CN103640622A - Automobile direction intelligent control method and control system based on driver model - Google Patents

Abstract

The invention discloses an automobile direction intelligent control method based on a driver model. The method includes the steps of firstly, predicting a position where an automobile can reach after a moment T of a preview period according to the current running state of the automobile, comparing the position with a target position after the moment T, obtaining deviation of the two positions, obtaining the ratio of the deviation and T to serve as a predicted automobile transverse velocity, and then comparing the predicted automobile transverse velocity with an actual automobile transverse velocity to obtain a predicted automobile transverse velocity difference; secondly, obtaining a steering wheel turning angle through calculation, and controlling an automobile steering wheel according to the obtained steering wheel turning angle. The invention further discloses an automobile direction intelligent control system based on the drive model. The system comprises a preview module, a prediction module, a comparing module, a calculation module and a control module. Compared with the prior art, parameters of the driver model built in the method are directly obtained through parameters of the whole automobile, the driver module has the advantages of being simple in parameter and clear in physical meaning, and control of the automobile is more accurate and real.

Description

A kind of automobile steering intelligent control method and control system based on pilot model

Technical field

The present invention relates to a kind of automobile intelligent control method, relate in particular to a kind of automobile steering intelligent control method and control system based on pilot model, belong to automatic control technology field.

Background technology

Along with scientific and technological progress and the raising of social level, people are increasing to the demand of automobile, and the owning amount of private car rises year by year, caused thus traffic safety problem and also become increasingly conspicuous.This situation is particularly evident in China, and China leaps to the whole world first for continuous ten years because of the number of traffic accidents death.Therefore, vehicle safety problem becomes most important.

For transport solution safety problem, not only need to pay close attention to the formulation of various traffic safety codes, the research to intelligent transportation system, intelligent vehicle and other active safety technology, also needs to be used to improve the drive safety of automobile.And driver modeling is the basis of these researchs, set up one more accurately, more real pilot model not only can be for intelligent vehicle, also can be for the detection of vehicle handling stability.

In recent years, both at home and abroad researchist and scholar have carried out large quantity research to automobile driver model, have successively proposed several pilot models, such as the people's such as the optimal preview control model of MacAdam, the preview follower model of Guo Konghui and Cole LQR model etc.But more conventional pilot model is both at home and abroad, the people's such as China Guo Konghui academician's preview follower model and external Cole LQR model.Preview follower model because of the physical meaning of its model parameter clear, be convenient to understand the mechanism of production of driving behavior, application is convenient.But it is when high speed steering, and tracking accuracy is not high and occur turning to wave phenomenon, makes its application under high speed have certain limitation; LQR model adopts optimization method, and theoretical rigorous, tracking accuracy is high.But its physical conception is unintelligible, adopt multipoint preview to cause optimizing calculated amount large, practical application difficulty simultaneously.

Summary of the invention

Technical matters to be solved by this invention is to overcome prior art deficiency, a kind of automobile steering intelligent control method and control system based on pilot model is provided, the parameter of the pilot model adopting is directly obtained by whole-car parameters, have that parameter is simple, physical meaning advantage clearly, to the control of automobile more accurately and true.

The present invention is specifically by the following technical solutions:

A kind of automobile steering intelligent control method based on pilot model, first according to automobile, the position that the constantly rear automobile of cycle T can arrive is taken aim in current running state prediction in advance, and compare with the target location of T after the moment, obtain both deviations, the ratio of both deviations and T is the automobile cross velocity of expectation, the automobile cross velocity of expectation and actual automobile cross velocity are compared, the automobile cross velocity that obtains expectation is poor again; Then according to following formula, obtain steering wheel angle, and according to the steering wheel angle obtaining, vehicle steering controlled:

θ(s)=e _vy(s)×G _d(s)

In formula, θ (s) represents steering wheel angle; e _vy(s) represent that the automobile cross velocity of expecting is poor; G _d(s) be transfer function, according to following formula, obtain:

$G_d\left(s\right)=\frac{{\mathrm{\ω}}_c}{v}(P+\mathrm{Ds})$

$P=\frac{k_0}{{\mathrm{<figure-callout\; id="0"\; label="l"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">l</figure-callout>}}_0},D=\frac{k_1+k_0{T}_h}{l_0}-\frac{k_0{l}_1}{{\mathrm{<figure-callout\; id="0"\; label="l"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">l</figure-callout>}}_0^2}$

${\mathrm{<figure-callout\; id="0"\; label="k"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">k</figure-callout>}}_0=\frac{l_r{c}_r-l_f{c}_f}{I_z}+\frac{c_f{c}_r{L}^2}{m{v}^2{I}_z},k_1=\frac{{l_f}^2{c}_f+{l_r}^2{c}_r}{v{I}_z}+\frac{c_f+c_r}{\mathrm{mv}},{\mathrm{<figure-callout\; id="0"\; label="l"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">l</figure-callout>}}_0=\frac{L{c}_f{c}_r}{\mathrm{mv}{I}_z},{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000414166380000013.png,HDA0000414166380000021.png"\; state="\{\{state\}\}">l</figure-callout>}}_1=\frac{l_{f}L{c}_f{c}_r}{m{v}^2{I}_z}$

Wherein, m is vehicle complete vehicle quality, and unit is kg; V is the speed of a motor vehicle, and unit is m/s; I _zfor automobile yaw rotor inertia, unit is kgm ²; L be automobile wheel base from, unit is m; l _ffor the distance of vehicle complete vehicle barycenter to front axle, unit is m; l _rfor the distance of vehicle complete vehicle barycenter to rear axle, unit is m; c _ffor the equivalent cornering stiffness of vehicle front, unit is N/rad; c _rfor the equivalent cornering stiffness of automobile back wheel, unit is N/rad; T _hfor the arm inertial delay of default chaufeur, unit is s; ω _cthe cutoff frequency of behaviour car closed loop system, unit is rad/s; S is granny rag Laplacian operater.

An automobile steering intelligent control system based on pilot model, comprising:

Take aim in advance module, utilize the road information observing to generate and take aim in advance the target location f (t+T) of cycle T after the moment;

Prediction module, dopes automobile according to the current running state information of automobile and is taking aim in advance the cycle T rear position y (t+T) that can arrive constantly;

Comparison module, compares target location f (t+T) and automobile taking aim in advance the position y (t+T) that cycle T can arrive constantly output bias e (t+T)=f (t+T)-y (t+T);

Computing module, divided by taking aim in advance cycle T, obtains the automobile cross velocity of expectation with described deviation e (t+T)

again with the cross velocity v of actual automobile _y(t) compare, obtain the poor e of automobile cross velocity of expectation _vy(t)=v _y ^*(t)-v _y(t);

Control module, poor according to the automobile cross velocity of expectation, according to following formula, obtain steering wheel angle, and according to the steering wheel angle obtaining, vehicle steering is controlled:

θ(s)=e _vy(s)×G _d(s)

In formula, θ (s) represents steering wheel angle; e _vy(s) represent that the automobile cross velocity of expecting is poor; G _d(s) be transfer function, according to following formula, obtain:

$G_d\left(s\right)=\frac{{\mathrm{\&omega;}}_c}{v}(P+\mathrm{Ds})$

$P=\frac{k_0}{{\mathrm{<figure-callout\; id="0"\; label="l"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">l</figure-callout>}}_0},D=\frac{k_1+k_0{T}_h}{l_0}-\frac{k_0{l}_1}{{\mathrm{<figure-callout\; id="0"\; label="l"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">l</figure-callout>}}_0^2}$

${\mathrm{<figure-callout\; id="0"\; label="k"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">k</figure-callout>}}_0=\frac{l_r{c}_r-l_f{c}_f}{I_z}+\frac{c_f{c}_r{L}^2}{m{v}^2{I}_z},k_1=\frac{{l_f}^2{c}_f+{l_r}^2{c}_r}{v{I}_z}+\frac{c_f+c_r}{\mathrm{mv}},{\mathrm{<figure-callout\; id="0"\; label="l"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">l</figure-callout>}}_0=\frac{L{c}_f{c}_r}{\mathrm{mv}{I}_z},{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000414166380000013.png,HDA0000414166380000021.png"\; state="\{\{state\}\}">l</figure-callout>}}_1=\frac{l_f{\mathrm{Lc}}_f{c}_r}{m{v}^2{I}_z}$

Wherein, m is vehicle complete vehicle quality, and unit is kg; V is the speed of a motor vehicle, and unit is m/s; I _zfor automobile yaw rotor inertia, unit is kgm ²; L be automobile wheel base from, unit is m; l _ffor the distance of vehicle complete vehicle barycenter to front axle, unit is m; l _rfor the distance of vehicle complete vehicle barycenter to rear axle, unit is m; c _ffor the equivalent cornering stiffness of vehicle front, unit is N/rad; c _rfor the equivalent cornering stiffness of automobile back wheel, unit is N/rad; T _hfor the arm inertial delay of default chaufeur, unit is s; ω _cthe cutoff frequency of behaviour car closed loop system, unit is rad/s; S is granny rag Laplacian operater.

The arm inertial delay T of described chaufeur _hspan be preferably 0.1-0.3s.

The cutoff frequency ω of described people's car closed loop system _cspan be preferably 1～6rad/s.

Compared to existing technology, the present invention has following beneficial effect:

The pilot model that the present invention sets up, its parameter is directly obtained by whole-car parameters, has that parameter is simple, a physical meaning advantage clearly, to the control of automobile more accurately and true.

Accompanying drawing explanation

Fig. 1 is the structural principle schematic diagram of pilot model of the present invention;

Fig. 2 is the control principle schematic diagram of automobile steering intelligent control system of the present invention;

Fig. 3 is the simulating, verifying comparison diagram of pilot model of the present invention and preview follower pilot model;

Fig. 4 is the track following error comparison diagram of pilot model of the present invention and preview follower pilot model.

The specific embodiment

Below in conjunction with accompanying drawing, technical scheme of the present invention is elaborated:

Automobile steering intelligent control system based on pilot model of the present invention, its core is set up pilot model, for example, by the observation system (image collecting device that automobile carries, radar etc.) outside condition of road surface is observed, internal sensor/observer by automobile detects the oneself state information of automobile, pilot model calculates outbound course dish corner according to observed road information and the automobile oneself state that detects, and according to steering wheel angle, vehicle steering is controlled, thereby realize the accurately intelligent vehicle direction closed loop control of reflection chaufeur driving behavior.

The pilot model that the present invention sets up as shown in Figure 1, comprising:

Take aim in advance module, the road information (also can directly input data by outside) that observation system data such as () image collecting device carrying such as automobile, radars is observed is processed (method that for example adopts fitting of a polynomial) and is generated and take aim in advance the target location f (t+T) of cycle T after constantly (or be called take aim in advance a position), and wherein t represents current time;

Prediction module, dopes automobile according to the current running state information of automobile and is taking aim in advance the cycle T position y (t+T) that rear (being the t+T moment) can arrive constantly;

Comparison module, compares target location f (t+T) and automobile taking aim in advance the position y (t+T) that cycle T can arrive constantly output bias e (t+T)=f (t+T)-y (t+T);

Computing module, first calculates the automobile cross velocity of expectation

again with the cross velocity v of vehicle condition information _y(t) compare, ask for the poor e of automobile cross velocity of expectation _vy(t)=v _y ^*(t)-v _y(t);

Control module, poor according to the automobile cross velocity of expectation, according to following formula, obtain steering wheel angle, and according to the steering wheel angle obtaining, vehicle steering is controlled:

θ(s)=e _vy(s)×G _d(s)

In formula, θ (s) represents steering wheel angle; e _vy(s) represent that the automobile cross velocity of expecting is poor; G _d(s) be transfer function, according to following formula, obtain:

$G_d\left(s\right)=\frac{{\mathrm{\&omega;}}_c}{v}(P+\mathrm{Ds})$

$P=\frac{k_0}{{\mathrm{<figure-callout\; id="0"\; label="l"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">l</figure-callout>}}_0},D=\frac{k_1+k_0{T}_h}{l_0}-\frac{k_0{l}_1}{{\mathrm{<figure-callout\; id="0"\; label="l"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">l</figure-callout>}}_0^2}$

${\mathrm{<figure-callout\; id="0"\; label="k"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">k</figure-callout>}}_0=\frac{l_r{c}_r-l_f{c}_f}{I_z}+\frac{c_f{c}_r{L}^2}{m{v}^2{I}_z},k_1=\frac{{l_f}^2{c}_f+{l_r}^2{c}_r}{v{I}_z}+\frac{c_f+c_r}{\mathrm{mv}},{\mathrm{<figure-callout\; id="0"\; label="l"\; filenames="HDA0000414166380000013.png"\; state="\{\{state\}\}">l</figure-callout>}}_0=\frac{L{c}_f{c}_r}{\mathrm{mv}{I}_z},{\mathrm{<figure-callout\; id="1"\; label="l"\; filenames="HDA0000414166380000013.png,HDA0000414166380000021.png"\; state="\{\{state\}\}">l</figure-callout>}}_1=\frac{l_{f}L{c}_f{c}_r}{m{v}^2{I}_z}$

Wherein, m is vehicle complete vehicle quality, and unit is kg; V is the speed of a motor vehicle, and unit is m/s; I _zfor automobile yaw rotor inertia, unit is kgm ²; L be automobile wheel base from, unit is m; l _ffor the distance of vehicle complete vehicle barycenter to front axle, unit is m; l _rfor the distance of vehicle complete vehicle barycenter to rear axle, unit is m; c _ffor the equivalent cornering stiffness of vehicle front, unit is N/rad; c _rfor the equivalent cornering stiffness of automobile back wheel, unit is N/rad; T _hfor the arm inertial delay of default chaufeur, unit is s, and its span is preferably 0.1-0.3s; ω _cthe cutoff frequency of behaviour car closed loop system, unit is rad/s, its span is preferably 1～6rad/s; S is granny rag Laplacian operater.

The control principle of automobile steering intelligent control system of the present invention as shown in Figure 2, internal sensor/the observer of automobile detects the status information of automobile and feeds back to pilot model, road information is directly inputted by outside, pilot model calculates steering wheel angle according to road information and vehicle condition information, and according to the steering wheel angle obtaining, controlled automobile travel direction is controlled, thereby form people-Che-Lu closed loop system, reach the object of following the tracks of path locus, can control for vehicle handling stability simultaneously.

In order to verify effect of the present invention, adopt pilot model of the present invention and existing preview follower pilot model to carry out respectively the l-G simulation test of two-track thread test.Two-track thread test is the closed test of synthesis measuring chaufeur-vehicle handling stability, can more fully study and evaluate the road-holding property of automobile.The reference road adopting in test is input as ISO-3888-1:1999 governing criterion two-track lineman condition; In order to embody validity of the present invention, now adopt identical whole-car parameters and samely take aim in advance the time, to pilot model of the present invention with take aim in advance and follow the tracks of pilot model and carry out simulation comparison analysis.The verification msg that emulation adopts has: m=1715Kg, v=30m/s, I _z=2697Kgm ², L=2.54m, l _f=1.07m, l _r=1.47m, c _f=2*89733N/rad, c _r=2*114100N/rad, T _h=0.15s, ω _c=2.5rad/s, T=0.8s.As shown in Figure 3, Figure 4, wherein Fig. 4 is the track following error comparison diagram of pilot model of the present invention and preview follower pilot model to checking comparing result.Track following error simulation result as shown in Figure 4 can find out, tracking accuracy of the present invention is higher, and maximum tracking error is reduced to 0.7m compared with the 1.2m of preview follower model, and maximum tracking error has reduced 42%.In addition, the tracking error of the inventive method does not have larger fluctuating as seen from Figure 4, shows that ride comfort of the present invention is better.Automobile, after lane change, finally reaches stable distance and is reduced to 220m by original 250m, shows that stabilization time of the present invention is less, and stability and responsibility are better.

Claims (6)

1. the automobile steering intelligent control method based on pilot model, is characterized in that, first according to automobile, the cycle is taken aim in current running state prediction in advance tthe position that constantly, automobile can arrive, and with ttarget location constantly compares, and obtains both deviations, both deviations with tratio be the automobile cross velocity of expectation, then the automobile cross velocity of expectation and actual automobile cross velocity are compared, the automobile cross velocity that obtains expectation is poor; Then according to following formula, obtain steering wheel angle, and according to the steering wheel angle obtaining, vehicle steering controlled:

In formula,

represent steering wheel angle;

the automobile cross velocity that represents expectation is poor;

for transfer function, according to following formula, obtain:

Wherein, mfor vehicle complete vehicle quality, unit is kg; vfor the speed of a motor vehicle, unit is m/s; i _zfor automobile yaw rotor inertia, unit is kgm ²; lfor automobile wheel base from, unit is m; l _f for the distance of vehicle complete vehicle barycenter to front axle, unit is m; l _r for the distance of vehicle complete vehicle barycenter to rear axle, unit is m; c _f for the equivalent cornering stiffness of vehicle front, unit is N/rad; c _r for the equivalent cornering stiffness of automobile back wheel, unit is N/rad; t _h for the arm inertial delay of default chaufeur, unit is s; ω _c the cutoff frequency of behaviour car closed loop system, unit is rad/s; sfor granny rag Laplacian operater.

2. the automobile steering intelligent control method based on pilot model as claimed in claim 1, is characterized in that, described in drive

The arm inertial delay of the person of sailing t _h span be 0.1-0.3s.

3. the automobile steering intelligent control method based on pilot model as claimed in claim 1, is characterized in that described people

The cutoff frequency of car closed loop system ω _c span be 1 ~ 6rad/s.

4. the automobile steering intelligent control system based on pilot model, is characterized in that, comprising:

Take aim in advance module, utilize the road information observing to generate the cycle of taking aim in advance ttarget location constantly f( t+T);

Prediction module, dopes automobile according to the current running state information of automobile and is taking aim in advance the cycle tthe position that can arrive constantly y( t+ t);

Comparison module, by target location f( t+T) taking aim in advance the cycle with automobile tthe position that can arrive constantly y( t+ t) compare output bias e( t+ t)= f( t+T)- y( t+ t);

Computing module, with described deviation e( t+ t) divided by taking aim in advance the cycle t, obtain the automobile cross velocity of expecting

; Again with the cross velocity of actual automobile v _y ( t) compare, the automobile cross velocity that obtains expectation is poor e _vy ( t)= v _y ^*( t)- v _y ( t);

Control module, poor according to the automobile cross velocity of expectation, according to following formula, obtain steering wheel angle, and according to the steering wheel angle obtaining, vehicle steering is controlled:

In formula,

represent steering wheel angle;

the automobile cross velocity that represents expectation is poor;

for transfer function, according to following formula, obtain:

Wherein, mfor vehicle complete vehicle quality, unit is kg; vfor the speed of a motor vehicle, unit is m/s; i _zfor automobile yaw rotor inertia, unit is kgm ²; lfor automobile wheel base from, unit is m; l _f for the distance of vehicle complete vehicle barycenter to front axle, unit is m; l _r for the distance of vehicle complete vehicle barycenter to rear axle, unit is m; c _f for the equivalent cornering stiffness of vehicle front, unit is N/rad; c _r for the equivalent cornering stiffness of automobile back wheel, unit is N/rad; t _h for the arm inertial delay of default chaufeur, unit is s; ω _c the cutoff frequency of behaviour car closed loop system, unit is rad/s; sfor granny rag Laplacian operater.

5. the automobile steering intelligent control system based on pilot model as claimed in claim 4, is characterized in that, described in drive

The arm inertial delay of the person of sailing t _h span be 0.1-0.3s.

6. the automobile steering intelligent control system based on pilot model as claimed in claim 4, is characterized in that described people

The cutoff frequency of car closed loop system ω _c span be 1 ~ 6rad/s.

Images