CN108859648B - Suspension shock absorber damping control switching weighting coefficient determination method - Google Patents

Suspension shock absorber damping control switching weighting coefficient determination method Download PDF

Info

Publication number
CN108859648B
CN108859648B CN201810933191.5A CN201810933191A CN108859648B CN 108859648 B CN108859648 B CN 108859648B CN 201810933191 A CN201810933191 A CN 201810933191A CN 108859648 B CN108859648 B CN 108859648B
Authority
CN
China
Prior art keywords
suspension
shock absorber
damping force
vehicle body
rear wheel
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201810933191.5A
Other languages
Chinese (zh)
Other versions
CN108859648A (en
Inventor
陈双
包黎明
赵凯旋
陈剑桥
李政原
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Liaoning University of Technology
Original Assignee
Liaoning University of Technology
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Liaoning University of Technology filed Critical Liaoning University of Technology
Priority to CN201810933191.5A priority Critical patent/CN108859648B/en
Publication of CN108859648A publication Critical patent/CN108859648A/en
Application granted granted Critical
Publication of CN108859648B publication Critical patent/CN108859648B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B60—VEHICLES IN GENERAL
    • B60G—VEHICLE SUSPENSION ARRANGEMENTS
    • B60G17/00—Resilient suspensions having means for adjusting the spring or vibration-damper characteristics, for regulating the distance between a supporting surface and a sprung part of vehicle or for locking suspension during use to meet varying vehicular or surface conditions, e.g. due to speed or load
    • B60G17/015—Resilient suspensions having means for adjusting the spring or vibration-damper characteristics, for regulating the distance between a supporting surface and a sprung part of vehicle or for locking suspension during use to meet varying vehicular or surface conditions, e.g. due to speed or load the regulating means comprising electric or electronic elements
    • B60G17/0152—Resilient suspensions having means for adjusting the spring or vibration-damper characteristics, for regulating the distance between a supporting surface and a sprung part of vehicle or for locking suspension during use to meet varying vehicular or surface conditions, e.g. due to speed or load the regulating means comprising electric or electronic elements characterised by the action on a particular type of suspension unit
    • B60G17/0157—Resilient suspensions having means for adjusting the spring or vibration-damper characteristics, for regulating the distance between a supporting surface and a sprung part of vehicle or for locking suspension during use to meet varying vehicular or surface conditions, e.g. due to speed or load the regulating means comprising electric or electronic elements characterised by the action on a particular type of suspension unit non-fluid unit, e.g. electric motor
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F30/00—Computer-aided design [CAD]
    • G06F30/10—Geometric CAD
    • G06F30/15—Vehicle, aircraft or watercraft design

Abstract

The invention discloses a method for determining a suspension shock absorber damping control switching weighting coefficient based on a neural network, which comprises the following steps: collecting automobile monitoring data, and converting the automobile monitoring data into sample data of a neural network based on a seven-degree-of-freedom motion differential equation with linear characteristics including automobile body vertical displacement, automobile body pitch angle, automobile body roll angle and vertical displacement of front and rear four wheels; classifying the sample data according to road condition types to obtain a sample data set corresponding to each road condition; according to the method, the damping force of the suspension shock absorber is controlled according to different working conditions, the safety and the comfort of the automobile are improved, and stable switching is realized.

Description

Suspension shock absorber damping control switching weighting coefficient determination method
Technical Field
The invention relates to the field of multi-working-condition switching control of an automobile active suspension, in particular to a neural network-based switching control characteristic weighting coefficient determination method.
Background
Suspension systems are important components of vehicles. With the development of science and technology and the continuous progress of control technology, the traditional passive suspension system restricts the exertion of the function of the suspension system because the parameters of the passive suspension system can not be changed, and the active suspension system can realize the optimal performance under various working conditions because the parameters of the active suspension system are adjustable, and accords with the development trend of low carbon, light weight, intellectualization and individualization of the future automobile, thereby becoming the popular and trend of research. In order to improve the comfort and safety of automobiles, the increasing development of modern automobile technology makes the suspension system more and more intelligent. The active suspension system and the semi-active suspension system can enable the vehicle to achieve the optimal running smoothness and the optimal operation stability under various road surface working conditions through real-time control, and have the advantages which cannot be compared with the traditional suspension.
In the vehicle driving process, the driving working condition is changed continuously, different control modes are adopted to control the active suspension according to different working conditions, and the comfort and the safety of the automobile are improved. When the road surface working condition changes, the switching control force can generate phenomena of blocking, sudden change and the like, so that the research of the smooth switching of the active suspension multi-working condition switching control is particularly important.
Disclosure of Invention
The invention designs and develops a suspension shock absorber damping control switching weighting coefficient determination method based on a neural network, establishes a neural network model aiming at different working conditions, obtains the switching control coefficient of the automobile suspension under different working conditions according to automobile state data monitored in real time, and controls the damping force of the suspension shock absorber according to the switching control coefficient, thereby improving the safety and comfort of the automobile and realizing stable switching.
The technical scheme provided by the invention is as follows:
a suspension shock absorber damping control switching weighting coefficient determination method based on a neural network comprises the following steps:
collecting automobile monitoring data, establishing a seven-degree-of-freedom motion differential equation, and converting vertical displacement of an automobile body, a pitch angle of the automobile body, a roll angle and vertical displacement of front and rear wheels into sample data of a neural network;
classifying the sample data according to road condition types to obtain a sample data set corresponding to each road condition;
respectively establishing a neural network model according to a sample data set corresponding to each road surface working condition, wherein the neural network model comprises the following steps:
constructing a neural network by taking the vertical displacement of the vehicle body, the pitch angle of the vehicle body, the roll angle and the vertical displacement of the front and rear wheels as input layer vectors, and analyzing the input layer vector characteristics in the neural network to obtain a vector group representing a switching weighting coefficient corresponding to the road surface working condition;
fusing all the neural network models into a neural network; and
the road condition is identified according to the real-time monitoring data of the automobile, and a control equation containing the switching weighting coefficient is used as a control strategy to be output;
wherein the road condition categories include: the deceleration strip comprises a straight road working condition, a slope road working condition and a continuous deceleration strip.
Preferably, the kinematic differential equation is:
vertical motion equation of the left front wheel:
Figure GDA0002220758990000021
wherein m isu1Left front wheel unsprung mass, zflThe left front wheel is vertically displaced,
Figure GDA0002220758990000022
the vertical speed of the left front wheel is the speed,
Figure GDA0002220758990000023
is the vertical acceleration, k, of the left front wheeltFor tire stiffness, kfFor suspension stiffness, q1Vertical displacement excited for the road surface of the left front wheel, Fk1Damping force of the left front suspension shock absorber;
equation of vertical motion of the right front wheel:
Figure GDA0002220758990000024
wherein m isu2For the unsprung mass of the right front wheel, zfrIs vertically displaced for the right front wheel,
Figure GDA0002220758990000025
the vertical speed of the right front wheel is set,
Figure GDA0002220758990000026
is the vertical acceleration of the right front wheel, q2Vertical displacement excited for the road surface of the front right wheel, Fk2Damping force of the shock absorber of the right front suspension;
left rear wheel vertical equation of motion:
Figure GDA0002220758990000027
wherein m isu3For left rear wheel unsprung mass, zrlThe left rear wheel is vertically displaced,the vertical speed of the left rear wheel is set,vertical acceleration of the left rear wheel, q3Vertical displacement excited for left rear wheel road surface, Fk3Damping force of the left rear suspension shock absorber;
vertical motion equation of the right rear wheel:
Figure GDA0002220758990000031
wherein m isu4Is the unsprung mass of the right rear wheel, zrrIs used for the vertical displacement of the right rear wheel,
Figure GDA0002220758990000032
is the vertical speed of the right rear wheel,
Figure GDA0002220758990000033
is the vertical acceleration of the right rear wheel, q4Vertical displacement excited for the road surface of the right rear wheel, Fk4Damping force of the rear right suspension shock absorber;
zs1is the vertical displacement of the joint of the left front wheel body and the suspension, zs1＝zs-Lfθ+φB/2；
zs2Is the vertical displacement of the joint of the right front wheel body and the suspension, zs2＝zs-Lfθ-φB/2；
zs3Is the vertical displacement of the joint of the left rear wheel body and the suspension, zs3＝zs+Lrθ+φB/2；
zs4Is the vertical displacement of the joint of the right rear wheel body and the suspension, zs4＝zs+Lrθ-φB/2；
zsFor vertical displacement of the car body, LfIs the distance of the center of mass from the front suspension, LrThe distance between the center of mass and the right hinge point of the suspension is shown, phi is the roll angle of the vehicle body, theta is the pitch angle of the vehicle body, and B is the left and right wheel distance;
the vertical motion equation of the vehicle body is as follows:
the pitching motion equation of the vehicle body is as follows:
Figure GDA0002220758990000035
vehicle body roll equation of motion:
Figure GDA0002220758990000036
wherein the content of the first and second substances,
Figure GDA0002220758990000037
the vertical speed of the joint of the left front wheel vehicle body and the suspension,
Figure GDA0002220758990000038
is the vertical speed of the joint of the right front wheel body and the suspension,
Figure GDA0002220758990000039
the vertical speed of the joint of the left rear wheel vehicle body and the suspension,
Figure GDA00022207589900000310
is the vertical speed of the joint of the right rear wheel vehicle body and the suspension,
Figure GDA00022207589900000311
for the pitch angle rate of the vehicle body,
Figure GDA00022207589900000312
for vehicle body roll angular velocity, IxMoment of inertia in the roll direction of the vehicle body, IyIs the moment of inertia in the pitch direction of the vehicle body, CfDamping coefficient of front shock absorber of suspension CrDamping coefficient of rear shock absorber of suspension, Fk1Damping force of left front suspension shock absorber, Fk2Damping force for shock absorber of front right suspension, Fk3Damping force of left rear suspension shock absorber, Fk4And damping force of the rear right suspension shock absorber.
Preferably, the neural network is a three-layer neural network model, and the input layer vectors are formatted in sequence to determine the input layer vectors of the three-layer neural network
Figure GDA00022207589900000313
The input layer vector is mapped to a hidden layer, and the hidden layer vector is Y ═ Y1,y2,y3,y4···ymM is the number of nodes, and the output layer vector is
Figure GDA00022207589900000314
Wherein z issIs a vector of the vertical displacement of the vehicle body,
Figure GDA0002220758990000041
is the vertical speed of the vehicle body, theta is the pitch angle of the vehicle body,
Figure GDA0002220758990000042
is the pitch angle speed of the vehicle body, phi is the roll angle of the vehicle body,
Figure GDA0002220758990000043
is the vehicle body roll angle velocity, zflThe front wheel is vertically displaced,is the vertical speed of the front wheel, zrlIn order to make the rear wheel vertically move,
Figure GDA0002220758990000045
is the rear wheel vertical speed; ladma1Switching the weighting coefficient matrix, ladma, for straight road conditions2For switching weighting coefficient matrix, ladma, for gradient road conditions3A weighting coefficient matrix is switched for the road condition of the continuous deceleration strip,
Figure GDA0002220758990000046
the characteristic weighting coefficient is switched for the left front wheel suspension shock absorber corresponding to the working condition,
Figure GDA0002220758990000047
the characteristic weighting coefficients are switched for the right front wheel suspension shock absorber corresponding to the operating conditions,
Figure GDA0002220758990000048
the weighting coefficients are switched for the left rear wheel suspension shock absorbers corresponding to the working conditions,
Figure GDA0002220758990000049
and switching the weighting coefficients for the right rear wheel suspension shock absorbers in the corresponding working conditions.
Preferably, the input layer parameters are formatted using the following formula:
Figure GDA00022207589900000410
wherein x isiFor the index coefficient after formatting, TiAs an input layer parameter, TimaxFor input layer parameters corresponding to a maximum value, TiminThe input layer parameters correspond to minimum values.
Preferably, the number of hidden layer nodes is 13.
Preferably, the control equation of the switching weight coefficient is:
Figure GDA00022207589900000411
wherein, Fk1Damping force of left front suspension shock absorber, Fk2Damping force for shock absorber of front right suspension, Fk3Damping force of left rear suspension shock absorber, Fk4The damping force of the shock absorber of the rear right suspension,
Figure GDA00022207589900000412
the damping force of the left front wheel suspension shock absorber under the straight road working condition,
Figure GDA00022207589900000413
the damping force of the left front wheel suspension shock absorber under the working condition of the slope road surface,
Figure GDA00022207589900000414
the damping force of the left front wheel suspension shock absorber under the working condition of the continuous speed bump,
Figure GDA00022207589900000415
for the damping force of the shock absorber of the right front wheel suspension under the straight road working condition,
Figure GDA00022207589900000416
the damping force of the shock absorber of the right front wheel suspension under the working condition of the slope road surface,
Figure GDA00022207589900000417
the damping force of the left front wheel suspension shock absorber under the working condition of the continuous speed bump is adopted;
Figure GDA00022207589900000418
the damping force of the left rear wheel suspension shock absorber under the straight road working condition,
Figure GDA00022207589900000419
the damping force of the left rear wheel suspension shock absorber under the working condition of the slope road surface,
Figure GDA00022207589900000420
the damping force of the left rear wheel suspension shock absorber is under the working condition of the continuous speed bump;
Figure GDA00022207589900000421
the damping force of the shock absorber of the right rear wheel suspension under the condition of straight road work,
Figure GDA00022207589900000422
the damping force of the rear right wheel suspension shock absorber under the working condition of the slope road surface,
Figure GDA00022207589900000423
the damping force of the left rear wheel suspension shock absorber is under the working condition of the continuous speed bump.
The invention has the advantages of
The invention designs and develops a method for determining a switching control characteristic weighting coefficient of a neural network, which is used for establishing a neural network model aiming at different working conditions, obtaining control coefficients of automobile suspensions under different working conditions according to real-time monitored automobile state data, controlling the damping force of a suspension shock absorber according to the control coefficients, improving the safety and the comfort of an automobile, realizing stable switching, and realizing smooth switching of active suspension multi-working condition control by adopting a neural network algorithm, wherein when the working conditions of a road surface change, the switching control force can generate phenomena of blocking, mutation and the like.
Drawings
FIG. 1 is a seven-degree-of-freedom vehicle dynamics model diagram according to the present invention.
Fig. 2 is a schematic diagram of a neural network model according to the present invention.
FIG. 3 is a diagram of a multi-condition road surface simulation result according to the present invention.
Fig. 4 is a straight road condition switching characteristic weighting coefficient according to the present invention.
Fig. 5 shows the weighting coefficients of the feature of switching between road conditions on a slope according to the present invention.
Fig. 6 shows the road condition switching characteristic weighting coefficients of the continuous deceleration strip according to the present invention.
Detailed Description
The present invention is further described in detail below with reference to the attached drawings so that those skilled in the art can implement the invention by referring to the description text.
As shown in FIG. 1, the method for determining the damping control switching characteristic weighting coefficient of the suspension shock absorber based on the neural network, provided by the invention, comprises the following steps: establishing a seven-degree-of-freedom motion differential equation which comprises vertical displacement of a vehicle body, a pitch angle of the vehicle body, a roll angle and vertical displacement of front and rear four wheels and has nonlinear characteristics:
vertical motion equation of the left front wheel:
Figure GDA0002220758990000051
wherein m isu1Left front wheel unsprung mass, zflIs a left front wheel plumbThe displacement is carried out towards the direction of the displacement,
Figure GDA0002220758990000052
the vertical speed of the left front wheel is the speed,
Figure GDA0002220758990000053
is the vertical acceleration, k, of the left front wheeltFor tire stiffness, kfFor suspension stiffness, q1Vertical displacement excited for the road surface of the left front wheel, Fk1Damping force of the left front suspension shock absorber;
equation of vertical motion of the right front wheel:
Figure GDA0002220758990000054
wherein m isu2For the unsprung mass of the right front wheel, zfrIs vertically displaced for the right front wheel,
Figure GDA0002220758990000055
the vertical speed of the right front wheel is set,
Figure GDA0002220758990000061
is the vertical acceleration of the right front wheel, q2Vertical displacement excited for the road surface of the front right wheel, Fk2Damping force of the shock absorber of the right front suspension;
left rear wheel vertical equation of motion:
wherein m isu3For left rear wheel unsprung mass, zrlThe left rear wheel is vertically displaced,
Figure GDA0002220758990000063
the vertical speed of the left rear wheel is set,
Figure GDA0002220758990000064
vertical acceleration of the left rear wheel, q3Vertical displacement excited for left rear wheel road surface, Fk3Left rear suspension dampingA damper damping force;
vertical motion equation of the right rear wheel:
Figure GDA0002220758990000065
wherein m isu4Is the unsprung mass of the right rear wheel, zrrIs used for the vertical displacement of the right rear wheel,
Figure GDA0002220758990000066
is the vertical speed of the right rear wheel,is the vertical acceleration of the right rear wheel, q4Vertical displacement excited for the road surface of the right rear wheel, Fk4Damping force of the rear right suspension shock absorber;
zs1is the vertical displacement of the joint of the left front wheel body and the suspension, zs1＝zs-Lfθ+φB/2；
zs2Is the vertical displacement of the joint of the right front wheel body and the suspension, zs2＝zs-Lfθ-φB/2；
zs3Is the vertical displacement of the joint of the left rear wheel body and the suspension, zs3＝zs+Lrθ+φB/2；
zs4Is the vertical displacement of the joint of the right rear wheel body and the suspension, zs4＝zs+Lrθ-φB/2；
zsFor vertical displacement of the car body, LfIs the distance of the center of mass from the front suspension, LrThe distance between the center of mass and the right hinge point of the suspension is shown, phi is the roll angle of the vehicle body, theta is the pitch angle of the vehicle body, and B is the left and right wheel distance;
the vertical motion equation of the vehicle body is as follows:
Figure GDA0002220758990000068
the pitching motion equation of the vehicle body is as follows:
Figure GDA0002220758990000069
vehicle body roll equation of motion:
Figure GDA00022207589900000610
wherein the content of the first and second substances,
Figure GDA00022207589900000611
the vertical speed of the joint of the left front wheel vehicle body and the suspension,
Figure GDA00022207589900000612
is the vertical speed of the joint of the right front wheel body and the suspension,
Figure GDA00022207589900000613
the vertical speed of the joint of the left rear wheel vehicle body and the suspension,
Figure GDA00022207589900000614
is the vertical speed of the joint of the right rear wheel vehicle body and the suspension,for the pitch angle rate of the vehicle body,
Figure GDA00022207589900000616
for vehicle body roll angular velocity, IxMoment of inertia in the roll direction of the vehicle body, IyIs the moment of inertia in the pitch direction of the vehicle body, CfDamping coefficient of front shock absorber of suspension CrDamping coefficient of rear shock absorber of suspension, Fk1Damping force of left front suspension shock absorber, Fk2Damping force for shock absorber of front right suspension, Fk3Damping force of left rear suspension shock absorber, Fk4And damping force of the rear right suspension shock absorber.
Collecting automobile monitoring data and classifying the sample data according to road condition types to obtain a sample data set corresponding to each road condition; respectively establishing a neural network model according to a sample data set corresponding to each road surface working condition, wherein the road condition types at least comprise: at least three neural network models are established according to three categories of straight road conditions, slope road conditions and continuous deceleration strips, the training process of one neural network model is given below, and the neural network model for straight road conditions is taken as an example:
as shown in fig. 2, the neural network module adopts a BP neural network to apply a state response vector of a seven-degree-of-freedom vehicle model
Figure GDA0002220758990000071
As training samples, input from the input layer and output as
Figure GDA0002220758990000072
Is the actual output vector;
zsis a vector of the vertical displacement of the vehicle body,
Figure GDA0002220758990000073
is the vertical speed of the vehicle body, theta is the pitch angle of the vehicle body,
Figure GDA0002220758990000074
is the pitch angle speed of the vehicle body, phi is the roll angle of the vehicle body,
Figure GDA0002220758990000075
is the vehicle body roll angle velocity, zflThe front wheel is vertically displaced,
Figure GDA0002220758990000076
is the vertical speed of the front wheel, zrlIn order to make the rear wheel vertically move,
Figure GDA0002220758990000077
is the rear wheel vertical speed.
The input layer parameters are formatted using the following formula:
Figure GDA0002220758990000078
wherein x isiIs the index coefficient after formatting，TiFor the input layer parameters, the input parameters include:Timaxis an input layer parameter
Figure GDA00022207589900000710
Corresponding to the maximum value, TiminIs an input layer parameter
Figure GDA00022207589900000711
Corresponding to the minimum value. z is a radical ofsIs a vector of the vertical displacement of the vehicle body,
Figure GDA00022207589900000712
is the vertical speed of the vehicle body, theta is the pitch angle of the vehicle body,
Figure GDA00022207589900000713
is the pitch angle speed of the vehicle body, phi is the roll angle of the vehicle body,is the vehicle body roll angle velocity, zflThe front wheel is vertically displaced,
Figure GDA00022207589900000715
is the vertical speed of the front wheel, zrlIn order to make the rear wheel vertically move,
Figure GDA00022207589900000716
is the rear wheel vertical speed.
Carrying out BP neural network training: after the BP neural network node model is established, the training of the BP neural network can be carried out. Obtaining training samples according to empirical data of the product, and giving a connection weight w between an input node i and a hidden layer node jijConnection weight w between hidden layer node j and output layer node kjkThreshold value theta of hidden layer node jjThreshold value w of node k of output layerij、wjk、θj、θkAre all random numbers between-1 and 1.
Continuously correcting w in the training processijAnd wjkUntil the system error is less than or equal to the expected error, the training process of the neural network is completed.
Wherein, delta is a training error,
Figure GDA0002220758990000081
for the output layer vector, n is 1,2,3,4,
Figure GDA0002220758990000082
are sample coefficients.
The learning rule of the system is to adjust the connection weight and the threshold vector by making the sum of the squares of the errors of the expected output and the actual output reach a certain set value. When the error is reduced to the set value, the system stops learning, and the weight value and the threshold value at the moment are kept in the system and become the system internal knowledge.
The number of nodes of the hidden layer is 13, the number of the activation functions is a symmetrical saturated linear transfer function, the number of the nodes of the output layer is 3, the activation functions are log-S type transfer functions, and the training method is selected as a self-learning applicability algorithm.
For the hidden layer, the activation function f [. cndot. ] adopts a symmetrical saturated linear transfer function (satlins)
Figure GDA0002220758990000083
For the output layer, the activation function f [. cndot. ] uses a log-S type transfer function,
Figure GDA0002220758990000084
IW {1,1} and LW {2,1} are connection weights between layers, and b {1,1} and b {2,1} are weights.
As shown in table 1, a set of training samples is given, along with the values of the nodes in the training process.
TABLE 1 training Process node values
Figure GDA0002220758990000085
Figure GDA0002220758990000101
Figure GDA0002220758990000111
Obtaining three categories of neural network models of a straight road working condition, a slope road working condition and a continuous deceleration strip according to the training method, and integrating the three neural network models into one neural network model
The output layer vector is
Figure GDA0002220758990000112
Wherein, ladma1Switching the weighting coefficient matrix, ladma, for straight road conditions2For switching weighting coefficient matrix, ladma, for gradient road conditions3A weighting coefficient matrix is switched for the road condition of the continuous deceleration strip,
Figure GDA0002220758990000113
to switch the weighting coefficients for the left front wheel using the operating conditions,
Figure GDA0002220758990000114
the weighting coefficients are switched for the right front wheel corresponding to the working condition,
Figure GDA0002220758990000115
the weighting coefficients are switched for the left rear wheel corresponding to the working condition,and switching the weighting coefficients for the right rear wheel of the corresponding working condition.
Obtaining a product comprising lamda1、lamda2、lamda3Is control of the characteristic weighting coefficients of the actual output vectorThe preparation process comprises the following steps:
Figure GDA0002220758990000117
wherein, Fk1Damping force of left front suspension shock absorber, Fk2Damping force for shock absorber of front right suspension, Fk3Damping force of left rear suspension shock absorber, Fk4Damping force of right rear suspension shock absorber, F10Damping force of the left front suspension shock absorber for straight road working conditions, F20Damping force of left front suspension shock absorber under working condition of slope road surface, F30The damping force of the left front suspension shock absorber is the damping force under the working condition of the continuous deceleration strip.
Examples of the experiments
As shown in FIG. 3, a multi-condition road surface simulation model of a random straight road, a slope road and a continuous deceleration strip is established for 0-15 s. Wherein 0-5s is the working condition of the random straight road, 5-10s is the working condition of the slope road surface, and 10-15s is the working condition of the continuous acceleration belt.
The working condition of a 0-5s random straight road is described by adopting a filtering white noise road surface model, and the motion equation is as follows:
wherein G is06.4e-6 is the road surface roughness coefficient, w (t) is the white gaussian noise input with the mean value at time t being zero, v 20m/s is the vehicle speed, f00.2Hz is the lower cut-off frequency.
A module with the gradient of 3% is added on the basis of a random flat road established by filtering white noise under the working condition of a road surface with the gradient of 5-10 s.
In the working condition of a 10-15s continuous speed bump, a gradient elimination module with the gradient of-3% is added at the end of 10s, and a pulse module is added at the end of 11s, wherein the pulse module describes that the height of the pulse module is 0.04m, the width of the pulse module is 0.3m, and the interval between two speed bumps is 2 m.
Collecting automobile monitoring data, classifying the sample data based on an automobile motion equation with seven degrees of freedom and simulating road condition types to obtain a sample data set corresponding to each road surface working condition
Figure GDA0002220758990000121
The training samples are input from the input layer of the neural network, and the learning rule of the system is to make the sum of the squares of the errors of the expected output and the actual output reach a certain set value so as to adjust the connection weight and the threshold vector. When the error is reduced to the set value, the system stops learning, and the weight value and the threshold value at the moment are kept in the system and become the system internal knowledge. As a preference, when training errors
Figure GDA0002220758990000122
When so, finishing the training; wherein, delta is a training error,
Figure GDA0002220758990000123
for the output layer vector, n is 1,2,3,4,
Figure GDA0002220758990000124
for the corresponding sample coefficient K as the threshold, lamda1、lamda2、lamda3Is the actual output vector.
The output layer vector is
Figure GDA0002220758990000125
As shown in fig. 4-6, a neural network algorithm is adopted to obtain three characteristic weighting coefficients corresponding to three road surface working conditions, mean filtering is performed on the three weighting coefficients respectively, the change of the changing time of the road surface working conditions is kept, the values of the rest time periods are subjected to filtering processing, and filtering is started from the 2 nd second, filtering is performed on lambda 1, and the change of 0.05 second before and after 7 seconds is kept; filtering lamda2, keeping the change of 0.05 second before and after 7 seconds and 12 seconds; lamda3 was filtered, retaining the 0.05 second change before and after 12 seconds. After filtering, three weighting coefficients are obtained.
Obtaining a product comprising lamda1、lamda2、lamda3The control equation for the weighting coefficients, which is the actual output vector, includes:
Figure GDA0002220758990000126
wherein, Fk1Damping force of left front suspension shock absorber, Fk2Damping force for shock absorber of front right suspension, Fk3Damping force of left rear suspension shock absorber, Fk4The damping force of the shock absorber of the rear right suspension,the vertical loading force of the left front wheel suspension spring under the straight road working condition,
Figure GDA0002220758990000128
the damping force of the left front wheel suspension shock absorber under the working condition of the slope road surface,
Figure GDA0002220758990000129
the damping force of the left front wheel suspension shock absorber under the working condition of the continuous speed bump,
Figure GDA00022207589900001210
for the damping force of the shock absorber of the right front wheel suspension under the straight road working condition,
Figure GDA00022207589900001211
the damping force of the shock absorber of the right front wheel suspension under the working condition of the slope road surface,
Figure GDA00022207589900001212
the damping force of the left front wheel suspension shock absorber under the working condition of the continuous speed bump is adopted;
Figure GDA00022207589900001213
the damping force of the left rear wheel suspension shock absorber under the straight road working condition,
Figure GDA0002220758990000131
the damping force of the left rear wheel suspension shock absorber under the working condition of the slope road surface,
Figure GDA0002220758990000132
for the shock absorber of the left rear wheel suspension under the working condition of the continuous deceleration stripDamping force;
Figure GDA0002220758990000133
the damping force of the shock absorber of the right rear wheel suspension under the condition of straight road work,
Figure GDA0002220758990000134
the damping force of the rear right wheel suspension shock absorber under the working condition of the slope road surface,the damping force of the left rear wheel suspension shock absorber is under the working condition of the continuous speed bump.
Therefore, the neural network algorithm is adopted to control the active suspension in different control modes according to different working conditions, the neural network model is established according to different working conditions, the control coefficients of the automobile suspensions under different working conditions are obtained according to the automobile state data monitored in real time, the damping force of the suspension shock absorber is controlled according to the control coefficients, the safety and the comfort of an automobile are improved, the phenomena of blocking, sudden change and the like of the switching control force can occur when the road surface working conditions are changed in stable switching, and the neural network algorithm is adopted to realize the smooth switching of the multi-working-condition control of the active suspension.
While embodiments of the invention have been described above, it is not limited to the applications set forth in the description and the embodiments, which are fully applicable in various fields of endeavor to which the invention pertains, and further modifications may readily be made by those skilled in the art, it being understood that the invention is not limited to the details shown and described herein without departing from the general concept defined by the appended claims and their equivalents.

Claims (6)

1. A suspension shock absorber damping control switching weighting coefficient determination method based on a neural network is characterized by comprising the following steps:
collecting automobile monitoring data, establishing a seven-degree-of-freedom motion differential equation, and converting vertical displacement of an automobile body, a pitch angle of the automobile body, a roll angle and vertical displacement of front and rear wheels into sample data of a neural network;
classifying the sample data according to road condition types to obtain a sample data set corresponding to each road condition;
respectively establishing a neural network model according to a sample data set corresponding to each road surface working condition, wherein the neural network model comprises the following steps:
constructing a neural network by taking the vertical displacement of the vehicle body, the pitch angle of the vehicle body, the roll angle and the vertical displacement of the front and rear wheels as input layer vectors, and analyzing the input layer vector characteristics in the neural network to obtain a vector group representing a switching weighting coefficient corresponding to the road surface working condition;
fusing all the neural network models into a neural network; and
the road condition is identified according to the real-time monitoring data of the automobile, and a control equation containing the switching weighting coefficient is used as a control strategy to be output;
wherein the road condition categories include: the working conditions of a straight road, a slope road surface and a continuous deceleration strip.
2. The neural network-based suspension shock absorber damping control switching weighting coefficient determining method as claimed in claim 1, wherein said differential equation of motion is:
vertical motion equation of the left front wheel:
Figure FDA0002220758980000011
wherein m isu1Left front wheel unsprung mass, zflThe left front wheel is vertically displaced,
Figure FDA0002220758980000012
the vertical speed of the left front wheel is the speed,
Figure FDA0002220758980000013
is the vertical acceleration, k, of the left front wheeltFor tire stiffness, kfFor suspension stiffness, q1Vertical displacement excited for the road surface of the left front wheel, Fk1Damping force of the left front suspension shock absorber;
equation of vertical motion of the right front wheel:
Figure FDA0002220758980000014
wherein m isu2For the unsprung mass of the right front wheel, zfrIs vertically displaced for the right front wheel,
Figure FDA0002220758980000015
the vertical speed of the right front wheel is set,
Figure FDA0002220758980000016
is the vertical acceleration of the right front wheel, q2Vertical displacement excited for the road surface of the front right wheel, Fk2Damping force of the shock absorber of the right front suspension;
left rear wheel vertical equation of motion:
Figure FDA0002220758980000021
wherein m isu3For left rear wheel unsprung mass, zrlThe left rear wheel is vertically displaced,
Figure FDA0002220758980000022
the vertical speed of the left rear wheel is set,
Figure FDA0002220758980000023
vertical acceleration of the left rear wheel, q3Vertical displacement excited for left rear wheel road surface, Fk3Damping force of the left rear suspension shock absorber;
vertical motion equation of the right rear wheel:
Figure FDA0002220758980000024
wherein m isu4Is the unsprung mass of the right rear wheel, zrrIs used for the vertical displacement of the right rear wheel,
Figure FDA0002220758980000025
is the vertical speed of the right rear wheel,
Figure FDA0002220758980000026
is the vertical acceleration of the right rear wheel, q4Vertical displacement excited for the road surface of the right rear wheel, Fk4Damping force of the rear right suspension shock absorber;
zs1is the vertical displacement of the joint of the left front wheel body and the suspension, zs1＝zs-Lfθ+φB/2；
zs2Is the vertical displacement of the joint of the right front wheel body and the suspension, zs2＝zs-Lfθ-φB/2；
zs3Is the vertical displacement of the joint of the left rear wheel body and the suspension, zs3＝zs+Lrθ+φB/2；
zs4Is the vertical displacement of the joint of the right rear wheel body and the suspension, zs4＝zs+Lrθ-φB/2；
zsFor vertical displacement of the car body, LfIs the distance of the center of mass from the front suspension, LrThe distance between the center of mass and the right hinge point of the suspension is shown, phi is the roll angle of the vehicle body, theta is the pitch angle of the vehicle body, and B is the left and right wheel distance;
the vertical motion equation of the vehicle body is as follows:
Figure FDA0002220758980000027
the pitching motion equation of the vehicle body is as follows:
Figure FDA0002220758980000028
vehicle body roll equation of motion:
Figure FDA0002220758980000029
wherein the content of the first and second substances,
Figure FDA00022207589800000210
the vertical speed of the joint of the left front wheel vehicle body and the suspension,
Figure FDA00022207589800000211
is the vertical speed of the joint of the right front wheel body and the suspension,
Figure FDA00022207589800000212
the vertical speed of the joint of the left rear wheel vehicle body and the suspension,
Figure FDA00022207589800000213
is the vertical speed of the joint of the right rear wheel vehicle body and the suspension,
Figure FDA00022207589800000214
for the pitch angle rate of the vehicle body,
Figure FDA00022207589800000215
for vehicle body roll angular velocity, IxMoment of inertia in the roll direction of the vehicle body, IyIs the moment of inertia in the pitch direction of the vehicle body, CfDamping coefficient of front shock absorber of suspension CrDamping coefficient of rear shock absorber of suspension, Fk1Damping force of left front suspension shock absorber, Fk2Damping force for shock absorber of front right suspension, Fk3Damping force of left rear suspension shock absorber, Fk4And damping force of the rear right suspension shock absorber.
3. The method for determining damping control switching weighting coefficient of suspension shock absorber based on neural network as claimed in claim 2, wherein said neural network is a three-layer neural network model, and the input layer vectors are formatted in sequence to determine the input layer vectors of the three-layer neural network
Figure FDA0002220758980000031
The input layer vector is mapped to a hidden layer, and the hidden layer vector is Y ═{y1,y2,y3,y4…ymM is the number of nodes, and the output layer vector is
Figure FDA0002220758980000032
Figure FDA0002220758980000033
Wherein z issIs a vector of the vertical displacement of the vehicle body,is the vertical speed of the vehicle body, theta is the pitch angle of the vehicle body,
Figure FDA0002220758980000035
is the pitch angle speed of the vehicle body, phi is the roll angle of the vehicle body,
Figure FDA0002220758980000036
is the vehicle body roll angle velocity, zflThe front wheel is vertically displaced,
Figure FDA0002220758980000037
is the vertical speed of the front wheel, zrlIn order to make the rear wheel vertically move,
Figure FDA0002220758980000038
is the rear wheel vertical speed; ladma1Switching the weighting coefficient matrix, ladma, for straight road conditions2For switching weighting coefficient matrix, ladma, for gradient road conditions3A weighting coefficient matrix is switched for the road condition of the continuous deceleration strip,
Figure FDA0002220758980000039
the characteristic weighting coefficient is switched for the left front wheel suspension shock absorber corresponding to the working condition,
Figure FDA00022207589800000310
for right front wheel suspension shock absorber switching characteristics corresponding to operating conditions plusThe weight coefficient is a function of the weight coefficient,
Figure FDA00022207589800000311
the weighting coefficients are switched for the left rear wheel suspension shock absorbers corresponding to the working conditions,
Figure FDA00022207589800000312
and switching the weighting coefficients for the right rear wheel suspension shock absorbers in the corresponding working conditions.
4. The neural network-based suspension shock absorber damping control switching weighting coefficient determining method as claimed in claim 3, wherein the input layer parameters are formatted using the following formula:
Figure FDA00022207589800000313
wherein x isiFor the index coefficient after formatting, TiAs an input layer parameter, TimaxFor input layer parameters corresponding to a maximum value, TiminThe input layer parameters correspond to minimum values.
5. The neural network-based suspension shock absorber damping control switching weighting coefficient determining method as claimed in claim 3, wherein the number of hidden layer nodes is 13.
6. The neural network-based suspension shock absorber damping control switching weighting coefficient determining method as claimed in claim 5, wherein the control equation of the switching weighting coefficient is:
Figure FDA0002220758980000041
wherein, Fk1Damping force of left front suspension shock absorber, Fk2Damping force for shock absorber of front right suspension, Fk3Damping force of left rear suspension shock absorber, Fk4The damping force of the shock absorber of the rear right suspension,the damping force of the left front wheel suspension shock absorber under the straight road working condition,the damping force of the left front wheel suspension shock absorber under the working condition of the slope road surface,
Figure FDA0002220758980000044
the damping force of the left front wheel suspension shock absorber under the working condition of the continuous speed bump,for the damping force of the shock absorber of the right front wheel suspension under the straight road working condition,
Figure FDA0002220758980000046
the damping force of the shock absorber of the right front wheel suspension under the working condition of the slope road surface,
Figure FDA0002220758980000047
the damping force of the left front wheel suspension shock absorber under the working condition of the continuous speed bump is adopted;
Figure FDA0002220758980000048
the damping force of the left rear wheel suspension shock absorber under the straight road working condition,the damping force of the left rear wheel suspension shock absorber under the working condition of the slope road surface,
Figure FDA00022207589800000410
the damping force of the left rear wheel suspension shock absorber is under the working condition of the continuous speed bump;the damping force of the shock absorber of the right rear wheel suspension under the condition of straight road work,
Figure FDA00022207589800000412
the damping force of the rear right wheel suspension shock absorber under the working condition of the slope road surface,
Figure FDA00022207589800000413
the damping force of the left rear wheel suspension shock absorber is under the working condition of the continuous speed bump.
CN201810933191.5A 2018-08-16 2018-08-16 Suspension shock absorber damping control switching weighting coefficient determination method Active CN108859648B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201810933191.5A CN108859648B (en) 2018-08-16 2018-08-16 Suspension shock absorber damping control switching weighting coefficient determination method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201810933191.5A CN108859648B (en) 2018-08-16 2018-08-16 Suspension shock absorber damping control switching weighting coefficient determination method

Publications (2)

Publication Number Publication Date
CN108859648A CN108859648A (en) 2018-11-23
CN108859648B true CN108859648B (en) 2020-02-18

Family

ID=64319227

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201810933191.5A Active CN108859648B (en) 2018-08-16 2018-08-16 Suspension shock absorber damping control switching weighting coefficient determination method

Country Status (1)

Country Link
CN (1) CN108859648B (en)

Families Citing this family (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110020499A (en) * 2019-04-19 2019-07-16 南京航空航天大学 A kind of road surface fluctuating disturbance identification method based on inverse Wilson method
JP7189514B2 (en) * 2020-06-02 2022-12-14 トヨタ自動車株式会社 Damping control device and damping control method
CN113752770B (en) * 2020-06-03 2023-09-15 观致汽车有限公司 Semi-active suspension control system and method, computer storage medium, and vehicle
CN113183707B (en) * 2021-05-21 2024-04-02 辽宁工业大学 Design method and system of self-adaptive fixed-period sampling controller of full-vehicle active suspension
CN113183709B (en) * 2021-06-04 2022-09-27 合肥工业大学 Preview control method for automobile electric control suspension
CN113570057B (en) * 2021-09-27 2021-12-14 岚图汽车科技有限公司 Vehicle wheel center vertical displacement measuring method and device based on model training

Family Cites Families (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
DE19527323A1 (en) * 1995-07-26 1997-01-30 Siemens Ag Circuit arrangement for controlling a device in a motor vehicle
US6212466B1 (en) * 2000-01-18 2001-04-03 Yamaha Hatsudoki Kabushiki Kaisha Optimization control method for shock absorber
CN101269618B (en) * 2008-04-25 2010-06-02 江苏大学 Control method for damping value of electronic control air spring damping apparatus with three-gear
CN101367324B (en) * 2008-10-15 2011-01-12 江苏大学 Pavement grade prediction technique based on electronic control air spring vehicle altimetric sensor
CN108274969A (en) * 2018-02-27 2018-07-13 江苏大学 A kind of feed energy suspension and its damping method for handover control based on speed subregion

Also Published As

Publication number Publication date
CN108859648A (en) 2018-11-23

Similar Documents

Publication Publication Date Title
CN108859648B (en) Suspension shock absorber damping control switching weighting coefficient determination method
CN106985627B (en) Vehicle road surface recognition system and suspension mode switching method
Yoshimura et al. Steering and suspension system of a full car model using fuzzy reasoning based on single input rule modules
CN105539052B (en) A kind of controllable suspension sliding formwork tracking controller using vehicle plateau as reference
CN108891220B (en) Ceiling control improvement method for semi-active suspension system of vehicle
CN110712490B (en) Active suspension system based on stack type self-coding and working method thereof
CN102501737B (en) Intelligent particle swarm fuzzy hybrid control method for automotive semi-active suspension systems
CN109591537A (en) A kind of automotive semi-active suspension control system and method
CN109334378B (en) Vehicle ISD suspension active control method based on single neuron PID control
CN102975587B (en) Vehicle semiactive suspension based on double controllable dampers and control method thereof
CN113183709A (en) Automobile electric control suspension pre-aiming control method
JPH04163220A (en) Vehicular suspension device
WO2023174327A1 (en) Cooperative control module, adaptive cruise control system as well as control method thereof, and vehicle
CN113591360A (en) Magnetorheological damper structure parameter optimization method based on complete vehicle dynamic model
CN112906133A (en) Vertical vibration negative effect suppression method for movable inertial suspension controlled by ground shed
CN110443003B (en) Control and optimal design method of active stabilizer bar system
CN115963836A (en) Path tracking and vehicle body posture cooperative control method
Liu et al. Study of ride comfort of active suspension based on model reference neural network control system
CN110765554A (en) Intelligent control method of automobile semi-active suspension system based on TS model
CN112948979A (en) Optimal Bingham numerical control method for impact working condition of automobile electric control damping suspension system
CN112590483A (en) Observer-based automobile lateral stability and active suspension multi-target integrated control method
CN115782496B (en) Intelligent evolution method of semi-active suspension system based on MAP control
CN113071282B (en) Vehicle adjusting method and device
CN117841591B (en) ISD suspension control method based on improved fuzzy neural network PID
Mohan et al. Comfort Parameters Tuning Analysis for Vehicle Suspension Pitch Performance

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant