CN-121989968-A - Longitudinal vehicle speed estimation method and system based on multi-sensor fusion

CN121989968ACN 121989968 ACN121989968 ACN 121989968ACN-121989968-A

Abstract

The invention provides a longitudinal vehicle speed estimation method and a longitudinal vehicle speed estimation system with multi-sensor fusion, which relate to the technical field of active safety control of vehicle engineering, wherein the method comprises the steps of acquiring vehicle sensor data, including driving steering related sensing data and motion state reference sensing data; the method comprises the steps of establishing a three-degree-of-freedom vehicle dynamics model, constructing a first extended Kalman filter estimator which calculates a first longitudinal vehicle speed estimated value and a first posterior covariance, constructing a second extended Kalman filter estimator which calculates a second longitudinal vehicle speed estimated value and a second posterior covariance based on a vehicle longitudinal kinematics relation and motion state reference sensing data, calculating a normalized innovation square value based on an innovation sequence of the first extended Kalman filter estimator and comparing the normalized innovation square value with a check threshold value, adaptively adjusting the first posterior covariance according to a comparison result to obtain an updated first posterior covariance, combining the second posterior covariance to weight and fuse the two vehicle speed estimated values, and outputting the vehicle longitudinal vehicle speed estimated value.

Inventors

ZUO ZHENGXING
AN NA
JIA BORU
LI GUANFU
WEI YIDI
LI JIAN
SUN XIAOHAN

Assignees

北京理工大学

Dates

Publication Date: 20260508
Application Date: 20260302

Claims (10)

1. A longitudinal vehicle speed estimation method for multi-sensor fusion, comprising: S1, acquiring vehicle sensor data, wherein the vehicle sensor data comprises driving steering related sensing data and motion state reference sensing data; s2, establishing a three-degree-of-freedom vehicle dynamics model; s3, constructing a first extended Kalman filter estimator through the three-degree-of-freedom vehicle dynamics model based on the driving steering related sensing data; S4, calculating a first longitudinal vehicle speed estimated value and a first posterior covariance through the first extended Kalman filter estimator; s5, constructing a second extended Kalman filter estimator based on the longitudinal kinematic relation of the vehicle and the motion state reference sensing data; s6, calculating a second longitudinal vehicle speed estimated value and a second posterior covariance through the second extended Kalman filter estimator; S7, calculating a normalized innovation square value based on the innovation sequence generated by the first extended Kalman filter estimator; S8, comparing the normalized innovation square value with a detection threshold value; s9, carrying out self-adaptive adjustment on the first posterior covariance according to a comparison result to obtain an updated first posterior covariance; And S10, combining the updated first posterior covariance and the updated second posterior covariance, carrying out weighted fusion on the first longitudinal vehicle speed estimated value and the second longitudinal vehicle speed estimated value, and outputting a vehicle longitudinal vehicle speed estimated value.
2. The multi-sensor fusion longitudinal vehicle speed estimation method of claim 1, wherein the drive steering related sensing data specifically includes wheel state data and wheel steering data, and the motion state reference sensing data specifically includes inertial measurement data and satellite positioning data.
3. The longitudinal vehicle speed estimation method of multi-sensor fusion according to claim 1, wherein S2 specifically comprises: s201, defining modeling assumption conditions of the three-degree-of-freedom vehicle dynamics model; s202, establishing a coordinate system of the three-degree-of-freedom vehicle dynamics model; s203, defining a system state quantity of the three-degree-of-freedom vehicle dynamics model based on the modeling assumption condition and the coordinate system; s204, according to Newton' S law of motion, combining the driving steering related sensing data and the system state quantity to construct the three-degree-of-freedom vehicle dynamics model.
4. A longitudinal vehicle speed estimation method according to claim 3, characterized in that the coordinate system comprises in particular a geodetic coordinate system, a vehicle coordinate system and a tire coordinate system.
5. The method for estimating a longitudinal vehicle speed by multi-sensor fusion according to claim 1, wherein S3 specifically comprises: s301, discretizing a state equation in the three-degree-of-freedom vehicle dynamics model through a forward Euler method; s302, defining an observation equation of the first extended Kalman filter estimator; s303, respectively linearizing the discretized state equation and the discretized observation equation; s304, constructing the first extended Kalman filter estimator based on the state equation and the observation equation after linearization processing.
6. The method for estimating a longitudinal vehicle speed by multi-sensor fusion according to claim 1, wherein S5 specifically comprises: S501, constructing a vehicle kinematic model based on the vehicle longitudinal kinematic relationship; S502, constructing the second extended Kalman filter estimator based on the vehicle kinematic model and the kinematic state reference sensing data.
7. The method for estimating a longitudinal vehicle speed by multi-sensor fusion according to claim 1, wherein S7 specifically comprises: S701, calculating an innovation item at each moment based on the first extended Kalman filter estimator to obtain the innovation sequence; S702, calculating the normalized innovation square value based on the innovation sequence.
8. The method for estimating a longitudinal vehicle speed by multi-sensor fusion according to claim 1, wherein S9 specifically comprises: S901, defining a reliability index of wheel speed observation in the first extended Kalman filter estimator according to a comparison result; S902, setting a numerical range of the reliability index; s903, determining an adjustment coefficient of the wheel speed observation noise covariance in the first extended Kalman filter estimator based on the numerical range; s904, calculating the adjusted observed noise covariance of the first extended Kalman filter estimator according to the adjustment coefficient and the initial wheel speed observed noise covariance; s905, inputting the adjusted observed noise covariance to the first extended Kalman filter estimator; s906, combining the adjusted observed noise covariance and the first posterior covariance to obtain the updated first posterior covariance.
9. The method for estimating a longitudinal vehicle speed by multi-sensor fusion according to claim 1, wherein S10 specifically comprises: S1001, respectively calculating a first weight and a second weight based on the updated first posterior covariance and the second posterior covariance; S1002, weighting the first longitudinal vehicle speed estimated value and the second longitudinal vehicle speed estimated value respectively based on the first weight and the second weight; And S1003, summing the weighted results to obtain the vehicle longitudinal speed estimated value.
10. A longitudinal vehicle speed estimation system with multi-sensor fusion is characterized by comprising a processor and a memory; The memory stores a program or instructions executable on the processor, which when executed by the processor, implement the steps of the multi-sensor fusion longitudinal vehicle speed estimation method of any one of claims 1 to 9.

Description

Longitudinal vehicle speed estimation method and system based on multi-sensor fusion Technical Field The invention relates to the technical field of active safety control of vehicle engineering, in particular to a longitudinal vehicle speed estimation method and system with multi-sensor fusion. Background The vehicle longitudinal speed estimation is a core component of vehicle motion state perception, provides key basic data support for various control strategies and navigation positioning functions of the vehicle, and becomes an important research direction in the field of vehicle perception by virtue of the advantages of instantaneity and suitability by means of a vehicle speed estimation method based on vehicle-mounted multi-sensor data. In the prior art, an extended Kalman filtering algorithm is often adopted to combine a vehicle dynamics model and a kinematic relation to realize longitudinal vehicle speed estimation, and simultaneously, a plurality of types of vehicle-mounted sensing data such as wheel speed, inertial measurement, satellite positioning and the like are fused to carry out estimation research. The vehicle speed dynamic tracking can be realized by means of the vehicle dynamics based on the three-degree-of-freedom vehicle dynamics model estimation mode, the vehicle speed absolute estimation accuracy can be guaranteed by combining satellite positioning data based on the vehicle longitudinal kinematics relation estimation mode, the characteristics of different sensing data are effectively integrated by the multi-sensor fusion technical thought, and an effective technical path is provided for improving the vehicle speed estimation accuracy. However, the single extended kalman filter estimator is easily limited by the working condition characteristics of the self observed data, the wheel speed observed data is easily influenced by the road surface and the tire state, the covariance stability of the observed noise is insufficient, and the stable estimation precision is difficult to maintain under the full working condition. Disclosure of Invention In view of the above-mentioned shortcomings of the prior art, an object of the present invention is to provide a multi-sensor fusion longitudinal vehicle speed estimation method, which can solve the technical problem that a single extended kalman filter estimator is easily limited by the working condition characteristics of self-observed data, and the wheel speed observed data is easily influenced by the road surface and the tire state, resulting in insufficient covariance stability of observed noise. In a first aspect of the embodiment of the present invention, a method for estimating a longitudinal vehicle speed by multi-sensor fusion is provided, including: s1, acquiring vehicle sensor data, wherein the vehicle sensor data comprises driving steering related sensing data and motion state reference sensing data; s2, establishing a three-degree-of-freedom vehicle dynamics model; S3, constructing a first extended Kalman filter estimator through a three-degree-of-freedom vehicle dynamics model based on driving steering associated sensing data; s4, calculating a first longitudinal vehicle speed estimated value and a first posterior covariance through a first extended Kalman filter estimator; S5, constructing a second extended Kalman filter estimator based on the longitudinal kinematic relation and the motion state reference sensing data of the vehicle; S6, calculating a second longitudinal vehicle speed estimated value and a second posterior covariance through a second extended Kalman filter estimator; s7, calculating a normalized innovation square value based on the innovation sequence generated by the first extended Kalman filter estimator; s8, comparing the normalized innovation square value with a detection threshold value; S9, carrying out self-adaptive adjustment on the first posterior covariance according to the comparison result to obtain an updated first posterior covariance; And S10, combining and updating the first posterior covariance and the second posterior covariance, carrying out weighted fusion on the first longitudinal vehicle speed estimated value and the second longitudinal vehicle speed estimated value, and outputting the longitudinal vehicle speed estimated value of the vehicle. In a second aspect of the embodiment of the invention, a longitudinal vehicle speed estimation system with multi-sensor fusion is provided, which comprises a processor and a memory; the memory stores a program or instructions executable on the processor which when executed by the processor performs the steps of the method for longitudinal vehicle speed estimation for multi-sensor fusion as described in the first aspect. The technical scheme provided by the embodiment of the invention has the beneficial effects that at least: In the embodiment of the invention, a first extended Kalman filter estimator based on a three-degree-of-freedom vehicle dynamics model