CN102009654B - Longitudinal speed evaluation method of full-wheel electrically-driven vehicle - Google Patents

Abstract

The present invention relates to a method for estimating the longitudinal vehicle speed of an all-wheel electric drive vehicle, which includes: 1) setting a vehicle speed measurement system; 2) real-time acquisition and

,

and

; 3) Adopt Kalman filter method to filter the collected signal; 4) Construct the vehicle speed estimation based on the Kalman filter space equation structure and the vehicle speed estimation based on the acceleration integral; 5) Use the vehicle speed estimation algorithm to switch the discrimination: set The threshold value of slip/slip rate absolute value |λ| is ε, when |λ|<ε, the vehicle speed estimation formula based on Kalman filter is used, and when |λ|≥ε, the vehicle speed estimation based on acceleration integral is used formula. The present invention is suitable for online vehicle speed estimation of all-wheel electric drive vehicles, including accurate observation of the longitudinal vehicle speed when the wheels appear to slip/skid excessively or even locked.

Description

A Longitudinal Velocity Estimation Method for All-Wheel Electric Drive Vehicles

technical field

The invention relates to a vehicle speed estimation method, in particular to a method for real-time online estimation of the longitudinal vehicle speed of an all-wheel electric drive vehicle. the

Background technique

In the process of vehicle dynamics control, the monitoring of the real-time state of the vehicle plays a very important role. The observation accuracy of some key vehicle states directly determines the effect of vehicle dynamics control. The vehicle state observation system conducts online real-time estimation of the key state parameters of the vehicle according to the on-board sensors, and transmits effective information to the control system, thereby realizing effective control of the vehicle state. Vehicle speed is an important reference for designing the parameters of vehicle stability controllers. Real-time online estimation of vehicle speed using common sensors on the vehicle is one of the key technologies required for vehicle stability control. the

There are two main methods for vehicle speed estimation based on vehicle-mounted common sensors. Method 1: Restore by wheel speed signal, that is, v _x = ωr±0.5wγ, where v _x is the estimated longitudinal vehicle speed, r is the tire radius, w is the wheelbase, and γ is the yaw rate. In the state of free rotation of the non-driving wheels, the longitudinal vehicle speed can be obtained directly according to this formula. However, all-wheel independent electric drive vehicles do not have non-driving wheels, and there is always slipping/slipping during the driving/braking process. If the non-driving wheel speed signal is directly used to restore, the slip/slip rate will cause a large error in the vehicle speed estimation. Therefore, the method of restoring vehicle speed based on wheel speed signals cannot be directly applied to all-wheel independent electric drive vehicles. Method 2: Restore the longitudinal vehicle speed based on the longitudinal acceleration signal, that is, v _x =v ₀ +∫a _x dt, where v ₀ is the initial speed of integration, a _x is the longitudinal acceleration signal, and t is the integration time length. The advantage of this method is that it is not affected by braking or driving conditions, but because the longitudinal acceleration signal is noisy, the long-term integration will cause the result to seriously deviate from the true value, so this method can only be used in a short time and is not applicable Long-term speed estimation for all-wheel independent electric drive vehicles. At the same time, in this method, how to determine the initial value of the integral is also an important issue.

Some scholars have also studied the speed observation algorithm based on the sliding mode algorithm and the speed observation algorithm based on the nonlinear state observer. These algorithms use complex nonlinear vehicle and tire models, and consider the influence of nonlinear characteristics on the accuracy of vehicle speed observation in vehicle speed observation. When these algorithms are used for vehicle speed estimation, the accuracy is generally high, but because more nonlinear iterative calculations are involved, the real-time performance is greatly affected. the

Contents of the invention

In view of the above problems, the object of the present invention is to provide an efficient and accurate method for estimating the longitudinal speed of an all-wheel electric drive vehicle online in real time. the

In order to achieve the above object, the present invention adopts the following technical solutions: a method for estimating the longitudinal speed of an all-wheel electric drive vehicle, which comprises the following steps:

1) A vehicle speed measurement system is set, which includes a longitudinal acceleration sensor, a steering wheel angle sensor and a tire rotational speed sensor;

、转向盘转角信号

和汽车质心纵向加速度信号，轮胎的轮边速度为

；  2) The tire speed signal is collected in real time through the vehicle speed measurement system

, steering wheel angle signal

and the longitudinal acceleration signal of the vehicle center of mass , the rim speed of the tire is

;

3) The collected signal is filtered by Kalman filter;

4) Using the filtered signal, two methods are used to estimate the vehicle speed online:

I. Construction of vehicle speed estimation based on Kalman filter space equation structure

① According to the vehicle kinematics, the following relationship is obtained:

a _x (t) = v′ _x -v _y γ (4)

v _w (t) = v _x (t) + Δv (5)

② Discretize equations (4) and (5) to get:

v _x (k+1)=v _x (k)+a _x ΔT+v _y (k)γ(k)ΔT (6)

v _w (k) = v _x (k) + Δv (7)

③Define w _s =v _y (k)γ(k)ΔT in formula (6) as the process noise of the system, and define w ₀ =Δv in formula (7) as the observation noise of the system, then the Kalman filter can be obtained The space equation structure of the device:

Equation of state v _x (k+1)=v _x (k)+a _x ΔT+w _s (8)

Observation equation v _w (k) = v _x (k) + w ₀ (9)

The online speed of the vehicle can be estimated through the space equation structure of the Kalman filter;

II. Construction of vehicle speed estimation based on acceleration integral

When the wheel enters the excessive slip/slip phase, at this time, the first derivative of the longitudinal vehicle speed and the longitudinal acceleration have the following relationship

v′ _x ＝a _x +v _y γ (10)

For (10) points get

v _x ＝v ₀ +∫(a _x +v _y γ)dt (11)

Wherein, v ₀ is the initial vehicle speed, at this moment, the vehicle speed estimation value v _x based on the Kalman filter is used as the initial value of the acceleration integral at this moment, and the vehicle speed estimation based on the acceleration integral is constructed;

5) Use the speed estimation algorithm to switch the discrimination:

Set the slip/slip rate absolute value |λ| threshold to ε,

When |λ|<ε, it is considered that there is no excessive slip/slip, and at this time, the vehicle speed estimation formula based on Kalman filter described in 4) is adopted,

When |λ|≥ε, it is considered that the wheel enters the excessive slip/slip phase, and the vehicle speed estimation formula based on acceleration integral described in 4) is used at this time. the

，其中k_Q是由车辆状态，尤其是车速决定的参数。  In step 4), assuming that w _s is an independent Gaussian distribution random signal, and assuming its variance matrix is Q, which is defined as process noise, it is adjusted as follows: According to the two-degree-of-freedom vehicle model, the yaw of the vehicle Both the angular velocity γ and the lateral velocity v _y are linear functions of the front wheel angle δ _f , and w _s can be regarded as a quadratic function of the steering wheel angle, namely

, where k _Q is a parameter determined by the vehicle state, especially the vehicle speed.

In step 4), it is assumed that w ₀ is an independent random signal with Gaussian distribution, and its variance matrix is assumed to be R, which is defined as observation noise, and it is adjusted as follows: When the tire slips/skids, the tire The absolute value of the slip/shift rate |λ|, and the absolute value of the difference between the angular acceleration a _w of the wheel and the longitudinal acceleration a _x of the center of mass |a _x -a _w | will all change, so when formulating fuzzy rules , select |λ| and Δa=|a _x -a _w | as the input of fuzzy rules, and establish the following relation of fuzzy inference adjustment rules:

Among them, S, M, L, and VL represent small, medium, large and very large. the

Due to the adoption of the above technical solutions, the present invention has the following advantages: 1. The method is suitable for online speed estimation of all-wheel electric drive vehicles; Accurately observe the vehicle speed, the method has high precision; 3. The invention considers the influence of steering on the wheel speed, so that the vehicle speed estimation during the steering process is also more accurate. the

Description of drawings

Fig. 1 is a schematic diagram of the overall architecture of the present invention;

Fig. 2 is process noise Q adjustment MAP figure of the present invention;

Fig. 3 is the |λ| membership function figure of the present invention;

Fig. 4 is the membership function diagram of |Δa| of the present invention;

Fig. 5 is the membership function figure of observation noise R of the present invention;

FIG. 6 is an adjustment map of observation noise R according to the present invention. the

Detailed ways

The present invention will be described in detail below in conjunction with the accompanying drawings and embodiments. the

The vehicle speed measuring system of the present invention collects vehicle-mounted general sensor signals (comprising longitudinal acceleration sensor signals, steering wheel angle sensor signals) and drive motor feedback signals (i.e. tire rotational speed signals), utilizes the vehicle speed estimation method based on Kalman filter and the vehicle speed based on acceleration integral The estimation method is estimated separately, and the slip rate threshold value is used to switch and judge, and then a more accurate vehicle speed value is obtained. the

As shown in Figure 1, the inventive method comprises the following steps:

1) Signal acquisition

、转向盘转角信号

和汽车质心纵向加速度信号。轮胎的轮边速度为

。  Real-time collection of vehicle wheel speed, longitudinal acceleration and steering wheel angle and other information. Collected tire speed signal

, steering wheel angle signal

and the longitudinal acceleration signal of the vehicle center of mass . The rim speed of the tire is

.

2) Filter the collected signal

Due to the inclusion of environmental noise in the process of signal acquisition and transmission, the acquired original signal often has glitches and errors. This signal with high-frequency noise not only cannot meet the requirements of use, but also will further amplify this error when obtaining the extended derivative signal, resulting in unrecognizable results. Therefore, these signals must be filtered before use, and the filter should meet the following requirements: the order is low and the signal is smooth. Based on this, any filter that meets the requirements can be designed to perform filtering processing. The present invention adopts a Kalman filtering method to perform filtering processing on the collected signals. the

Kalman filtering is an optimal filtering method based on time series theory. Using the Kalman filter method can not only filter the original signal online in real time, but also use the characteristics of the Kalman filter based on the state space equation to spontaneously restore the extended derivative signal of the original signal from the Kalman filter. the

Through the filtering signal processing in this step, relatively smooth wheel edge velocity v _w , wheel edge acceleration a _w , longitudinal acceleration a _x and steering wheel angle δ _w signals are obtained.

3) Construct a vehicle speed estimation method based on Kalman filter

Set the threshold value of the absolute value of the slip/slip rate |λ| to ε, when |λ|<ε, it is considered that there is no excessive slip/slip, at this time, using the Kalman filter-based vehicle speed estimation method will get better results. the

According to the vehicle kinematics, the following relationship is obtained:

a _x (t) = v′ _x -v _y γ (4)

v _w (t) = v _x (t) + Δv (5)

Discretize equations (4) and (5) to get:

v _x (k+1)=v _x (k)+a _x ΔT+v _y (k)γ(k)ΔT (6)

v _w (k) = v _x (k) + Δv (7)

Define w _s =v _y (k)γ(k)ΔT in formula (6) as the process noise of the system, define w ₀ =Δv in formula (7) as the observation noise of the system, formulas (6), (7 ) can be rewritten as follows:

Equation of state v _x (k+1)=v _x (k)+a _x ΔT+w _s (8)

Observation equation v _w (k) = v _x (k) + w ₀ (9)

Assume w _s and w ₀ are both independent Gaussian distributed random signals, and assume their variance matrices are Q and R, respectively. The space equation structure of the complete Kalman filter can be formed by formulas (8) and (9).

The process noise and observation noise in the Kalman filter algorithm are affected by the state of the vehicle. When estimating the vehicle speed, the process noise and observation noise can be adjusted online according to the current state of the vehicle to obtain more accurate results. the

，其中k_Q是由车辆状态，尤其是车速决定的参数。图2是过程噪声方差Q随前轮转角和车速变化的调节曲线。  Adjustment of process noise variance Q: Process noise w _s =v _y (k)γ(k)ΔT depends on whether the vehicle turns or not. According to the two-degree-of-freedom vehicle model, the vehicle's yaw rate γ and lateral velocity v _y are both linear functions of the front wheel rotation angle δ _w . Therefore, the process noise w _s can be regarded as a quadratic function of the steering wheel angle, namely

, where k _Q is a parameter determined by the vehicle state, especially the vehicle speed. Fig. 2 is the adjustment curve of process noise variance Q changing with front wheel angle and vehicle speed.

Adjustment of observed noise variance R: observed noise w ₀ =Δv characterizes the difference between wheel speed and vehicle speed. When the slip/shift rate is low, the difference between wheel speed and vehicle speed is small, so the observed noise variance R should be correspondingly small. When the tire starts to slip/skid, the difference between the wheel speed and the vehicle speed gradually increases, and the observed noise variance R should also increase accordingly. In order to better adjust the noise variance R, a fuzzy adjustment method is adopted here. When the tire slips/slips, the absolute value of the tire's slip/move rate |λ|, and the absolute value of the difference between the angular acceleration of the wheel rotation (converted to tangential acceleration) and the longitudinal acceleration of the center of mass |a _x -a _w | both change. Therefore, when formulating fuzzy rules, choose |λ| and Δa=|a _x -a _w | as the input of fuzzy rules. The following table shows the fuzzy inference regulation rules for observation noise R.

Table 1 Fuzzy rule adjustment of observation noise R

Among them, S, M, L, VL represent small, medium, large and very large respectively, and design the membership function of each input and output (as shown in Fig. 3, Fig. 4 and Fig. 5). Figure 6 shows the adjustment surface for the observed noise R. the

4) Construct a vehicle speed estimation method based on acceleration integral

When |λ|≥ε, it is considered that the wheel enters the excessive slip/slip stage. At this time, switching to the speed estimation algorithm based on acceleration integration will get better results. the

As shown in formula (10), the first derivative of longitudinal vehicle speed and longitudinal acceleration have the following relationship

v′ _x ＝a _x +v _y γ (10)

For (10) points get

v _x ＝v ₀ +∫(a _x +v _y γ)dt (11)

Among them, v ₀ is the initial vehicle speed. When switching to the estimation algorithm based on the acceleration integral, the estimated value v _x of the vehicle speed based on the Kalman filter at the previous moment is used as the initial value of the acceleration integral. Generally, when the estimation algorithm is switched to an algorithm based on acceleration integral, it must be the case of emergency braking and high acceleration. In this case, the term v _y γ is a minor term relative to a _x . At the same time, on vehicles equipped with ABS, the time for tires to lock up is generally very short. In this case, equation (11) can be approximated as

v _x ＝v ₀ +∫a _x dt (12)

And the error caused by acceleration noise integration can be ignored in this short integration time. the

5) Speed estimation algorithm switching discrimination

Set the threshold value of the absolute value of slip/slip ratio |λ| to ε, when |λ|<ε, it is considered that there is no excessive slip/slip, and when |λ|≥ε, it is considered that the wheel enters excessive slip turn/slip phase. the

When |λ|<ε, it is considered that there is no excessive slip/slip, and at this time, the vehicle speed estimation method based on Kalman filter described in 3) can be used to estimate the vehicle speed at each moment through Kalman filter iterations . the

When |λ|≥ε, it is considered that the wheel enters the excessive slip/slip phase, and at this time switch to the vehicle speed estimation method based on acceleration integration described in 4). This is because the Kalman filter based on formula (9) uses the wheel speed signal v _w as the feedback correction mechanism. When the wheel slips/slips excessively, the correlation between the wheel speed and the vehicle speed is small; In the event of a lockup or full spin, there is no correlation between wheel speed and vehicle speed. At this time, if the Kalman filter algorithm is still used, the obtained results will have a large deviation.

When the absolute value of the slip rate |λ| satisfies |λ|<ε again, it is considered that the wheel has exited the excessive slip/slip phase, and it can be switched back to the Kalman filter-based vehicle speed estimation method described in 3), Kalman The initial value of the filter is the velocity value obtained based on the acceleration integration at the previous moment. the

。  Through the above vehicle speed estimation method, the estimated value of vehicle speed can be obtained at each sampling time k

.

Claims (3)

1. vertical vehicle speed estimation method of a full wheel electro-motive vehicle, it may further comprise the steps:

1) vehicle speed measurement system is set, it comprises longitudinal acceleration sensor, steering wheel angle sensor and tire rotational speed sensor;

2) arrive the tire rotational speed signal by the vehicle speed measurement system Real-time Collection

, the steering wheel angle signal

With automobile barycenter longitudinal acceleration signal , the wheel limit speed of tire is

, wherein r is tire radius;

3) taking the Kalman filtering mode that the signal that collects is carried out filtering processes;

4) signal after utilizing filtering to process, adopt two kinds of methods that the speed of a motor vehicle is carried out On-line Estimation:

I, structure are estimated based on the speed of a motor vehicle of Kalman filter space equation structure

1. obtain following relational expression according to vehicle kinematics:

a _x(t)＝v′ _x-v _yγ (4)

v _w(t)＝v _x(t)+Δv (5)

Wherein, a _x(t) be t longitudinal acceleration constantly, v ' _xBe vertical speed of a motor vehicle derivative, v _yBe the side direction speed of a motor vehicle, γ is yaw velocity, v _w(t) be t wheel limit speed constantly, v _x(t) be t vertical speed of a motor vehicle constantly, Δ v is the observation noise of system;

2. the discretization processing being done in formula (4), (5) obtains:

v _x(k+1)＝v _x(k)+a _xΔT+v _y(k)γ(k)ΔT (6)

v _w(k)＝v _x(k)+Δv (7)

Wherein, v _x(k+1) be k+1 vertical speed of a motor vehicle constantly, v _x(k) be k vertical speed of a motor vehicle constantly, a _xBe longitudinal acceleration, v _y(k) be the k side direction speed of a motor vehicle constantly, γ (k) is k yaw velocity constantly, v _w(k) be k wheel limit speed constantly;

3. with w in the formula (6) _s=v _y(k) γ (k) Δ T is defined as the process noise of system, with w in the formula (7) ₀=Δ v is defined as the observation noise of system, can obtain the space equation structure of Kalman filter:

Equation of state v _x(k+1)=v _x(k)+a _xΔ T+w _s(8)

Observational equation v _w(k)=v _x(k)+w ₀(9)

Can estimate the online speed of a motor vehicle of vehicle by the space equation structure of Kalman filter;

II, structure are estimated based on the speed of a motor vehicle of integrated acceleration

Trackslip when wheel enters excessively/the slippage stage, at this moment, vertically first derivative and the longitudinal acceleration of the speed of a motor vehicle have following relation

v′ _x＝a _x+v _yγ (10)

(10) integration is obtained

v _x＝v ₀+∫(a _x+v _yγ)dt (11)

Wherein, v ₀Be the initial speed of a motor vehicle, this moment is with the speed of a motor vehicle estimated valve v of a upper moment based on Kalman filtering space equation structure _xAs the initial value of integrated acceleration, make up based on the speed of a motor vehicle of integrated acceleration and estimate;

5) utilizing speed of a motor vehicle algorithm for estimating to switch differentiates:

Setting is trackslipped/the slip rate absolute value | λ | threshold values be ε,

As | λ | during＜ε, think not occur excessively not trackslip/slippage that adopt 4 this moment) described in the speed of a motor vehicle estimation formulas based on Kalman filtering,

As | λ | during 〉=ε, think that wheel enters excessively to trackslip/the slippage stage that adopt 4 this moment) described in the speed of a motor vehicle estimation formulas based on integrated acceleration.

2. vertical vehicle speed estimation method of a kind of full wheel electro-motive vehicle as claimed in claim 1 is characterized in that: in step 4) (8) formula in, suppose w _sBe Gaussian distribution stochastic signal independently, and suppose that its variance battle array is Q, be defined as process noise, then it is regulated as follows:

According to the two degrees of freedom auto model, the yaw velocity γ of vehicle and side direction speed of a motor vehicle v _yAll are front wheel angle δ _fLinear function, w _sThe quadratic function that can regard steering wheel angle as, namely

, k wherein _QIt is the parameter that is determined by the speed of a motor vehicle.

3. vertical vehicle speed estimation method of a kind of full wheel electro-motive vehicle as claimed in claim 1 is characterized in that: in step 4) (9) formula in, suppose w ₀Be Gaussian distribution stochastic signal independently, and suppose that its variance battle array is R, be defined as observation noise, then it is regulated as follows:

When tire occur trackslipping/during slippage, the absolute value of the trackslipping of tire/slip rate | λ |, and the absolute difference between longitudinal acceleration and the vehicle wheel rotation angular acceleration | a _x-a _w| all can change, therefore when formulating fuzzy rule, choose | λ | and Δ a=|a _x-a _w| as the input of fuzzy rule, set up following fuzzy reasoning and regulate rule relation:

S wherein, M, L, the VL representative is little, in, greatly and very large.

Images