or abnormal yaw conditions, the weighted strategy is used as the wheel speed arbitration strategy; if the operating conditions are slipping or lock-up conditions, the extreme value strategy is used as the wheel speed arbitration strategy.

[0019] For different operating conditions, corresponding arbitration strategies are matched. Under normal and abnormal yaw conditions, weighted fusion is used to balance estimation accuracy and result stability. Under slippage and lock-up conditions, extreme value selection is used to quickly avoid abnormal wheel speed interference. This ensures that the original reference speed maintains relatively optimal reliability and representativeness under various operating conditions.

[0020] Optionally, state prediction and update processing is performed using state vectors and observation vectors to iteratively calculate the target reference vehicle speed. This includes: determining the stability of the target vehicle based on basic measurement data to obtain a vehicle stability level; wherein the vehicle stability level includes a first stable state, a second stable state, and a third stable state; the first stable state includes the target vehicle being in a four-wheel slippage or four-wheel lockup situation; the second stable state includes the target vehicle being in a non-four-wheel slippage / lockup situation or an abnormal yaw situation; determining the filtering parameters of the Kalman filter algorithm based on the vehicle stability level; and performing state prediction and update processing based on the filtering parameters, state vectors, and observation vectors to iteratively calculate the target reference vehicle speed.

[0021] By using dynamically adjusted filtering parameters, the Kalman filter can adaptively balance the weights between predicted and observed values ​​under different operating conditions, thereby maintaining the robustness and accuracy of vehicle speed estimation across the entire operating range.

[0022] Optionally, obtaining estimation reference data for the target vehicle includes: obtaining basic measurement data of the target vehicle; and performing mathematical transformation processing based on the basic measurement data to obtain derived measurement data.

[0023] The system adopts a hierarchical data processing architecture that first collects basic measurement data and then generates derived measurement data through mathematical transformation. This architecture not only preserves the true attributes of the original sensor signals but also generates specific parameters required for each computational stage. All derived data are derived from existing vehicle-mounted sensors, eliminating the need for additional hardware acquisition equipment. This architecture can improve the vehicle speed estimation capability under conventional vehicle sensor configurations.

[0024] Optionally, the basic measurement data includes real-time wheel speed data and inertial measurement data; mathematical transformations are performed on the basic measurement data to obtain derived measurement data, including: integrating the pitch angular velocity in the inertial measurement data to obtain the vehicle's pitch angle; performing trigonometric function calculations based on the vehicle's pitch angle to obtain gravity component data; and performing differentiation based on the gravity component data to obtain gravity variation parameters.

[0025] Using only the pitch angular velocity signal from the inertial measurement unit (IMU), gravity component data and gravity variation parameters can be independently calculated. The calculation process does not rely on wheel speed signals or vehicle speed estimation results, yielding gravity parameters completely decoupled from vehicle speed. This provides a speed-independent calculation approach for subsequent Kalman filtering. It fundamentally solves the problem that IMU acceleration measurements cannot distinguish between actual motion acceleration and gravity components under slope conditions.

[0026] Secondly, this application provides a vehicle speed estimation device, the device comprising: The data acquisition module is used to acquire estimation reference data for the target vehicle. The estimation reference data includes basic measurement data and derived measurement data. The derived measurement data includes gravity component data and gravity variation parameters. The gravity component data refers to the gravitational acceleration component in the longitudinal direction of the target vehicle. The gravity variation parameters refer to the rate of change of the gravitational acceleration component over time. The operating condition judgment module is used to judge the operating condition of the target vehicle and arbitrate the wheel speed based on the basic measurement data in order to obtain the original reference speed of the target vehicle. The speed estimation module is used to estimate the vehicle speed based on the estimated reference data and the original reference speed using the Kalman filter algorithm, so as to obtain the target reference speed of the target vehicle; wherein, the observation vector of the Kalman filter algorithm includes the original reference speed and the gravity change parameters.

[0027] Thirdly, this application provides a hybrid vehicle including a memory and a processor, the memory including instructions loaded and executed by the processor, which, when executed, cause the processor to perform the vehicle speed estimation method described above. Attached Figure Description

[0028] To more clearly illustrate the technical solutions in the specific embodiments of this application or the prior art, the drawings used in the description of the specific embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of this application. For those skilled in the art, other drawings can be obtained from these drawings without creative effort.

[0029] Figure 1aThis is a schematic diagram of a vehicle stationary on a ramp according to an embodiment of this application; Figure 1b This is a schematic diagram of a vehicle accelerating on a slope according to an embodiment of this application; Figure 1c This is a flowchart of a vehicle speed estimation method provided according to an embodiment of this application; Figure 1d A comparison diagram of vehicle speed estimation effects under different degrees of slippage in an uphill scenario according to an embodiment of this application; Figure 2 This is a flowchart of a reference speed determination method according to an embodiment of this application; Figure 3 A structural block diagram of a vehicle speed estimation device according to an embodiment of this application. Detailed Implementation

[0030] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0031] Understandably, since the longitudinal reference speed of a vehicle cannot be directly and accurately measured by sensors, the following two approaches are commonly used in related technologies for estimation.

[0032] One approach is a kinematic scheme that combines condition identification with inertial measurement unit (IMU) integral correction. This scheme typically uses information such as wheel speeds, IMU acceleration, and steering wheel angle to identify extreme conditions such as vehicle slippage, wheel lock-up, or steering. Under extreme conditions, the longitudinal acceleration integral result from the IMU is used to replace the wheel speed, and finally, a corrected optimal wheel speed is selected as the vehicle's reference speed.

[0033] It should be noted that the longitudinal acceleration mentioned refers to the acceleration signal detected by the Inertial Measurement Unit (IMU) along the vehicle's direction of travel. However, in extreme conditions such as going uphill or downhill, due to the influence of gravity, the longitudinal acceleration measured by the IMU is coupled with a gravitational factor. That is, the longitudinal acceleration measured by the IMU consists of two superimposed parts: one part is the vehicle's actual acceleration, and the other part is the gravitational component caused by the vehicle's pitch angle. Therefore, when the vehicle slips or locks up in uphill or downhill scenarios, if the longitudinal acceleration output by the IMU is still used for integral correction of wheel speed, the gravity deviation caused by the slope will continue to accumulate in the vehicle speed result. The larger the slope angle, the more significant the integral error.

[0034] Furthermore, when all four wheels of the vehicle slip or lock simultaneously, the wheel speed signals of all wheels become meaningless. In slope scenarios, the inertial measurement unit's acceleration is already inaccurate due to gravitational coupling, and the wheel speeds, even after integral correction, also exhibit significant deviations. At this point, the solution lacks any reliable signal as an estimation benchmark, and the overall logic fails directly.

[0035] Another approach is the Kalman vehicle speed-gradient joint estimation scheme, which incorporates a Kalman filter framework. This type of scheme introduces a Kalman filter framework to attempt to separate the gradient gravity component while estimating the vehicle speed. Specifically, this type of scheme has the following two typical implementation methods.

[0036] (1) Two-state variable dual-observation scheme (vehicle speed-gradient joint estimation) The scheme uses wheel speed and longitudinal acceleration as two observation inputs, and sets vehicle speed and slope together as state variables for Kalman estimation. During the filtering iteration process, the gravity component is separated and the reference vehicle speed is output simultaneously.

[0037] However, its limitation lies in the fact that vehicle speed and gradient are highly coupled, each being a prerequisite for the other. That is, accurate gradient estimation depends on reliable vehicle speed observations, while the separation of clean vehicle acceleration depends on accurate gradient results. When four wheels slip on a slope, wheel speed observations become completely inaccurate, directly causing the gradient estimate to deviate from the true value. An incorrect gradient further renders acceleration separation ineffective, which in turn worsens the vehicle speed estimation results, ultimately creating a vicious cycle of "inaccurate vehicle speed → incorrect gradient estimation → even more inaccurate acceleration → even worse vehicle speed," with estimation results completely diverging under extreme conditions.

[0038] (2) Three-Observation Scheme for Predicting Slope This scheme pre-estimates the gradient using vehicle longitudinal dynamics, then uses the estimated gradient as the third observation input for Kalman filtering, ultimately achieving independent estimation of vehicle speed. Theoretically, it has better anti-interference capabilities than the first scheme.

[0039] However, its limitation lies in the fact that it merely moves the coupling process forward, without eliminating the root cause of the coupling. Specifically, the predicted gradient is usually calculated based on the vehicle dynamics model, which itself relies on a reference vehicle speed. Essentially, it simply shifts the speed-gradient coupling relationship from the filtering process to the preprocessing stage. When wheel slippage causes the vehicle speed reference to become inaccurate, the forward gradient estimation will also be biased, and the resulting gradient observation input to the filter remains unreliable, failing to fundamentally solve the failure problem under extreme slope conditions.

[0040] In other words, the Kalman speed-gradient joint estimation schemes in related technologies have not escaped the coupling dependence between speed and gradient. Speed ​​estimation and gradient estimation form a nested closed loop, and inaccuracy in one will directly lead to synchronous deterioration of the other.

[0041] In summary, both of these approaches share a common fundamental flaw: the acquisition of the gradient gravity component always depends on vehicle speed observation, making the two deeply intertwined. Once the wheel speed signal becomes completely inaccurate under extreme slope conditions, the entire estimation system loses its benchmark and continues to accumulate errors, failing to guarantee the accuracy of vehicle speed estimation under extreme conditions.

[0042] Therefore, to address the issue of unpredictable reference vehicle speed under extreme conditions on slopes, this application introduces pitch angular velocity measurement using a 6-axis IMU. It should be noted that a 6-axis IMU can measure acceleration in the X, Y, and Z directions and angular velocity around these three directions; pitch angular velocity is one of the three angular velocities.

[0043] For example, please refer to Figure 1a and Figure 1b . Figure 1a It shows a vehicle stationary on a ramp. Figure 1b This diagram illustrates a vehicle accelerating down a slope. In the diagram, θ represents the vehicle's pitch angle, vx' represents the vehicle's actual longitudinal acceleration relative to the ground, and IMU_ax represents the raw measurement from the IMU longitudinal acceleration sensor (i.e., the raw longitudinal acceleration).

[0044] It is understandable that the IMU longitudinal acceleration sensor can essentially be regarded as a spring-mass system: the greater the slope (the greater the pitch angle θ), or the more intense the vehicle's acceleration, the greater the spring compression, and the greater the value output by the sensor.

[0045] Specifically, when the vehicle is stationary on the slope (e.g.) Figure 1a As shown in the diagram, the vehicle's own acceleration is vx'=0, but the component of gravitational acceleration g in the longitudinal direction of the vehicle, g×sinθ, still exists. This component acts on the sensor and causes it to generate an output. At this time, the IMU's measurement value is entirely contributed by the gravitational component, i.e., IMU_ax=g×sinθ.

[0046] When a vehicle accelerates on a slope (e.g.) Figure 1b As shown in the figure, two accelerations are superimposed on the longitudinal axis of the IMU: one is the actual acceleration of the vehicle, vx', and the other is the longitudinal component of gravity, g×sinθ. Therefore, the measurement value of the IMU is: IMU_ax = g×sinθ + vx'.

[0047] Theoretically, if we consider the vehicle pitch angle θ and the slope angle as the same physical quantity, under ideal conditions (assuming the IMU's pitch angular velocity gyroscope does not experience low-frequency drift), the pitch angle θ can be directly obtained by time integration of the pitch angular velocity output by the IMU. Once an accurate θ is obtained, the gravitational component g × sinθ can be calculated. Subtracting this gravitational component from the original measurement of the original longitudinal acceleration obtained by the IMU yields the true acceleration generated purely by the vehicle's motion, excluding slope interference, i.e., vx' = IMU_ax - g × sinθ.

[0048] Therefore, as long as the current pitch angle θ can be accurately obtained, the real vehicle acceleration can be completely separated from the original longitudinal acceleration measured by the IMU, thus fundamentally decoupling the vehicle speed estimation and the slope estimation.

[0049] Based on this, according to the embodiments of this application, a vehicle speed estimation method embodiment is provided. It should be noted that the steps shown in the flowchart in the accompanying drawings can be executed in a computer system such as a set of computer-executable instructions. Furthermore, although a logical order is shown in the flowchart, in some cases, the steps shown or described may be executed in a different order than that shown here.

[0050] This embodiment provides a vehicle speed estimation method, such as Figure 1c As shown, the method includes the following steps: S110. Obtain estimated reference data for the target vehicle.

[0051] The target vehicle can refer to the motor vehicle whose speed needs to be estimated, such as a vehicle capable of operating in hybrid mode. The estimation reference data can refer to various data required by the controller during the speed estimation process. Specifically, the estimation reference data can include basic measurement data and derived measurement data.

[0052] Among them, basic measurement data can refer to the raw measurement signals directly acquired by sensors, which may include real-time wheel speed data, inertial measurement data, and rotation angle data.

[0053] The real-time wheel speed data refers to the rotational speed of each of the four wheels of the target vehicle. The inertial measurement data refers to the motion data measured by a 6-axis inertial measurement unit, including three raw accelerations and three raw angular velocities. The three raw accelerations are the raw longitudinal acceleration along the vehicle's X-axis (vehicle longitudinal direction), the raw lateral acceleration along the vehicle's Y-axis (vehicle transverse direction), and the raw vertical acceleration along the vehicle's Z-axis (vertical direction). The three raw angular velocities are the raw yaw rate around the Z-axis, the raw pitch rate around the Y-axis, and the raw roll rate around the X-axis. The steering angle data refers to the steering wheel rotation angle of the target vehicle, which can be acquired by a steering wheel angle sensor.

[0054] Derivative measurement data can refer to intermediate parameters derived from basic measurement data through mathematical operations. In other words, it is derived data obtained from basic measurement data through mathematical processing such as statistics, conversion, integration, and differentiation. Examples include wheel speed statistics, front wheel angle, gravity component data, and gravity variation parameters.

[0055] Specifically, the derived measurement data can include gravity component data and gravity variation parameters. Gravity component data refers to the gravitational acceleration component in the longitudinal direction of the target vehicle, that is, the projected component of gravitational acceleration along the longitudinal axis of the vehicle body, generated by the vehicle's pitch attitude, and superimposed on the longitudinal acceleration signal measured by the IMU. Gravity variation parameters refer to the rate of change of the gravitational acceleration component over time, that is, the amount of change of the pitch angle gravity component per unit time, which reflects the rate of change of the vehicle's pitch attitude.

[0056] Specifically, obtaining the estimated reference data for the target vehicle includes: first, obtaining the basic measurement data of the target vehicle; and then performing mathematical transformations based on the basic measurement data to obtain derived measurement data.

[0057] For example, the body controller can read the pulse counts of four wheel speed sensors at fixed sampling intervals (e.g., 10 milliseconds) via a CAN bus or a dedicated sensor interface. After automatic conversion by the circuit, real-time wheel speed data is obtained. Simultaneously, the three-axis acceleration and three-axis angular velocity output in real time from the inertial measurement unit can also be directly read as inertial measurement data. Furthermore, the steering angle value output from the steering wheel angle sensor can also be read simultaneously as steering angle data.

[0058] Furthermore, the above reading operation is performed once at each sampling time to obtain the basic measurement data at that time.

[0059] After obtaining the basic measurement data, the gravity-related parameters (gravity component data and gravity variation parameters) in the derived measurement data can be integrated based on the pitch angular velocity in the inertial measurement data to obtain the vehicle's pitch angle; trigonometric function calculations can be performed based on the vehicle's pitch angle to obtain the gravity component data; and differentiation can be performed based on the gravity component data to obtain the gravity variation parameters.

[0060] Specifically, the pitch angular velocity is first integrated over time to determine the vehicle's pitch angle at the current moment. Then, using trigonometric relationships, the vehicle's pitch angle is converted into a longitudinal gravity component, allowing the calculation of the gravity component data at the current moment. Finally, the gravity component data is differentiated over time to obtain the change per unit time, which serves as the gravity change parameter.

[0061] For example, let the algorithm's running step size be dt, the gravitational acceleration be g, and the pitch angular velocity in the inertial measurement data at time k be ω. m,k The specific calculation process is as follows: First, the pitch angular velocity ω m,k Perform time-by-time integration to obtain the vehicle pitch angle θ at time k. m,k The vehicle pitch angle is then converted into the longitudinal gravity component. Using trigonometric functions, the gravity component at time k is gsinθ. m,k Then, by performing time differentiation on the gravity component data, the gravity change parameters at time k can be obtained. .

[0062] The gravity component data and gravity variation parameters can be independently calculated using only the pitch angular velocity signal of the inertial measurement unit. The calculation process does not depend on the wheel speed signal or the vehicle speed estimation result, and the gravity parameters can be obtained completely decoupled from the vehicle speed. This provides a calculation method for subsequent Kalman filtering that does not depend on the vehicle speed, and fundamentally solves the problem that the original longitudinal acceleration value measured by the IMU under the slope condition cannot distinguish between the real motion acceleration and the gravity component.

[0063] In addition, other derived measurement data can also be derived from the basic measurement data through corresponding mathematical transformations.

[0064] For example, regarding wheel speed statistics, the real-time wheel speed data of the four wheels can be sorted and statistically analyzed to obtain the maximum wheel speed, minimum wheel speed, second minimum wheel speed, and average wheel speed. Regarding wheel acceleration, the real-time wheel speed data of each wheel can be differentiated over time to obtain the wheel acceleration of the corresponding wheel. Regarding front wheel steering angle, the steering wheel angle data can be multiplied by the fixed transmission ratio of the vehicle's steering system to calculate the actual steering angle of the vehicle's front wheels.

[0065] The system adopts a hierarchical data processing architecture that first collects basic measurement data and then generates derived measurement data through mathematical transformation. This architecture not only preserves the true attributes of the original sensor signals but also generates specific parameters required for each computational stage. All derived data are derived from existing vehicle-mounted sensors, eliminating the need for additional hardware acquisition equipment. This architecture can improve the vehicle speed estimation capability under conventional vehicle sensor configurations.

[0066] S120. Based on the basic measurement data, the operating conditions of the target vehicle are judged and the wheel speed is arbitrated to obtain the original reference speed of the target vehicle.

[0067] The original reference speed can refer to a temporary vehicle speed that can represent the longitudinal motion state of the vehicle under different operating conditions, but it may still contain a large error under extreme conditions.

[0068] Specifically, firstly, the current state of the target vehicle is determined by comparing basic measurement data with four preset judgment conditions. Based on this, it is further determined whether the target vehicle is currently in a normal operating condition, a slipping condition, a locked-up condition, or an abnormal yaw condition. Subsequently, a corresponding wheel speed arbitration strategy is matched according to the target vehicle's current operating condition. Following this strategy, real-time wheel speed data and relevant derived measurement data are fused together to obtain a preliminary wheel speed result. Finally, the preliminary wheel speed result is subjected to slope limiting processing to constrain the speed change amplitude per unit time, ultimately obtaining the original reference speed.

[0069] S130. Using the Kalman filter algorithm, the vehicle speed is estimated based on the estimated reference data and the original reference speed to obtain the target reference speed of the target vehicle.

[0070] It should be noted that, theoretically, as long as the current vehicle pitch angle can be accurately obtained, the actual vehicle acceleration can be completely separated from the raw longitudinal acceleration of the IMU. However, in the actual vehicle operating environment, the IMU measurement signal is affected by factors such as temperature drift, vehicle vibration, and random noise. Even if the actual vehicle acceleration is separated through the aforementioned calculations, it still contains a large amount of noise and residual errors. If the vehicle speed is directly calculated by integrating this acceleration, the integration drift and accumulated error will continue to increase over time, causing the results to diverge rapidly and failing to meet the accuracy requirements of engineering applications.

[0071] Therefore, based on the aforementioned kinematic principles, it is necessary to further construct a linear Kalman filter to fuse wheel speed observations, acceleration observations, and gravity change observations from multiple sources, thereby achieving robust estimation of the reference vehicle speed.

[0072] As is understandable, the Kalman filter algorithm is a recursive state estimation algorithm, which executes cyclically through two stages: prediction and update. In the prediction stage, based on the state estimate from the previous time step and the system's kinematic model, the prior state prediction (i.e., the state estimate) is calculated for the current time step. In the update stage, actual observations are introduced, the observation residuals and Kalman gain are calculated, and the state estimate is weighted and corrected to obtain the posterior state estimate for the current time step. Through repeated iterations time-by-time, the estimate is continuously adjusted using observations, ultimately enabling the extraction of the optimal state estimate from the noisy observation signal, which serves as the final state output.

[0073] Specifically, the Kalman filter algorithm is used to estimate the vehicle speed based on the estimated reference data and the original reference speed to obtain the target reference vehicle speed. This can include: first, constructing a state vector and an observation vector based on the estimated reference data and the original reference speed; then, using the state vector and the observation vector to perform state prediction and update processing to iteratively calculate the target reference vehicle speed.

[0074] The state vector is the set of physical quantities that need to be continuously estimated iteratively, while the observation vector is the measurement input vector of the Kalman filter, which can be used to correct and calibrate the state prediction values.

[0075] Specifically, the observation vector can include the original reference speed and gravity variation parameters to simultaneously constrain the system state using wheel speed arbitration results and independently calculated slope dynamic information within a single calculation time. The state vector can include the vehicle's longitudinal acceleration, gravity component data, and gravity variation parameters to achieve joint estimation of vehicle speed, acceleration, slope disturbance, and their changing trends. Specifically, a five-dimensional state vector is first constructed (containing the target reference vehicle speed, vehicle longitudinal acceleration, vehicle longitudinal jerk, gravity component data, and gravity variation parameters). Subsequently, a three-dimensional observation vector is constructed (containing the original reference speed, longitudinal acceleration, and gravity variation parameters).

[0076] For example, suppose the discretization formula of the state equation is as follows: In the formula, X k+1 X represents the state vector at time k+1; F represents the state transition matrix, used to describe the change of state variables from time k to time k+1; X k w represents the state vector at time k; k Let k represent the observation vector at time k, also known as the process noise vector. It follows a zero-mean normal distribution, and its covariance matrix is ​​the Q matrix, i.e., ... N represents the normal distribution; Q represents the process noise covariance matrix, used to quantify the uncertainty of the state transition model.

[0077] Wherein, the state vector at time k It can be as follows: In the formula, The target reference speed; This represents the longitudinal acceleration of the vehicle body, which is the first derivative of the target reference vehicle speed. The jerk represents the longitudinal acceleration of the vehicle body, which is the first derivative of the acceleration. The purpose of introducing this state is to improve the response speed of vehicle speed tracking by using second-order Taylor extrapolation under conditions of rapid acceleration or emergency braking. Represents the gravitational component data, where The pitch angle represents the vehicle body angle, and g represents the acceleration due to gravity. This represents the parameter representing the change in gravity.

[0078] The observation vector at time k As shown below: In the formula, This represents the original reference speed, which is the vehicle speed observation value obtained at time k after arbitration of operating conditions; The raw longitudinal acceleration representing the inertial measurement data, i.e., the longitudinal acceleration measurement value directly acquired by the IMU at time k; This represents the gravity change parameter, specifically the observed rate of change of the gravity component obtained from the pitch angular velocity at time k.

[0079] It's important to note that the gravity change parameter in the observation vector is a value obtained through actual measurement or estimation. It originates from the integration of the pitch angular velocity, trigonometric transformations, and differentiation operations, representing a known observational input. The gravity change parameter in the state vector, however, is the state variable to be estimated within the filter, representing the optimal estimate of the rate of change of the gravity component at the current moment. In the Kalman update step, the two are mapped through the observation matrix. That is, the observation matrix directly maps the gravity change parameter in the state vector to its corresponding position in the observation vector. Then, the observation residual (measured value minus predicted value) is calculated to correct the state estimate. In other words, the gravity change parameter in the observation vector is the measured value, while the gravity change parameter in the state vector is the estimated value. The Kalman filter uses the difference between the two to iteratively optimize the estimation result.

[0080] Next, the state transition matrix and observation matrix are constructed. The state transition matrix maps the state vector from the previous time step to the prior prediction value at the current time step based on the kinematic relationship.

[0081] Specifically, the prediction of vehicle speed can be achieved using a second-order Taylor expansion, meaning the vehicle speed at the next moment (k+1) is equal to the vehicle speed (v) at the current moment (k). k) plus the acceleration at this moment (a) k The product of the algorithm's running step (dt) and half of the jerk at this moment (j) k Multiply by the square of the algorithm's running step ( ).

[0082] The acceleration prediction uses a first-order Taylor expansion, i.e., the acceleration (a) at the next moment. k+1 ) equals the acceleration at this moment (a) k Add the jerk (j) at this moment k Multiply by the algorithm's running step size (dt).

[0083] The prediction of jerk is achieved using a random walk model, i.e., the jerk at the next time step (j) is predicted. k+1 The acceleration (j) at this moment is equal to the jerk at this moment. k The noise from the superposition process is not used for deterministic extrapolation.

[0084] Prediction of gravity component data: gravity component data at the next moment ( ) equals the gravitational component data at this moment ( ) plus the gravity change parameters at this moment ( Multiply by the algorithm's running step size (dt).

[0085] The prediction of gravity change parameters also uses a random walk model, and the gravity change parameters at the next time step ( ) equals the gravitational change parameter at this moment ( (Superposition process noise)

[0086] For example, the state transition matrix (i.e., the F matrix) can be as follows: In the formula, dt represents the algorithm's step size, which is the time interval between two adjacent Kalman filter calculations.

[0087] The discretization formula for the observation equation is as follows: In the formula, H represents the observation matrix, which is used to map the state vector to the observation space; X k c represents the state vector at time k; k Let represent the measurement noise vector at time k, which follows a zero-mean normal distribution and has a covariance matrix of R.

[0088] The observation matrix establishes a linear mapping between the state vector and the observation vector: the first observation (original reference velocity) directly corresponds to the reference vehicle speed in the state vector. The second observation (original longitudinal acceleration) corresponds to the sum of the acceleration and gravity component data in the state vector. The third observation (measured value of gravity change parameter) directly corresponds to the gravity change parameter in the state vector.

[0089] For example, the H matrix, i.e. the observation matrix, can be defined as follows: c k The measurement noise vector, specifically the measurement noise vector, has its covariance matrix R (measurement noise covariance matrix), which can be defined as follows: In the formula, R represents the measurement noise covariance matrix, which is used to quantify the measurement noise of the three observations. It is usually set as a diagonal matrix, with the diagonal elements corresponding to the noise variances of the original reference velocity, the original longitudinal acceleration, and the gravity variation parameters, respectively.

[0090] The standard recursive process of Kalman filtering is then executed. At each computation time step, the prior state prediction and prior covariance matrix are first calculated using the state transition matrix and the posterior state estimate from the previous time step. Then, the Kalman gain matrix is ​​calculated; this gain reflects the relative confidence between the predicted and observed values ​​and is determined by the prediction covariance, the observation matrix, and the measurement noise covariance matrix. Next, the observation vector is subtracted from the predicted observation value calculated using the observation matrix to obtain the observation residual. The prior state prediction is corrected by multiplying the Kalman gain matrix by the observation residual to obtain the posterior state estimate for the current time step. Finally, the posterior covariance matrix is ​​updated for the next time step's recursion. After several iterations, the first component of the state vector (target reference vehicle speed) will converge to a smooth and accurate estimate.

[0091] It is important to note that during rapid acceleration or emergency braking, the IMU longitudinal acceleration measurement (i.e., the raw longitudinal acceleration) will experience a drastic jump, and the acceleration channel residual component in the observation residual will increase significantly. The Kalman gain matrix distributes this residual to all state variables, including the longitudinal acceleration state of the vehicle body. By appropriately setting the corresponding jerk term in the process noise covariance matrix (making its value relatively large to allow for rapid changes in jerk), the jerk is quickly assigned a non-zero value reflecting the rate of change of acceleration.

[0092] In the subsequent prediction step, using the jerk estimate, the vehicle speed prediction employs a second-order Taylor expansion with quadratic terms. This allows for the prediction of acceleration trends before the measurement occurs, pushing the vehicle speed estimate forward and significantly reducing tracking lag under rapid acceleration and deceleration conditions. In contrast, without the jerk state (degenerating into a four-state Kalman calorimeter), acceleration is modeled as a pure random walk, and speed prediction is merely a first-order linear extrapolation. This results in a "catch-lag-catch-up" cycle when acceleration changes drastically, leading to a significantly insufficient response speed.

[0093] By using Kalman filtering design, on the one hand, the slope gravity interference and the actual motion acceleration are decoupled by using an independent pitch angular velocity calculation channel. On the other hand, second-order advance tracking under rapid acceleration and deceleration conditions is realized by using the jerk state. Compared with traditional Kalman filters, the accuracy, smoothness and response speed of vehicle speed estimation are significantly improved under extreme conditions such as four-wheel slippage, wheel lock-up, rapid acceleration and emergency braking on slopes.

[0094] As can be understood from the above Kalman filter design, after constructing the observation vector and state vector, it is necessary to further determine the filtering parameters of the Kalman filter algorithm (i.e., the process noise covariance matrix Q and the measurement noise covariance matrix R in the Kalman filter). Specifically, this can be achieved in the following way: First, the stability of the target vehicle is judged based on the basic measurement data to obtain the vehicle stability level; then, the filtering parameters of the Kalman filter algorithm are determined based on the vehicle stability level; finally, state prediction and update processing are performed based on the filtering parameters, state vector, and observation vector to iteratively calculate the target reference vehicle speed.

[0095] Among them, vehicle stability level can refer to a classification label that characterizes the current degree of instability of the target vehicle, that is, a graded label for vehicle driving stability.

[0096] For example, the vehicle stability level may include a first stable state, a second stable state, and a third stable state. The first stable state may include the target vehicle being in a four-wheel slippage or four-wheel lockup situation, that is, all four wheels simultaneously excessively slip (slippage) or are simultaneously locked by the braking system (lockup). At this time, all wheel speed signals of the vehicle deviate significantly from the actual vehicle speed, and this situation can be regarded as the vehicle being in an extremely unstable state.

[0097] The second stable state includes situations where the target vehicle is not experiencing simultaneous slippage / lockup of all four wheels or abnormal yaw. For example, the target vehicle may be experiencing one of the following: simultaneous slippage of all four wheels, simultaneous lockup of all four wheels, or abnormal yaw. That is, the target vehicle has one or more wheels slipping or locked (but not all four wheels), or the vehicle has undergone unexpected yaw movements such as fishtailing or sideslipping, but has not yet reached the point of simultaneous instability of all four wheels. This situation can be considered as a moderately unstable state.

[0098] The third stable state can include a situation where the target vehicle is in a state that is neither the first nor the second stable state, that is, all wheels are not slipping or locked, and the yaw motion is normal, and the vehicle is in a stable driving state.

[0099] Specifically, the system first checks whether the triggering conditions for the first stable state are met, i.e., whether all four wheels are slipping or locked. This can be determined by comparing the wheel acceleration of each wheel with a preset threshold. If the wheel acceleration of all wheels exceeds the slippage threshold or is below the lockup threshold, the system is considered to be in the first stable state. If the conditions for the first stable state are not met, the system continues to check whether the triggering conditions for the second stable state are met, i.e., whether at least one (but not all) wheel is slipping or locked, or whether the abnormal yaw condition is met. If these conditions are met, the system is considered to be in the second stable state. If neither of these conditions is met, the system is considered to be in the third stable state.

[0100] After obtaining the vehicle stability level, the filtering parameters of the Kalman filter algorithm can be determined based on the vehicle stability level.

[0101] It is understandable that the filtering parameters can refer to the values ​​or adjustment factors of the process noise covariance matrix Q and the measurement noise covariance matrix R in the Kalman filter. They can control the weight distribution between state prediction and observation update. That is, when the vehicle is at different stability levels, the degree of confidence of the filter in the predicted and observed values ​​can be changed by adjusting Q and R.

[0102] Specifically, the process noise covariance matrix Q and the measurement noise covariance matrix R in the Kalman filter can be dynamically adjusted according to the real-time state of the vehicle to improve the adaptability of the estimator.

[0103] Specifically, the adjustment rules can be as follows: First, for the measurement noise covariance matrix R, the diagonal elements corresponding to the original reference speed (i.e., the optimal wheel speed measurement value) are mainly adjusted.

[0104] For example, if the vehicle stability level indicates that the target vehicle is in the first stable state (all four wheels slipping or all four wheels locking up at the same time), then the reliability of its original reference speed is the lowest. Therefore, the R term can be set to a large value (e.g., 100), indicating that the measurement noise is large and the filter should reduce the trust in this measurement value.

[0105] If the vehicle stability level indicates that the target vehicle is in the second stable state (non-four-wheel slippage or lock-up, or the vehicle is in an abnormal yaw condition), then the reliability of the target vehicle's original reference speed is moderate, and the R term can be set to a moderate value (e.g., 20).

[0106] If the vehicle stability level indicates that the target vehicle is in the third stable state (entering this state when neither the conditions of level one nor level two are met), then the original reference speed of the target vehicle has the highest reliability, and the corresponding R term is set to a small value (e.g., 0.1).

[0107] It should be noted that the above values ​​are merely examples. In practical applications, calibration is required based on the specific vehicle's wheel speed sensor noise level, IMU noise level, and algorithm step size. This is typically determined through repeated adjustments using real vehicle data. The more unstable the vehicle, the greater the noise of the original reference speed, the less reliable its measurement value, and the larger the value of the R term.

[0108] Secondly, the process noise covariance matrix Q can be scaled as a whole according to the intensity of the vehicle's lateral movement.

[0109] For example, the lateral motion percentage can be defined to describe the intensity of the vehicle's lateral movement. The formula for calculating this percentage is: Lateral motion percentage = (lateral acceleration squared) / (lateral acceleration squared + longitudinal acceleration squared). The closer this ratio is to 1, the higher the lateral motion percentage, indicating more intense lateral movement. In this case, the modeling error of the longitudinal kinematics model (the state transition equation of the Kalman filter) will increase significantly, meaning increased process noise. Therefore, the original Q-matrix needs to be multiplied by a scaling factor greater than 1 (this scaling factor can be obtained through table lookup or linear interpolation and is positively correlated with the lateral motion percentage). Conversely, when the lateral motion percentage is small (the vehicle mainly moves longitudinally), the process noise is small, and the scaling factor is close to 1 or slightly less than 1.

[0110] The specific values ​​of the scaling factor were also obtained through calibration using real vehicle data. This dynamic adjustment mechanism enables the filter to automatically reduce its reliance on the kinematic model under conditions where lateral motion is dominant, such as vehicle turning and lane changing, thus avoiding estimation bias caused by model mismatch.

[0111] By using dynamically adjusted Q and R matrices, the Kalman filter can adaptively balance the weights between predicted and observed values ​​under different operating conditions, thereby maintaining the robustness and accuracy of vehicle speed estimation across the entire operating range.

[0112] For example, Figure 1d The image shows a comparison of vehicle speed estimation results under different degrees of slippage in an uphill scenario. Multiple colored curves correspond to the real-time wheel speeds of the four wheels. The dashed line represents the vehicle's true reference speed (true value baseline), and the black solid line represents the target reference speed estimated by this solution. The vertical axis is in kilometers per hour, and the horizontal axis is the running time.

[0113] The period from 57.5s to 59s represents a typical four-wheel slippage condition, corresponding to an extremely unstable vehicle state. During this time, the speeds of all four wheels oscillate violently, deviating significantly from the actual vehicle speed, making the reliability of the original reference speed extremely low. This solution adjusts the measurement noise term of the original reference speed to its maximum value to reduce the interference of unreliable wheel speeds on the estimation results. It primarily relies on independently calculated gravity variation parameters and longitudinal acceleration observations to maintain the state estimation. The black estimated vehicle speed curve consistently closely approximates the actual vehicle speed benchmark and does not deviate significantly with slippage wheel speeds.

[0114] From 59s to 61s, the vehicle is in an unstable state, with some wheels still experiencing slippage. However, the wheel speeds of some wheels (represented by the blue line) are relatively close to the actual vehicle speed. This solution adjusts the measurement noise term of the original reference speed to an intermediate value, appropriately accepts wheel speed observations, and combines multi-source information fusion to maintain a smooth estimated vehicle speed curve with a small deviation from the actual vehicle speed.

[0115] After 62.5 seconds, the vehicle enters a stable driving condition with no wheel slippage or lock-up. The vehicle is decelerating, and the speeds of all four wheels follow the trend of the actual vehicle speed. This scheme minimizes the measurement noise of the original reference speed, fully utilizes wheel speed observations, and estimates the vehicle speed curve to closely match the actual vehicle speed, achieving both high estimation accuracy and good smoothness.

[0116] In the above implementation, by introducing independently calculated gravity component data and gravity change parameters, a pitch gravity observation channel completely decoupled from vehicle speed is constructed, solving the estimation failure problem in extreme conditions such as high coupling between vehicle speed and slope and four-wheel slippage on slopes in traditional schemes; through a wheel speed arbitration mechanism for different operating conditions, the optimal wheel speed observation results can be matched under different operating conditions, improving the observation accuracy under normal operating conditions; through a multi-state Kalman filter framework including jerk, both steady-state estimation accuracy and dynamic operating condition tracking response speed are taken into account; the overall scheme can achieve stable and accurate longitudinal reference vehicle speed estimation under all operating conditions with conventional vehicle-mounted sensor configuration, especially improving the robustness of vehicle speed estimation under extreme slope conditions.

[0117] In some implementations, the target vehicle's operating condition is determined and wheel speed is arbitrated based on fundamental measurement data to obtain the target vehicle's original reference speed. Please refer to the appendix. Figure 2 ,include: S210. Based on the basic measurement data, the operating condition of the target vehicle is determined to obtain the operating condition.

[0118] Among them, operating conditions can refer to graded indicators that characterize the current wheel-road adhesion state and vehicle motion state of the target vehicle.

[0119] Specifically, operating conditions can include normal operating conditions, slipping operating conditions, seizing operating conditions, and abnormal yaw operating conditions.

[0120] Normal operating conditions can refer to a stable driving state in which all wheels of the vehicle do not slip or lock up, and the yaw motion of the vehicle body is within the normal range.

[0121] Slippage can refer to a condition where the wheel speed of one or more wheels is significantly higher than the actual speed of the vehicle, that is, the wheels excessively slip relative to the road surface.

[0122] Lock-up conditions can refer to a situation where the wheel speed of one or more wheels is significantly lower than the actual speed of the vehicle or even approaches zero, meaning that the wheels are locked by the braking system and are dragging.

[0123] Abnormal yaw conditions can refer to a situation where the vehicle's rotational angular velocity (yaw rate) around its vertical axis deviates significantly from the theoretical value calculated based on the steering wheel angle and vehicle speed, or the measured lateral acceleration deviates significantly from the theoretical value, indicating that the vehicle may experience unexpected rotational movements such as fishtailing and sideslipping.

[0124] Specifically, the operation condition of the target vehicle is determined based on the basic measurement data. This can include: first, using the basic measurement data and preset conditions to determine the state, and then determining the operation condition based on the state determination result.

[0125] Among them, the status judgment result can refer to the judgment result that reflects the matching relationship between a certain basic measurement data and the judgment conditions of each working condition, and is used to identify whether the current vehicle meets the trigger conditions of the corresponding working condition.

[0126] The preset conditions can refer to the judgment rules and threshold parameters that are pre-set and stored in the controller, which are used to determine whether the vehicle is currently in a specific working condition such as normal, slipping, locking up or abnormal yaw.

[0127] Understandably, these thresholds are typically obtained through real-vehicle calibration and implemented using lookup tables. For example, the preset acceleration threshold for judging slippage and wheel lockup can be determined using a one-dimensional lookup table based on the longitudinal acceleration in the inertial measurement data. The lookup rule is: the greater the longitudinal acceleration, the smaller the corresponding threshold, ensuring that the timing of slippage or wheel lockup judgments under high acceleration conditions such as uphill or downhill is basically consistent with that under flat road conditions. For the threshold for judging abnormal wheel speed differences under normal operating conditions, a two-dimensional lookup table correction can be performed based on the steering wheel angle and the lateral acceleration in the inertial measurement data to account for normal changes in the inner and outer wheel speed differences under turning conditions.

[0128] Specifically, the preset conditions may include a first condition, a second condition, a third condition, and a fourth condition.

[0129] The first condition can be a normal driving condition, such as a preset wheel speed threshold. The second condition can be a slippage condition, such as a preset slippage threshold. The third condition can be a lock-up condition, such as a preset lock-up threshold. The fourth condition can be an abnormal yaw condition, which can be a preset yaw threshold or a preset lateral threshold.

[0130] Specifically, at each calculation moment, the controller first extracts real-time wheel speed data (the rotational speed of each of the four wheels) and longitudinal acceleration, lateral acceleration, yaw rate, and steering angle data (steering wheel rotation angle) from the basic measurement data. Simultaneously, it extracts pre-calculated maximum wheel speed, minimum wheel speed, second-lowest wheel speed, average wheel speed, wheel acceleration of each wheel (obtained by time differentiation of the real-time wheel speed data), and the front wheel steering angle calculated by multiplying the steering wheel angle by the steering ratio from the derived measurement data.

[0131] Subsequently, the controller independently compares the above data with the four preset conditions in turn to determine whether each condition is met, so as to obtain the state judgment result of whether the condition is met.

[0132] For example, for the first condition, the controller can calculate the difference between the maximum and minimum wheel speeds and compare this difference with a preset wheel speed threshold. It should be noted that this preset wheel speed threshold is not a fixed value, but a dynamic threshold obtained by looking up and correcting a two-dimensional mapping table based on the current steering wheel angle and lateral acceleration in the inertial measurement data. If the difference between the maximum and minimum wheel speeds is less than this dynamic threshold, a state judgment result satisfying the first condition can be output; otherwise, a state judgment result not satisfying the first condition can be output.

[0133] It should be noted that the above-mentioned two-dimensional mapping table can be established as follows: First, using steering wheel angle and lateral acceleration as two input dimensions, the reasonable difference between the maximum and minimum wheel speeds under normal driving conditions during different turning scenarios is measured in advance during actual vehicle calibration. This difference is then stored in the table as a baseline threshold. The rows of the table correspond to discrete values ​​of steering wheel angle (e.g., 0°, 30°, 60°, 90°, 120°, 150°, 180°), and the columns correspond to discrete values ​​of lateral acceleration (e.g., 0 m / s², 2 m / s², 4 m / s², 6 m / s², 8 m / s²). When looking up the table, the controller takes the steering wheel angle and lateral acceleration at the current moment, finds the four nearest grid points in the table, and calculates the preset wheel speed threshold corresponding to the current moment through bilinear interpolation. If the actual steering wheel angle or lateral acceleration exceeds the range of the table, the threshold corresponding to the boundary value is taken.

[0134] Regarding the second condition, the controller can iterate through the wheel accelerations of all four wheels and compare the wheel acceleration of each wheel with a preset slippage threshold. This threshold is also a dynamic threshold, determined by looking up a one-dimensional mapping table based on the longitudinal acceleration in the current inertial measurement data. Understandably, the creation and lookup method of this one-dimensional mapping table is the same as that of the two-dimensional mapping table described above.

[0135] If the wheel acceleration of any wheel is greater than the preset slip threshold, the state judgment result that satisfies the second condition is output; otherwise, the state judgment result that does not satisfy the second condition is output.

[0136] Regarding the third condition, the controller can also iterate through the wheel accelerations of all four wheels, comparing the wheel acceleration of each wheel with a preset lock-up threshold (usually a negative value, indicating deceleration). This can also be determined by looking up a one-dimensional mapping table based on the longitudinal acceleration.

[0137] If the wheel acceleration of any wheel is less than the preset lock-up threshold, the state judgment result that satisfies the third condition is output; otherwise, the state judgment result that does not satisfy the third condition is output.

[0138] Regarding the fourth condition, the controller can simultaneously judge two sub-conditions. If either condition is satisfied, the controller outputs a state judgment result that satisfies the fourth condition; otherwise, it outputs a state judgment result that does not satisfy the fourth condition.

[0139] Specifically, the first sub-condition is to determine whether the absolute value of the difference between the target vehicle's current yaw rate (from inertial measurement data) and the nominal yaw rate is greater than a preset yaw threshold.

[0140] The nominal yaw rate can be calculated using the previous reference vehicle speed (i.e., the target reference speed output at the previous calculation time), the current front wheel steering angle, and the vehicle wheelbase, according to the following formula: In the formula, Represents the nominal yaw rate; This represents the previous reference speed; This represents the front wheel steering angle (calculated from the steering wheel angle via the steering ratio). Represents the vehicle's wheelbase; The stability factor representing the vehicle can be considered a known quantity.

[0141] The second sub-condition is whether the absolute value of the difference between the target vehicle's current lateral acceleration (from inertial measurement data) and the theoretical lateral acceleration (obtained by multiplying the previous reference vehicle speed and yaw rate) is greater than a preset lateral threshold.

[0142] After obtaining the state judgment results, the operating conditions can be further determined based on the state judgment results using the following methods: if the state judgment results indicate that the real-time wheel speed data meets the first condition, the operating condition is determined to be the normal operating condition; if the state judgment results indicate that the inertial measurement data meets the second condition, the operating condition is determined to be the slipping operating condition; if the state judgment results indicate that the inertial measurement data meets the third condition, the operating condition is determined to be the locking operating condition; if the state judgment results indicate that the inertial measurement data meets the fourth condition, the operating condition is determined to be the abnormal yaw operating condition.

[0143] It should be noted that the normal operating condition requires all wheels to be free from slippage and lockup, and that yaw motion be normal. This condition is the most stringent, requiring the first condition to be met simultaneously while the second, third, and fourth conditions are not met. Slippage, lockup, and abnormal yaw are considered abnormal conditions, which more directly and severely interfere with vehicle speed estimation, and therefore have a higher priority in the judgment.

[0144] In other words, in the comprehensive determination of the state judgment results, the following priority order can be adopted: if the second condition is met, it is directly determined to be a slipping condition, and other conditions are not checked. Similarly, if the third condition is met, it is directly determined to be a locking condition. If the fourth condition is met, it is directly determined to be an abnormal yaw condition.

[0145] However, if the first condition is met and the second, third, and fourth conditions mentioned above are not met, then the operating condition will be determined as the normal operating condition.

[0146] By subdividing the operating conditions into four states—normal, slippage, lock-up, and abnormal yaw—and defining clear judgment conditions for each state based on real-time wheel speed data and inertial measurement data, the system can accurately identify the current adhesion and stability state of the vehicle under different operating conditions, thereby providing a reliable decision-making basis for subsequent wheel speed arbitration strategy selection.

[0147] The system employs a method of comparing basic measurement data with preset conditions one by one. Its computational logic is simple and highly real-time, enabling rapid identification of operating conditions within each algorithm step. Simultaneously, the judgment threshold is dynamically adjusted using a lookup table, adapting to different vehicle speeds, slopes, and steering angles, effectively improving the accuracy and scenario adaptability of the judgment results.

[0148] S220: Based on operating conditions and basic measurement data, wheel speed arbitration calculation is performed to obtain the original reference speed of the target vehicle.

[0149] It should be noted that the reliability and representativeness of wheel speeds vary significantly depending on the vehicle's stability level. For example, under normal driving conditions, all wheel speeds are valuable for reference. However, under slippage conditions, the wheel speed of the slipping wheel is much higher than the actual speed, so lower wheel speeds can be prioritized. Under lock-up conditions, the wheel speed of the locked wheel is much lower than the actual speed, so higher wheel speeds can be prioritized. Under abnormal yaw conditions, the wheel speeds may deviate from their actual values ​​due to sideslip; multi-wheel weighted fusion can be used to reduce fluctuations. Therefore, it is essential to select an appropriate wheel speed arbitration strategy based on the operating conditions to obtain reliable original reference speeds.

[0150] Specifically, wheel speed arbitration can be calculated using the following method: First, determine the wheel speed arbitration strategy for the target vehicle based on the operating conditions; then, calculate the original reference speed based on the wheel speed arbitration strategy and basic measurement data.

[0151] Among them, the wheel speed arbitration strategy can refer to the calculation rules that are predefined for different operating conditions and can determine the optimal temporary speed value.

[0152] Specifically, wheel speed arbitration strategies can include weighted strategies and extreme value strategies. A weighted strategy refers to a method of obtaining a temporary speed value by fusing multiple sets of wheel speeds according to their corresponding weights, while an extreme value strategy refers to a method of directly selecting the maximum or minimum value among the four wheel speeds as the output temporary speed value.

[0153] For example, determining the wheel speed arbitration strategy for the target vehicle based on the operating conditions may include: if the operating conditions are normal or abnormal yaw conditions, then the weighted strategy is used as the wheel speed arbitration strategy; if the operating conditions are slipping or lock-up conditions, then the extreme value strategy is used as the wheel speed arbitration strategy.

[0154] It should be noted that under normal operating conditions, since all wheel speeds are relatively reliable, there is a speed difference between the inner and outer wheels when turning. Directly taking the extreme value will introduce deviation. Therefore, a weighted strategy can be used to integrate the information of multiple wheel speeds, and the influence of excessively high outer wheel speed when turning can be suppressed by weighting coefficients. In other words, the weighted strategy is used as the wheel speed arbitration strategy to calculate the original reference speed.

[0155] Specifically, the second smallest wheel speed and the combined wheel speed can be weighted to determine the original reference speed.

[0156] The fused wheel speed refers to the comprehensive wheel speed value obtained by weighted averaging of the four real-time wheel speeds based on the credibility coefficients of each wheel. It is used to characterize the weighted fusion result of the full wheel speed information. The credibility coefficient of each wheel can be calculated first, and then the fused wheel speed can be obtained through weighted averaging.

[0157] For example, the confidence coefficient for each wheel is first calculated. For the i-th wheel (i=1,2,3,4), the controller receives two inputs: the first input is the absolute value of the difference between the real-time wheel speed of that wheel (i.e., the wheel rotation speed value read from the wheel speed sensor at the current moment) and the previous reference vehicle speed. The second input is the absolute value of the difference between the wheel acceleration of that wheel (obtained by taking the time derivative of the real-time wheel speed of that wheel) and the derivative of the target reference vehicle speed at the previous moment.

[0158] The absolute values ​​of these two differences are then used as input to a two-dimensional lookup table, which yields a confidence coefficient bi between 0 and 1. The lookup rules are as follows: the smaller the absolute value of the two differences, the closer the wheel speed is to the vehicle speed and its trend at the previous moment, and the higher the confidence level, with bi closer to 1; conversely, the larger the absolute value of the two differences, the more suspicious the wheel speed, with bi closer to 0.

[0159] Next, the fused wheel speed is calculated. The fused wheel speed is the weighted average of the wheel speeds of the four wheels multiplied by their respective confidence coefficients. The calculation formula is as follows: In the formula, Represents the combined wheel speed; i represents the wheel number, which can be any value from 1, 2, 3, or 4; The credibility coefficient represents wheel i; This represents the wheel speed of wheel i.

[0160] Then, the second smallest wheel speed is selected from the four wheel speeds, which is the second value after sorting the four wheel speed values ​​from smallest to largest. Compared with the smallest wheel speed, the second smallest wheel speed can eliminate the influence of a single abnormally low value, and compared with the largest wheel speed, it can eliminate the influence of a single abnormally high value, making it a robust statistic.

[0161] Next, the weighting coefficient f for the second smallest wheel speed is determined. This coefficient can be obtained by looking up a two-dimensional table based on the lateral acceleration and steering wheel angle from the inertial measurement data at the current moment.

[0162] Understandably, a larger lateral acceleration or steering wheel angle indicates a sharper turn, resulting in a greater natural difference in wheel speed between the inner and outer wheels. This reduces the representativeness of the second smallest wheel speed, thus a smaller value for f increases the weight (1-f) of the combined wheel speed, thereby suppressing the deviation caused by relying solely on the second smallest wheel speed during cornering. Conversely, when driving straight or making gentle turns, a value closer to 1 indicates greater reliance on the second smallest wheel speed.

[0163] Finally, the original reference speed can be calculated using the following formula: In the formula, c1 represents the normal operating condition; The original reference speed representing normal operating conditions; Represents the second smallest wheel speed; f represents the weighting coefficient; This represents the speed of the wheel.

[0164] Under abnormal yaw conditions, the speed of a single wheel is easily affected by lateral slippage and deviates. By merging multiple wheel speeds proportionally and weighted, the impact of single wheel anomalies on the overall results can be reduced, and the stability of the observation results can be improved. In other words, the weighting strategy can also be used as a wheel speed arbitration strategy to calculate the original reference speed.

[0165] For example, first, the second smallest wheel speed, the combined wheel speed (calculated in the same way as under normal operating conditions), and the average wheel speed of the four wheel speeds are calculated. Then, they are weighted and combined according to a fixed ratio. For example, the second smallest wheel speed has the largest proportion (e.g., 50%), the combined wheel speed is the second largest (e.g., 30%), and the average wheel speed has the smallest proportion (e.g., 20%). The specific proportions can be determined based on the actual vehicle calibration. This weighted fusion method can integrate multiple wheel speed statistics and reduce the impact of abnormal fluctuations in individual wheel speeds caused by sideslip on the arbitration results.

[0166] Furthermore, under slipping conditions, the wheel speed of the slipping wheel will be higher than the actual vehicle speed, while the wheel speed of the non-slipping wheel will be relatively lower and closer to the actual vehicle speed. Therefore, the minimum wheel speed can be selected to avoid the artificially high deviation of the slipping wheel speed to the greatest extent. That is, the extreme value strategy can be used as the wheel speed arbitration strategy to calculate the original reference speed.

[0167] For example, if the operating condition is a slippage condition, the extreme value strategy is executed as follows: iterate through the real-time wheel speed data of the four wheels, and directly select the minimum value as the initial wheel speed result to avoid interference from the slippage wheel speed being too high.

[0168] In the case of wheel lockup, the wheel speed of the locked wheel will be lower than the actual vehicle speed, while the wheel speed of the non-locked wheel will be relatively higher and closer to the actual vehicle speed. Therefore, the maximum wheel speed can be selected to avoid the low deviation of the locked wheel speed to the greatest extent. In other words, the extreme value strategy can also be used as the wheel speed arbitration strategy to calculate the original reference speed.

[0169] For example, if the operating condition is a lock-up condition, the extreme value strategy is executed as follows: iterate through the real-time wheel speed data of the four wheels, and directly select the maximum value as the initial wheel speed result to avoid interference from the low wheel speed caused by lock-up.

[0170] For different stability levels, corresponding arbitration strategies are matched. Under normal and abnormal yaw conditions, weighted fusion is used to balance estimation accuracy and result stability. Under slippage and lock-up conditions, extreme value selection is used to quickly avoid abnormal wheel speed interference. This ensures that the original reference speed maintains relatively optimal reliability and representativeness under various conditions.

[0171] Furthermore, regardless of whether the result comes from a weighted strategy or an extreme value strategy, after obtaining the initial wheel speed result, a slope constraint can be applied to it to obtain the final original reference speed.

[0172] Specifically, the upper limit of the slope can be taken as the vehicle's maximum acceleration capability on a high-friction surface. The lower limit of the descent slope can be taken as the maximum deceleration capacity of the vehicle on a high-friction surface. For example, maximum acceleration capability Positive values ​​are acceptable; this represents the maximum deceleration capability. Negative values ​​are acceptable. Clamping operations are used to suppress wheel speed jumps during operating condition changes, thus obtaining the original reference speed.

[0173] For example, the slope constraint can be calculated as follows: In the formula, This represents the original reference velocity output at the current time k after slope limitation; This represents the preliminary wheel speed result obtained through arbitration at the current moment; represents the original reference velocity output after slope limiting at the previous moment; t represents the algorithm's running step size; clamp() represents the clamping function.

[0174] Among them, the clamping function can... (Difference) is limited to the lower limit and upper limit The value is determined by the following criteria: if the difference is less than the lower limit, the lower limit value is used; if it is greater than the upper limit, the upper limit value is used; if the difference is between the lower and upper limits, the difference itself is used.

[0175] By limiting the slope as described above, it can be ensured that the rate of change of the output speed is always within the range of the vehicle's physical capabilities.

[0176] By classifying wheel speed arbitration strategies into two main types—weighted strategies and extreme value strategies—the logic becomes simpler, facilitating engineering implementation and parameter tuning. Furthermore, it achieves optimal filtering and fusion of wheel speed signals under different attachment states, providing high-quality speed observation input for subsequent Kalman filtering.

[0177] In the above implementation, through a two-level processing architecture of working condition identification and wheel speed arbitration, the operating conditions of the driving state are first classified based on multi-source sensor signals, and then the corresponding rules are matched to output reliable wheel speed observation values. This can effectively filter wheel speed interference under abnormal working conditions such as slippage, lock-up, and yaw instability, and provide vehicle speed observation input with controllable quality for subsequent Kalman filtering, thereby improving the robustness and accuracy of the final vehicle speed estimation result from the observation source.

[0178] It should be understood that although the steps in the flowchart above are shown sequentially as indicated by the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowchart above may include multiple steps or stages, which are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages in other steps.

[0179] This specification also provides a vehicle speed estimation device 300, such as... Figure 3 As shown, it includes: a data acquisition module 310, a working condition judgment module 320, and a speed estimation module 330, wherein: The data acquisition module 310 is used to acquire estimated reference data of the target vehicle; wherein, the estimated reference data includes basic measurement data and derived measurement data; the derived measurement data includes gravity component data and gravity change parameters; the gravity component data refers to the gravitational acceleration component in the longitudinal direction of the target vehicle; the gravity change parameters refer to the rate of change of the gravitational acceleration component over time.

[0180] The operating condition judgment module 320 is used to judge the operating condition of the target vehicle and arbitrate the wheel speed based on the basic measurement data in order to obtain the original reference speed of the target vehicle.

[0181] The speed estimation module 330 is used to estimate the vehicle speed based on the estimated reference data and the original reference speed using the Kalman filter algorithm, so as to obtain the target reference vehicle speed of the target vehicle; wherein, the observation vector of the Kalman filter algorithm includes the original reference speed and the gravity change parameters.

[0182] In some implementations, the speed estimation module 330 is also used to construct a state vector and an observation vector based on the estimated reference data and the original reference speed; wherein, the state vector includes the longitudinal acceleration of the vehicle body, gravity component data and gravity change parameters; and the state vector and observation vector are used to perform state prediction and update processing to iteratively calculate the target reference vehicle speed.

[0183] In some embodiments, the vehicle speed estimation device 300 further includes a wheel speed determination module, which is used to determine the operating condition of the target vehicle based on basic measurement data to obtain the operating condition; wherein, the operating condition includes normal operating condition, slipping operating condition, lock-up operating condition and abnormal yaw operating condition; and to perform wheel speed arbitration calculation based on the operating condition and basic measurement data to obtain the original reference speed of the target vehicle.

[0184] In some embodiments, the vehicle speed estimation device 300 further includes a stability judgment module, which is used to make a state judgment using basic measurement data and preset conditions to obtain a state judgment result; wherein, the basic measurement data includes real-time wheel speed data and inertial measurement data; and the operating conditions are determined based on the state judgment result.

[0185] In some implementations, the preset conditions include a first condition, a second condition, a third condition, and a fourth condition; the stability judgment module is further configured to determine the operating condition as a normal operating condition if the state judgment result indicates that the real-time wheel speed data meets the first condition; determine the operating condition as a slipping operating condition if the state judgment result indicates that the inertial measurement data meets the second condition; determine the operating condition as a lock-up operating condition if the state judgment result indicates that the inertial measurement data meets the third condition; and determine the operating condition as an abnormal yaw operating condition if the state judgment result indicates that the inertial measurement data meets the fourth condition.

[0186] In some implementations, the wheel speed determination module is also used to determine the wheel speed arbitration strategy of the target vehicle based on the operating conditions; wherein the wheel speed arbitration strategy includes a weighted strategy and an extreme value strategy; and the original reference speed is calculated based on the wheel speed arbitration strategy and the basic measurement data.

[0187] In some embodiments, the vehicle speed estimation device 300 further includes a wheel speed arbitration module, which is used to use a weighted strategy as the wheel speed arbitration strategy if the operating condition is a normal operating condition or an abnormal yaw condition; and to use an extreme value strategy as the wheel speed arbitration strategy if the operating condition is a slipping condition or a lock-up condition.

[0188] In some implementations, the speed estimation module 330 is further used to determine the stability of the target vehicle based on basic measurement data to obtain a vehicle stability level; wherein, the vehicle stability level includes a first stable state, a second stable state, and a third stable state; the first stable state includes the target vehicle being in a four-wheel slippage or four-wheel lockup situation; the second stable state includes the target vehicle being in a non-four-wheel slippage / lockup situation or an abnormal yaw situation; the filtering parameters of the Kalman filter algorithm are determined based on the vehicle stability level; state prediction and update processing are performed based on the filtering parameters, state vector, and observation vector to iteratively calculate the target reference vehicle speed.

[0189] In some implementations, the data acquisition module 310 is also used to acquire basic measurement data of the target vehicle; and to perform mathematical transformation processing based on the basic measurement data to obtain derived measurement data.

[0190] In some implementations, the basic measurement data includes real-time wheel speed data and inertial measurement data. The data acquisition module 310 is also used to perform integral processing based on the pitch angular velocity in the inertial measurement data to obtain the vehicle body pitch angle; perform trigonometric function calculation processing based on the vehicle body pitch angle to obtain gravity component data; and perform differentiation based on the gravity component data to obtain gravity change parameters.

[0191] For specific limitations regarding a vehicle speed estimation device, please refer to the limitations of a vehicle speed estimation method described above, which will not be repeated here. Each module in the aforementioned vehicle speed estimation device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device in hardware form, or stored in the memory of a computer device in software form, so that the processor can call and execute the corresponding operations of each module.

[0192] In this embodiment, a vehicle speed estimation device is presented in the form of a functional unit. Here, a unit refers to an ASIC (Application Specific Integrated Circuit) circuit, a processor and memory that execute one or more software or fixed programs, and / or other devices that can provide the above-mentioned functions.

[0193] This specification also provides a hybrid vehicle, including a memory and a processor. The memory includes instructions loaded and executed by the processor, which, when executed, cause the processor to perform the vehicle speed estimation method described above. Specific limitations regarding the hybrid vehicle can be found in the limitations of the vehicle speed estimation method described above, and will not be repeated here. The apparatus, module, or unit illustrated in the above embodiments can be implemented by a computer chip or entity, or by a product with a certain function. A typical implementation device is a computer. Specifically, the computer can be, for example, a personal computer, laptop computer, cellular phone, camera phone, smartphone, personal digital assistant, media player, navigation device, email device, game console, tablet computer, wearable device, or any combination of these devices.

[0194] For ease of description, the above devices are described separately by function as various units. Of course, in implementing this application, the functions of each unit can be implemented in one or more software and / or hardware.

[0195] Those skilled in the art will understand that embodiments of this application can be provided as methods, apparatus, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0196] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus, and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in one or more flowchart illustrations and / or one or more block diagrams.

[0197] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means that implement the functions specified in one or more flowcharts and / or one or more block diagrams.

[0198] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer-implemented process, such that the instructions, which execute on the computer or other programmable apparatus, provide steps for implementing the functions specified in one or more flowcharts and / or one or more block diagrams.

[0199] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes said element.

[0200] The various embodiments in this specification are described in a progressive manner. Similar or identical parts between embodiments can be referred to mutually. Each embodiment focuses on describing the differences from other embodiments. In particular, the apparatus embodiments are basically similar to the method embodiments, so the description is relatively simple; relevant parts can be referred to the descriptions of the method embodiments.

[0201] The above description is merely an embodiment of this application and is not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

[0202] Although embodiments of this application have been described in conjunction with the accompanying drawings, those skilled in the art can make various modifications and variations without departing from the spirit and scope of this application, and such modifications and variations all fall within the scope defined by the appended claims.

Claims

1. A method for estimating vehicle speed, characterized in that, The method includes: Obtain estimated reference data for the target vehicle; wherein, the estimated reference data includes basic measurement data and derived measurement data; the derived measurement data includes gravity component data and gravity variation parameters; the gravity component data refers to the gravitational acceleration component in the longitudinal direction of the target vehicle; the gravity variation parameters refer to the rate of change of the gravitational acceleration component over time; Based on the aforementioned basic measurement data, the operating conditions of the target vehicle are determined and wheel speed is arbitrated to obtain the original reference speed of the target vehicle. Using the Kalman filter algorithm, vehicle speed is estimated based on the estimated reference data and the original reference speed to obtain the target reference speed of the target vehicle; wherein, the observation vector of the Kalman filter algorithm includes the original reference speed and the gravity change parameter.

2. The method according to claim 1, characterized in that, The step of using the Kalman filter algorithm to estimate the vehicle speed based on the estimated reference data and the original reference speed to obtain the target reference vehicle speed includes: A state vector and an observation vector are constructed based on the estimated reference data and the original reference velocity; wherein, the state vector includes jerk, gravity component data, and gravity variation parameters; The state vector and the observation vector are used for state prediction and update processing to iteratively calculate the target reference vehicle speed.

3. The method according to claim 1, characterized in that, The step of determining the operating condition and arbitrating wheel speed of the target vehicle based on the basic measurement data to obtain the original reference speed of the target vehicle includes: Based on the aforementioned basic measurement data, the operating condition of the target vehicle is determined to obtain the operating condition; wherein, the operating condition includes normal operating condition, slipping operating condition, lock-up operating condition, and abnormal yaw operating condition; Wheel speed arbitration calculation is performed based on the operating conditions and the basic measurement data to obtain the original reference speed of the target vehicle.

4. The method according to claim 3, characterized in that, The step of determining the operating condition of the target vehicle based on the basic measurement data to obtain the operating condition includes: The state is determined by using the basic measurement data and preset conditions to obtain the state determination result; wherein, the basic measurement data includes real-time wheel speed data and inertial measurement data; The operating conditions are determined based on the status judgment results.

5. The method according to claim 4, characterized in that, The preset conditions include a first condition, a second condition, a third condition, and a fourth condition; determining the operating condition based on the state judgment result includes: If the status judgment result indicates that the real-time wheel speed data meets the first condition, the operating condition is determined as the normal operating condition. If the state judgment result indicates that the inertial measurement data meets the second condition, the operating condition is determined to be the slipping condition; If the state judgment result indicates that the inertial measurement data meets the third condition, the operating condition is determined as the locking condition. If the state judgment result indicates that the inertial measurement data meets the fourth condition, the operating condition is determined to be the abnormal yaw condition.

6. The method according to claim 2, characterized in that, The step of performing wheel speed arbitration calculation based on the operating conditions and the basic measurement data to obtain the original reference speed of the target vehicle includes: The wheel speed arbitration strategy for the target vehicle is determined based on the operating conditions; wherein the wheel speed arbitration strategy includes a weighted strategy and an extreme value strategy. The original reference speed is calculated based on the wheel speed arbitration strategy and the basic measurement data.

7. The method according to claim 6, characterized in that, The wheel speed arbitration strategy for determining the target vehicle based on the operating conditions includes: If the operating condition is a normal operating condition or an abnormal yaw condition, then the weighting strategy will be used as the wheel speed arbitration strategy. If the operating condition is a slipping condition or a locking condition, then the extreme value strategy will be used as the wheel speed arbitration strategy.

8. The method according to claim 2, characterized in that, The process of using the state vector and the observation vector to perform state prediction and update, and iteratively calculating the target reference vehicle speed, includes: Based on the aforementioned basic measurement data, the stability of the target vehicle is assessed to obtain a vehicle stability level; wherein, the vehicle stability level includes a first stable state, a second stable state, and a third stable state; the first stable state includes the target vehicle being in a situation where all four wheels are slipping or all four wheels are locked; the second stable state includes the target vehicle being in a situation where not all four wheels are slipping / locking simultaneously or in a situation where the vehicle is experiencing abnormal yaw. The filtering parameters of the Kalman filter algorithm are determined based on the vehicle stability level. Based on the filtering parameters, the state vector, and the observation vector, state prediction and update processing are performed to iteratively calculate the target reference vehicle speed.

9. The method according to claim 1, characterized in that, The acquisition of the estimated reference data for the target vehicle includes: Obtain basic measurement data of the target vehicle; The derived measurement data is obtained by performing mathematical transformations on the basic measurement data.

10. The method according to claim 9, characterized in that, The basic measurement data includes real-time wheel speed data and inertial measurement data; the mathematical transformation processing based on the basic measurement data to obtain the derived measurement data includes: The pitch angular velocity in the inertial measurement data is integrated to obtain the vehicle body pitch angle of the target vehicle. The gravity component data is obtained by performing trigonometric function calculations based on the vehicle pitch angle. The gravity change parameters are obtained by differentiating the gravity component data.

11. A vehicle speed estimation device, characterized in that, The device includes: The data acquisition module is used to acquire estimated reference data of the target vehicle; wherein, the estimated reference data includes basic measurement data and derived measurement data; the derived measurement data includes gravity component data and gravity variation parameters; the gravity component data refers to the gravitational acceleration component in the longitudinal direction of the target vehicle; the gravity variation parameters refer to the rate of change of the gravitational acceleration component over time; The operating condition judgment module is used to judge the operating condition and arbitrate the wheel speed of the target vehicle based on the basic measurement data, so as to obtain the original reference speed of the target vehicle. The speed estimation module is used to estimate the vehicle speed based on the estimated reference data and the original reference speed using the Kalman filter algorithm, so as to obtain the target reference vehicle speed; wherein, the observation vector of the Kalman filter algorithm includes the original reference speed and the gravity change parameters.

12. A hybrid vehicle, characterized in that, The device includes a memory and a processor, the memory including instructions loaded and executed by the processor, the instructions causing the processor to perform the vehicle speed estimation method according to any one of claims 1-9 when executed.

Images