CN117331312A - Multi-depth neural fuzzy network control method and system for semi-active suspension of whole vehicle - Google Patents
Multi-depth neural fuzzy network control method and system for semi-active suspension of whole vehicle Download PDFInfo
- Publication number
- CN117331312A CN117331312A CN202311243373.7A CN202311243373A CN117331312A CN 117331312 A CN117331312 A CN 117331312A CN 202311243373 A CN202311243373 A CN 202311243373A CN 117331312 A CN117331312 A CN 117331312A
- Authority
- CN
- China
- Prior art keywords
- difference value
- control
- speed
- change rate
- speed difference
- 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.)
- Pending
Links
- 230000001537 neural effect Effects 0.000 title claims abstract description 71
- 239000000725 suspension Substances 0.000 title claims abstract description 57
- 238000000034 method Methods 0.000 title claims abstract description 51
- 230000008859 change Effects 0.000 claims abstract description 90
- 239000006096 absorbing agent Substances 0.000 claims description 35
- 230000035939 shock Effects 0.000 claims description 25
- 230000006870 function Effects 0.000 claims description 16
- 238000005096 rolling process Methods 0.000 claims description 13
- 230000003044 adaptive effect Effects 0.000 claims description 6
- 230000000694 effects Effects 0.000 abstract description 10
- 230000001133 acceleration Effects 0.000 description 27
- 230000005284 excitation Effects 0.000 description 14
- 238000011156 evaluation Methods 0.000 description 9
- 230000004044 response Effects 0.000 description 8
- 238000012360 testing method Methods 0.000 description 7
- 238000013461 design Methods 0.000 description 6
- 230000009467 reduction Effects 0.000 description 4
- 238000012549 training Methods 0.000 description 4
- 238000010276 construction Methods 0.000 description 3
- 238000011217 control strategy Methods 0.000 description 3
- 238000013016 damping Methods 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 3
- 238000006073 displacement reaction Methods 0.000 description 3
- 239000011159 matrix material Substances 0.000 description 3
- 238000004088 simulation Methods 0.000 description 3
- 239000013598 vector Substances 0.000 description 3
- 230000006978 adaptation Effects 0.000 description 2
- 230000009286 beneficial effect Effects 0.000 description 2
- 230000008878 coupling Effects 0.000 description 2
- 238000010168 coupling process Methods 0.000 description 2
- 238000005859 coupling reaction Methods 0.000 description 2
- 238000011161 development Methods 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 238000013139 quantization Methods 0.000 description 2
- 238000013528 artificial neural network Methods 0.000 description 1
- 230000000903 blocking effect Effects 0.000 description 1
- 230000001427 coherent effect Effects 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 230000007704 transition Effects 0.000 description 1
- 238000012795 verification Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Vehicle Body Suspensions (AREA)
Abstract
The invention discloses a control method and a system of a full-vehicle semi-active suspension multiple depth neural fuzzy network, which are used for carrying out fuzzy reasoning on a speed difference value and a difference value change rate of each direction through a variable domain module of each direction to determine an input variable expansion factor and an output variable expansion factor of each direction; the speed difference value and the difference value change rate of the speed difference value in each direction are adjusted through the input variable expansion factors in each direction, and the speed difference value and the difference value change rate of the speed difference value after adjustment in each direction are obtained; carrying out fuzzy reasoning on the speed difference value and the difference value change rate thereof after each direction is adjusted through a main controller module of each direction, and determining theoretical control force or theoretical control moment of each direction; the theoretical control force or the theoretical control moment in each direction is adjusted through the output variable expansion factors in each direction, and the control force or the control moment in each direction is determined; and controlling the whole vehicle according to the control force or the control moment in each direction. The method has the optimal control effect.
Description
Technical Field
The invention relates to the technical field of automobile electronic control, in particular to a method and a system for controlling a multi-depth neural fuzzy network of a semi-active suspension of a whole automobile.
Background
The statements in this section merely provide background information related to the present disclosure and may not necessarily constitute prior art.
The main function of the suspension system is to transmit forces and moments between the wheels and the frame, absorb vibrational responses caused by road irregularities, and thereby improve ride comfort and steering stability of the vehicle. The semi-active suspension has the advantages of both passive and active suspensions, the damping force can be adjusted in real time according to the driving working condition, the damping effect similar to that of the active suspension can be achieved, the structure is simple, a power source is not needed, and the semi-active suspension is an important direction of future development of the intelligent suspension. The semi-active suspension of the vehicle is a complex vibration system, and to fully embody the vertical vibration, pitching change, rolling motion and mutual coupling influence of the vehicle and comprehensively grasp the comprehensive problem of motion response and control of the vehicle, a whole vehicle model is needed.
When the semi-active suspension of the whole vehicle is controlled, the speed difference value and the change rate of the speed difference value in each direction are mainly obtained through a fuzzy control strategy to infer, and the control force in each direction is determined; however, the conventional fuzzy control strategy has the problems of large difficulty in determining the fuzzy rules, imperfect rules and excessive dependence on expert experience, and the control accuracy is too low. In order to realize the adjustment of the fuzzy domain of the fuzzy controller and improve the control precision of the system, some scholars propose the idea of fuzzy control of the variable domain, however, the input variable expansion factor and the output variable expansion factor in the traditional variable domain fuzzy controller are designed by adopting functional expansion factors or fuzzy expansion factors, so that the problems that the expansion factor parameters are difficult to determine, the expert experience is excessively depended, the expansion factor parameter values are fixed, the dynamic design cannot be carried out according to the feedback information of the system, and the optimal control effect cannot be ensured when the semi-active suspension of the vehicle is controlled according to the control forces in different directions.
Disclosure of Invention
In order to solve the problems, the invention provides a control method and a system for a multi-depth neural fuzzy network of a semi-active suspension of a whole vehicle, which are used for carrying out fuzzy reasoning on a speed difference value and a difference change rate of the speed difference value in each direction to determine the input variable expansion factor and the output variable expansion factor, so that the real-time adjustment of the input variable expansion factor and the output variable expansion factor according to the speed is realized, the optimal vibration reduction control effect of the vehicle is ensured when the vehicle is controlled according to the control force and the control moment which are obtained through fuzzy reasoning, and the running smoothness and the control stability of the vehicle are improved.
In order to achieve the above purpose, the invention adopts the following technical scheme:
in a first aspect, a method for controlling a multi-depth neural fuzzy network of a semi-active suspension of a whole vehicle is provided, which comprises the following steps:
acquiring the vertical speed, the pitch angle speed and the roll angle speed of the vehicle body;
calculating the obtained difference value between each direction speed and the corresponding direction reference speed and the difference value change rate thereof;
fuzzy reasoning is carried out on the speed difference value and the difference value change rate of each direction through a variable domain module of each direction, and the input variable expansion factor and the output variable expansion factor of each direction are determined;
the speed difference value and the difference value change rate of the speed difference value in each direction are adjusted through the input variable expansion factors in each direction, and the speed difference value and the difference value change rate of the speed difference value after adjustment in each direction are obtained;
carrying out fuzzy reasoning on the speed difference value and the difference value change rate thereof after each direction is adjusted through a main controller module of each direction, and determining theoretical control force or theoretical control moment of each direction;
the theoretical control force or the theoretical control moment in each direction is adjusted through the output variable expansion factors in each direction, and the control force or the control moment in each direction is determined;
and controlling the whole vehicle according to the control force or the control moment in each direction.
In a second aspect, a full vehicle semi-active suspension multiple depth neural fuzzy network control system is provided, including:
the speed acquisition module is used for acquiring the vertical speed, the pitch angle speed and the roll angle speed of the vehicle body;
the input module is used for calculating the difference value between each acquired direction speed and the corresponding direction reference speed and the difference value change rate thereof;
the neural fuzzy network control module is used for carrying out fuzzy reasoning on the speed difference value and the difference value change rate of each direction through the variable domain module of each direction, and determining the input variable expansion factor and the output variable expansion factor of each direction; the speed difference value and the difference value change rate of the speed difference value in each direction are adjusted through the input variable expansion factors in each direction, and the speed difference value and the difference value change rate of the speed difference value after adjustment in each direction are obtained; carrying out fuzzy reasoning on the speed difference value and the difference value change rate thereof after each direction is adjusted through a main controller module of each direction, and determining theoretical control force or theoretical control moment of each direction; the theoretical control force or the theoretical control moment in each direction is adjusted through the output variable expansion factors in each direction, and the control force or the control moment in each direction is determined;
and the whole vehicle control module is used for controlling the whole vehicle according to the vertical control force, the pitching control moment and the rolling control moment.
In a third aspect, an electronic device is provided, including a memory, a processor, and computer instructions stored in the memory and running on the processor, where the computer instructions, when executed by the processor, perform the steps described in the method for controlling a multi-depth neural fuzzy network for a semi-active suspension of a whole vehicle.
In a fourth aspect, a computer readable storage medium is provided for storing computer instructions that, when executed by a processor, perform the steps described in the overall vehicle semi-active suspension multiple depth neural-fuzzy network control method.
Compared with the prior art, the invention has the beneficial effects that:
1. according to the invention, fuzzy reasoning is carried out on the speed difference value and the difference value change rate of each direction through each direction change domain module, and the input variable expansion factor and the output variable expansion factor of each direction are determined; the speed difference value and the difference value change rate of each direction are adjusted through the input variable expansion factors of each direction, the speed difference value and the difference value change rate of each direction after adjustment are used as the input of the main controller module of each direction, the main controller module carries out fuzzy reasoning on the speed difference value and the difference value change rate after adjustment, and theoretical control force or theoretical control moment is determined; then, the fuzzy reasoning result is adjusted through the output variable expansion factor, and the final control force or control moment is obtained; the real-time adjustment of the input variable expansion factor and the output variable expansion factor according to the speed is realized, and the problems that the traditional variable-domain fuzzy controller is excessively dependent on expert experience, is difficult to obtain perfect fuzzy control rules, has low system control precision and the like are effectively solved.
2. After the control force or the control moment in each direction is determined, the force and the moment are distributed to four vibration absorbers through the force controller module, so that ideal control force of each vibration absorber is obtained, actual control current of the vibration absorber is determined according to the ideal control force, and the actual control current is input into the vibration absorber, so that the actual control force is obtained; controlling the vehicle by the actual control force; the coordination control of the whole vehicle semi-active suspension system is realized, and the running smoothness and the operating stability of the vehicle are improved.
3. The invention determines the actual control current of the shock absorber through the shock absorber inverse model, wherein the shock absorber inverse model takes ideal control force as input and actual control current as output, and is obtained by constructing a self-adaptive neural fuzzy network, so that the real-time adjustment of the actual control current according to the ideal control force is realized, and the obtained actual control current can realize the optimal control of the vehicle after being input into the shock absorber.
4. The variable domain module and the main controller module in different directions are cascaded by adopting a serial idea to form the deep neural fuzzy network control module in the direction of the invention. The deep neural fuzzy network modules in different directions are connected in parallel, and then the force controller module and the vibration damper inverse model modules in different directions are combined to form the multiple deep neural fuzzy network control module of the whole vehicle semi-active suspension, so that the optimal control of vibration damping of the vehicle is realized.
Additional aspects of the invention will be set forth in part in the description which follows and, in part, will be obvious from the description, or may be learned by practice of the invention.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiments of the application and together with the description serve to explain the application and do not constitute an undue limitation to the application.
FIG. 1 is a flow chart of an embodiment disclosed control method;
FIG. 2 is a block diagram of an ANFIS according to an exemplary embodiment;
FIG. 3 is a block diagram of a system according to an embodiment;
FIG. 4 is a graph of random road surface excitation time domain response of a four-wheel vehicle according to an embodiment;
FIG. 5 is a 7 degree of freedom whole vehicle dynamics model disclosed in the examples;
FIG. 6 is a sprung mass acceleration response curve of an embodiment disclosed;
FIG. 7 is a pitch angle acceleration response curve of the embodiment disclosure;
fig. 8 is a roll acceleration response curve of the embodiment disclosed.
Detailed Description
The invention will be further described with reference to the drawings and examples.
It should be noted that the following detailed description is illustrative and is intended to provide further explanation of the present application. Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs.
Example 1
In this embodiment, a method for controlling a multi-depth neural fuzzy network of a semi-active suspension of a whole vehicle is disclosed, as shown in fig. 1 to 8, including:
acquiring the vertical speed, the pitch angle speed and the roll angle speed of the vehicle body;
calculating the obtained difference value between each direction speed and the corresponding direction reference speed and the difference value change rate thereof;
fuzzy reasoning is carried out on the speed difference value and the difference value change rate of each direction through a variable domain module of each direction, and the input variable expansion factor and the output variable expansion factor of each direction are determined;
the speed difference value and the difference value change rate of the speed difference value in each direction are adjusted through the input variable expansion factors in each direction, and the speed difference value and the difference value change rate of the speed difference value after adjustment in each direction are obtained;
carrying out fuzzy reasoning on the speed difference value and the difference value change rate thereof after each direction is adjusted through a main controller module of each direction, and determining theoretical control force or theoretical control moment of each direction;
the theoretical control force or the theoretical control moment in each direction is adjusted through the output variable expansion factors in each direction, and the control force or the control moment in each direction is determined;
and controlling the whole vehicle according to the control force or the control moment in each direction.
The variable domain module of each direction is connected with the main controller module in series to form the deep neural fuzzy network control of the direction; and then, the depth neural fuzzy network control in different directions is connected in parallel to form a multiple depth neural fuzzy network control method, and the distributed force and moment are connected in series with the inverse models of the shock absorbers in different directions to realize the multiple depth fuzzy network control method of the whole vehicle semi-active suspension.
Specifically, the difference between the vertical speed of the vehicle body and the vertical reference speed and the change rate of the difference are calculated to obtain a vertical speed difference e v Rate of change of difference ec v The method comprises the steps of carrying out a first treatment on the surface of the Calculating the difference between the pitch angle speed and the pitch angle reference speed and the change rate of the difference, and obtaining a pitch angle speed difference e p Rate of change of difference ec p The method comprises the steps of carrying out a first treatment on the surface of the Calculating the difference between the roll angle speed and the roll angle reference speed and the change rate of the difference to obtain a roll angle speed difference e r Rate of change of difference ec r 。
After the speed difference value and the difference value change rate of each direction are calculated, a deep neural fuzzy network control module is further arranged for each direction, each deep neural fuzzy control module comprises a variable domain module and a main controller module, the variable domain module of each direction takes the speed difference value and the difference value change rate of the speed difference value of the corresponding direction as input, takes two input variable expansion factors and two output variable expansion factors as output, and is constructed by adopting a self-adaptive neural fuzzy network; the speed difference value and the difference value change rate of the speed difference value in each direction are adjusted through the input variable expansion factors in each direction, and the speed difference value and the difference value change rate of the speed difference value after adjustment in each direction are obtained; the main controller module in each direction takes the speed difference value and the difference value change rate thereof after the corresponding direction is adjusted as input, takes the theoretical control force or the theoretical control moment in the corresponding direction as output, and is constructed and obtained by adopting a self-adaptive neural fuzzy network; and adjusting the theoretical control force or the theoretical control moment in the corresponding direction through the output variable expansion factors in each direction, and determining the control force or the control moment in each direction.
The speed difference value and the difference value change rate of the speed difference value in each direction are divided by the input variable expansion factor in the corresponding direction, so that the speed difference value and the difference value change rate of the speed difference value after adjustment in each direction are obtained, and the speed difference value and the difference value change rate of the speed difference value are adjusted through the input variable expansion factor.
Multiplying the theoretical control force or the theoretical control moment in each direction by the output variable expansion factor in the corresponding direction to obtain the control force or the control moment in each direction; the adjustment of theoretical control force and theoretical control moment by the output variable expansion factor is realized.
Specific:
the vertical depth neural fuzzy network control module is arranged vertically and comprises a vertical variable domain module and a vertical vehicle main controller module, wherein the input of the vertical variable domain module is a vertical speed difference e v Rate of change of difference ec v . The output is the expansion factor alpha of the vertical speed difference ve Expansion factor alpha of vertical speed difference change rate vec Expansion factor beta of vertical control force vu The method comprises the steps of adopting an adaptive neural fuzzy network (ANFIS) to construct and obtain; will e v Divided by alpha ve Obtaining the velocity difference after vertical adjustmentThe method comprises the steps of carrying out a first treatment on the surface of the Will e cv Divided by alpha vec Obtaining the change rate of the speed difference value after vertical adjustment; the input of the vertical vehicle main controller module is a vertical adjusted speed difference value and a vertical adjusted speed difference value change rate, the output is a vertical theoretical control force, and the vertical theoretical control force is obtained by adopting an ANFIS construction; multiplying the vertical theoretical control force by beta vu And obtaining the vertical control force.
The pitching depth neural fuzzy network control module is arranged for the pitching direction and comprises a pitching variable domain module and a pitching vehicle main controller module, wherein the input of the pitching variable domain module is a pitching angle speed difference e p Rate of change of difference ec p Output is the expansion factor alpha of the pitch angle speed difference pe Expansion factor alpha of pitch angle speed difference change rate pec Expansion factor beta of pitching control moment pu The method comprises the steps of adopting an adaptive neural fuzzy network (ANFIS) to construct and obtain; will e p Divided by alpha pe Obtaining an angular velocity difference value after pitching adjustment; will ec p Divided by alpha pec Obtaining the change rate of the angular velocity difference after pitching adjustment; the input of the pitching vehicle main controller module is a pitching adjusted angular velocity difference value and a pitching adjusted angular velocity difference value change rate, and the output is a pitching theoretical control moment, and the pitching theoretical control moment is obtained by adopting an ANFIS construction; multiplying pitch theory control moment by beta pu Pitch control moment is obtained.
The roll depth neural fuzzy network control module is arranged for the roll direction and comprises a roll variation domain module and a roll vehicle main controller module, wherein the roll variation domain module is input into a roll angle speed difference e r Rate of change of difference ec r The output is the expansion factor alpha of the roll angle speed difference value re Expansion factor alpha of roll angle speed difference change rate rec Expansion factor beta of roll control moment ru The method comprises the steps of adopting an adaptive neural fuzzy network (ANFIS) to construct and obtain; will e r Divided by alpha re Obtaining a roll adjusted angular velocity difference; will ec r Divided by alpha rec Obtaining the change rate of the angular velocity difference after the roll adjustment; roll vehicle master controller moduleThe input is a roll adjusted angular velocity difference value and a roll adjusted angular velocity difference value change rate, the output is a roll theoretical control moment, and the roll theoretical control moment is obtained by adopting an ANFIS construction; multiplying roll theory control moment by beta ru A roll control moment is obtained.
As shown in FIG. 2, ANFIS is an adaptive network system that integrates a neural network and a T-S fuzzy inference system, which can approximate the T-S fuzzy inference system with arbitrary precision to express a nonlinear function.
In FIG. 2, L 1 The layer is responsible for inputting variable x 1 ,x 2 Is fuzzified by L 2 The layer is responsible for calculating the adaptation degree value of the fuzzy rule, L 3 The layer is responsible for normalizing the adaptation degree value of the fuzzy rule, L 4 The layer is responsible for calculating output, L 5 The layer is responsible for computing the total output y of the system.
The T-S fuzzy controller may be constructed from the following equation:
Q i if x 1 Is A i1 ，x 2 Is A i2 ,x j Is A ij ,…，x n Is A in Then:
y i (X)＝b i0 +b i1 x 1 +b i2 x 2 +…+b ij x j +…+b in x n (1)
A ij (x j )＝p ij (x j ) (4)
in which Q i (i=1, 2, …, Q) is the ith rule, x= [ X ] 1 ,x 2 ,…,x n ] T For inputting vectors, A ij To blur set, y i (X) is the output of the ith rule, f (X) is the output of the T-S fuzzy controller, beta i (X) degree of membership satisfied by rule i, p ij (x j ) Is a membership function of the input domain.
The T-S fuzzy controller in the main controller module inputs a speed difference value and a difference value change rate thereof in the corresponding direction after being adjusted by the input variable expansion factor, and outputs a theoretical control force or a theoretical control moment in the corresponding direction; the T-S fuzzy controller in the variable domain module inputs the speed difference value and the difference value change rate in the corresponding direction, and outputs the input variable expansion factor and the output variable expansion factor in the corresponding direction. The real-time adjustment of the input variable expansion factor and the output variable expansion factor according to the vehicle speed is realized, so that the optimal control of vibration reduction of the whole vehicle can be realized, and the problems that the traditional variable-domain fuzzy controller is excessively dependent on expert experience, is difficult to obtain perfect fuzzy control rules, has low system control precision and the like are effectively solved.
According to the method, the vibration reduction effect of the whole vehicle is evaluated through the evaluation function, wherein the evaluation indexes of the evaluation function comprise the vertical acceleration, the pitch acceleration and the roll acceleration of the vehicle body, the evaluation function calculates a root mean square value for each evaluation index, all root mean square values are weighted and summed, a final evaluation function value is determined, and the minimum evaluation function value indicates that the vibration reduction effect of the whole vehicle is optimal.
The membership functions of the main controller module and the variable domain module are all Gaussian membership functions.
The language fuzzy subsets adopted by the main controller module and the variable domain module are: { NB, NM, NS, Z, PS, PM, PB }; wherein NB is negative big, NM is negative middle, NS is negative small, Z is zero, PS is positive small, PM is positive, PB is positive big.
Taking a vertical deep neural fuzzy network control module as an example for explanation, a double-input-single-output T-S fuzzy controller is established as a vertical vehicle main controller, a vertical adjusted speed difference value and a difference value change rate thereof are taken as inputs, and a vertical theoretical control force is taken as an output. The fuzzy domains of the input and output variables are [ -1,1], the initial quantization factor ke=10, the key=0.5, the initial scaling factor ku=600, and the magnitude values of the quantization factor and the scaling factor are determined by repeated debugging according to the control effect. The design method of the pitching vehicle main controller and the rolling vehicle main controller is similar to the design method of the vertical vehicle main controller, and is not repeated here, so as to ensure the smooth transition of the output of the algorithm.
The method comprises the steps of selecting an evaluation function taking vehicle body vertical acceleration, pitch angle acceleration and roll angle acceleration as main indexes, wherein the simulation time is 10s, the step length is 0.001, and 10000 groups of data taking vehicle body vertical speed and vehicle body vertical acceleration, pitch angle speed and pitch angle acceleration, roll angle speed and roll angle acceleration as input and vehicle vertical control force, pitch control moment and roll control moment as output are obtained. The obtained data is trained by a double-input-single-output T-S fuzzy controller in a corresponding direction, so that a vertical vehicle main control module, a pitching vehicle main control module and a rolling vehicle main control module based on ANFIS are constructed.
The T-S fuzzy controller with double input and three output is established as a vertical variable domain module, and the module takes a vertical speed difference value and a difference value change rate thereof as input, and a vertical input variable expansion factor alpha ve 、α vec Vertical output variable expansion factor beta vu As an output. The fuzzy domains of the input variables are [ -1,1]Fuzzy arguments of output variable are [0,1]. The membership functions of the 7 language fuzzy subsets { NB (negative big), NM (negative medium), NS (negative small), Z (zero), PS (positive small), PM (median), PB (positive big) } are used. The design method of the pitch variable domain module and the roll variable domain module is similar to the design method of the vertical variable domain module, and is not described in detail herein.
The method comprises the steps of selecting an evaluation function taking vertical acceleration, pitch angle acceleration and roll angle acceleration of a vehicle body as main indexes, wherein simulation time is 10s, step length is 0.001, and 10000 groups of data which are respectively input by the vertical speed and the vehicle body acceleration of the vehicle body, the pitch angle speed and the pitch angle acceleration, and the roll angle speed and the roll angle acceleration and are output by the vertical input variable expansion factor, the vertical output variable expansion factor, the pitch input variable expansion factor, the pitch output variable expansion factor, the roll input variable expansion factor and the roll output variable expansion factor are obtained. Training the obtained data by using a double-input-three-output T-S fuzzy controller in a corresponding direction, thereby obtaining a vertical variable domain module, a pitching variable domain module and a rolling variable domain module based on ANFIS.
The embodiment also acquires a pitch angle and a roll angle;
determining an ideal control force of each vehicle shock absorber according to the pitch angle, the roll angle and the control forces or control moments in all directions;
determining an actual control current for each shock absorber according to the desired control force for each vehicle shock absorber;
inputting actual control current into the shock absorber to obtain actual control force;
the vehicle is controlled by the actual control force.
Specifically, determining the actual control current of the shock absorber according to a shock absorber inverse model of the vehicle;
the inverse model of the shock absorber takes ideal control force and the relative speed of a piston rod of the shock absorber as input, takes actual control current as output, and is obtained by constructing a self-adaptive neural fuzzy network.
The damper adopted in this embodiment is a CDC damper, and the pitch angle, the roll angle, and the vertical control force, the pitch control moment, and the roll control moment obtained from the multiple depth neural fuzzy control module are sent to the force controller module, and the force controller module effectively distributes the ideal control force to four vehicle dampers in combination with the vehicle geometry relationship, respectively as F 1 、F 2 、F 3 、F 4 . Wherein F is 1 、F 2 、F 3 、F 4 The force is desirably controlled for each vehicle shock absorber.
The vibration damper inverse model of the CDC vibration damper in four directions is respectively used for F 1 、F 2 、F 3 、F 4 Reasoning is carried out to obtain the actual control current i required by the four shock absorbers 1 、i 2 、i 3 、i 4 Four dampers are arrangedThe actual control current is input into the CDC vibration damper in four directions to obtain the actual control force F of the vibration damper 5 、F 6 、F 7 、F 8 。
And acquiring test data of the external characteristics of the CDC damper under different excitation frequencies and currents through a field bench test, and training the acquired data by using the ANFIS so as to construct an inverse model of the CDC damper based on the ANFIS. The input of the inverse model of the four CDC vibration absorbers is the ideal control force of each vibration absorber and the relative speed of the piston rod of the vibration absorber, the output is four actual control currents of the vibration absorbers, and the obtained four actual control currents are respectively sent into the four CDC vibration absorbers to obtain the actual control force of the four CDC vibration absorbers.
As shown in FIG. 3, the present embodiment also uses the actual control forces F of the four CDC dampers 5 、F 6 、F 7 、F 8 And the suspension is sent into a whole vehicle semi-active suspension system module.
Road surface excitation of four wheels is generated by a four-wheel road surface excitation module.
Preferably, the embodiment adopts a filtering white noise method to generate a single-wheel random pavement excitation time domain signal, and combines a parameterized model of a left-right track coherent function and a delay relation between front wheels and rear wheels of the same track to establish four-wheel vehicle pavement excitation. The research on single-wheel pavement excitation in the related literature is mature, and a single-wheel pavement excitation time domain model is shown in a formula (5).
Wherein v is the vehicle running speed, n 0 For reference spatial frequency, G xr (n 0 ) F is the road surface unevenness coefficient min F is the road surface space cut-off frequency min ＝0.011m -1 ，W rlf And (t) is single-round unit white noise excitation.
And constructing a four-wheel pavement excitation time domain model in Matlab/Simulink. Taking the B-class random road surface as an example, when the vehicle speed is 50km/h, the simulation results in the time domain response of the four-wheel vehicle random road surface excitation as shown in figure 4.
x ri Indicating the excitation magnitude of the elevation of the pavement, x rlf Represents the road surface height Cheng Jili, x of the left front wheel rrf Represents the road surface height Cheng Jili, x of the right front wheel rlr Represents the road surface height Cheng Jili, x of the left rear wheel rrr Indicating the right rear wheel road elevation stimulus.
The whole vehicle semi-active suspension system module mainly comprises a 7-degree-of-freedom whole vehicle dynamics system, 7 degrees of freedom refer to vertical, pitching and rolling of the mass center of a vehicle body and vertical movement of four vehicles, and the established 7-degree-of-freedom whole vehicle dynamics system is shown in fig. 5. In the figure, M represents the sprung mass of the vehicle, M ui (i=lf, lr, rf, rr) denotes unsprung mass, k i Representing suspension spring rate, k ti (i=lf, lr, rf, rr) represents tire stiffness, F i (i=lf, lr, rf, rr) denotes semi-active suspension control force, x ri (i=lf, lr, rf, rr) represents road surface excitation, x ti (i=lf, lr, rf, rr) expressed as unsprung mass vertical displacement, x bi (i=lf, lr, rf, rr) expressed as sprung mass vertical displacement, x c Represents the displacement of the mass center of the vehicle body, theta represents the pitch angle,the roll angle is represented, a and b represent the front and rear wheelbase, respectively, and c and d represent the left and right wheelbase, respectively.
According to darang Bei Yuanli, the kinetic equation of the semi-active suspension of the whole vehicle is established as follows:
vertical vibration of a vehicle body:
vehicle body roll vibration:
vehicle body pitching vibration:
non-sprung mass vertical vibration:
the vehicle model dynamics equation can be expressed as the following matrix form:
wherein M is 1 For semi-active suspension system mass matrix, K 1 As a rigidity coefficient matrix, ω= [ x ] rlf ,x rrf ,x rlr ,x rrr ] T For road surface disturbance excitation vector, u= [ F ] lf ,F rf ,F lr ,F rr ]A control force is input to the CDC damper,is a system state vector.
In order to verify the effectiveness of the method disclosed by the embodiment, the working conditions of a B-level random road surface and a vehicle speed of 50km/h are selected for test verification, and response curves of sprung mass acceleration, pitch angle acceleration and roll angle acceleration under the control strategies of passive, traditional fuzzy control and multiple depth neural fuzzy networks are obtained as shown in figures 6-8. The root mean square values of the corresponding indicators of the suspension system are shown in table 1.
TABLE 1 suspension System Performance index
As can be seen from fig. 5 and table 1, compared with the passive suspension, the root mean square values of the sprung mass acceleration, the pitch angle acceleration and the roll angle acceleration controlled by the multi-depth neural fuzzy network control method disclosed in the embodiment are respectively reduced by 38.65%, 30.27% and 9.86%, while root mean square indexes of the sprung mass acceleration, the pitch angle acceleration and the roll angle acceleration controlled by the traditional fuzzy control method are respectively reduced by 22.48%, 12.96% and 1.74%, and the proposed multi-depth neural fuzzy network control method has better control effect and can effectively improve the running smoothness of the vehicle. As can be seen from fig. 7, 8 and table 1, the left front, right front, left rear, right rear suspension deflection and left front, right front, left rear, right rear tire dynamic loads controlled by the multi-depth neural fuzzy network control method according to the present embodiment are reduced by 31.94%, 43.94%, 25%, 44.64%, 17.95%, 14.39%, 19.14% and 25.81%, respectively, while the left front, right front, left rear, right rear suspension deflection and left front, right front, left rear, right rear tire dynamic loads controlled by the conventional fuzzy control method are reduced by 22.22%, 21.21%, 11.36%, 8.93%, 5.51%, 5.04%, 5.00% and 4.72%, respectively, and the multi-depth neural fuzzy network control method according to the present embodiment has significantly better effects than the conventional fuzzy control method, and can effectively reduce the probability of collision between the suspension system and the vehicle stopper, thereby greatly improving the operation stability of the vehicle.
The multi-depth neural fuzzy network control method for the whole vehicle semi-active suspension disclosed by the embodiment can realize coordination control of the vertical, pitching and rolling of the whole vehicle semi-active suspension system and coupling vibration of four wheel directions, and effectively improves the running smoothness and the operation stability of the vehicle. The system mainly comprises a variable domain module, a vehicle main controller module, a force controller module, a shock absorber module and the like, wherein each module is obtained by training a self-adaptive neural fuzzy network (ANFIS) according to data. The modules are cascaded through a serial idea to form a deep neural fuzzy network control module, and the whole vehicle semi-active suspension mainly coordinates vibration in three directions of vertical direction, pitching direction, rolling direction and the like of a vehicle, so that the multiple deep neural fuzzy network control method disclosed by the embodiment mainly comprises three deep neural fuzzy network control modules, namely a vertical deep neural fuzzy network control module, a pitching deep neural fuzzy control module and a rolling deep neural fuzzy network control module; the main controllers in the three deep neural fuzzy control modules are mainly responsible for controlling the vertical direction, the pitching direction and the rolling direction of the vehicle, and the three deep neural fuzzy network control modules send the output control force and the control moment into the force controllers of the vehicle to realize the force distribution of the CDC shock absorbers in four different directions; and the variable domain module is arranged in the deep neural fuzzy control module and used for determining the input variable expansion factor and the output variable expansion factor according to the speed difference value and the difference value change rate of the direction through fuzzy reasoning, and the input and output of the main controller module are adjusted through the input variable expansion factor and the output variable expansion factor, so that the real-time adjustment of the input variable expansion factor and the output variable expansion factor according to the speed is realized, and the problems that the traditional variable domain fuzzy controller is excessively dependent on expert experience, is difficult to obtain perfect fuzzy control rules, has low system control precision and the like are effectively solved.
The damper module mainly comprises a CDC damper inverse model module and a CDC damper based on test data. According to the external characteristic test data of the CDC damper, which are obtained through bench tests under different test frequencies and currents, ANFIS training is adopted to obtain a CDC damper inverse model, ideal control force is converted into actual control current of the CDC damper through the CDC damper inverse model, the actual control current is sent to the CDC damper to obtain actual output force of the CDC damper, and the actual output force is sent to a semi-active suspension system of the whole vehicle to realize vibration control of the vehicle, so that the vibration control effect of vehicle control is further ensured.
The method disclosed by the embodiment of the invention can realize the coordination control of the whole vehicle semi-active suspension system, adopts a multiple system design of layering and blocking from top to bottom, from left to right, reduces the operation amount and complexity of the method, and improves the self-adaption capability and control precision of the whole vehicle semi-active suspension system.
The method disclosed by the embodiment provides a new thought for the coordination control of the whole vehicle semi-active suspension system, is beneficial to the development and engineering practical application of the whole vehicle semi-active suspension system controller, and can effectively improve the running smoothness and the operation stability of the whole vehicle semi-active suspension system.
Example 2
In this embodiment, a full vehicle semi-active suspension multiple depth neural fuzzy network control system is disclosed, comprising:
the speed acquisition module is used for acquiring the vertical speed, the pitch angle speed and the roll angle speed of the vehicle body;
the input module is used for calculating the difference value between each acquired direction speed and the corresponding direction reference speed and the difference value change rate thereof;
the neural fuzzy network control module is used for carrying out fuzzy reasoning on the speed difference value and the difference value change rate of each direction through the variable domain module of each direction, and determining the input variable expansion factor and the output variable expansion factor of each direction; the speed difference value and the difference value change rate of the speed difference value in each direction are adjusted through the input variable expansion factors in each direction, and the speed difference value and the difference value change rate of the speed difference value after adjustment in each direction are obtained; carrying out fuzzy reasoning on the speed difference value and the difference value change rate thereof after each direction is adjusted through a main controller module of each direction, and determining theoretical control force or theoretical control moment of each direction; the theoretical control force or the theoretical control moment in each direction is adjusted through the output variable expansion factors in each direction, and the control force or the control moment in each direction is determined;
and the whole vehicle control module is used for controlling the whole vehicle according to the vertical control force, the pitching control moment and the rolling control moment.
The speed acquisition module comprises a sensor for acquiring each speed.
The neural fuzzy network control module comprises a vertical neural fuzzy network control module, a pitching neural fuzzy network control module and a rolling neural fuzzy network control module.
The vertical neural fuzzy network control module is used for carrying out fuzzy reasoning on the vertical speed difference value and the difference change rate thereof through the vertical variable domain module to determine a vertical input variable expansion factor and a vertical output variable expansion factor; the vertical speed difference value and the difference value change rate thereof are adjusted through a vertical input variable expansion factor, and the speed difference value and the difference value change rate thereof after vertical adjustment are obtained; the vertical main controller module performs fuzzy reasoning on the vertical speed difference value after vertical adjustment and the difference value change rate thereof to determine a vertical theoretical control force; the vertical theoretical control force is adjusted through a vertical output variable expansion factor, and the vertical control force is determined;
the pitching neural fuzzy network control module is used for carrying out fuzzy reasoning on the pitch angle speed difference value and the difference change rate thereof through the pitching variable domain module to determine a pitching input variable expansion factor and a pitching output variable expansion factor; the pitch input variable expansion factor is used for adjusting the pitch angle speed difference value and the difference change rate thereof, so as to obtain the pitch-adjusted angular speed difference value and the difference change rate thereof; the pitching main controller module carries out fuzzy reasoning on the difference value of the angular velocity after pitching adjustment and the change rate of the difference value, and determines the theoretical control moment of pitching; the theoretical pitching control moment is adjusted through the pitching output variable expansion factor, and the pitching control moment is determined;
the roll neural fuzzy network control module is used for carrying out fuzzy reasoning on the roll angle speed difference value and the difference change rate thereof through the roll change domain module to determine roll input variable expansion factors and roll output variable expansion factors; the roll input variable expansion factor is used for adjusting the roll inclination angle speed difference value and the difference change rate thereof to obtain the roll adjusted angular speed difference value and the difference change rate thereof; the roll main controller module performs fuzzy reasoning on the difference value of the angular velocity after the roll adjustment and the change rate of the difference value, and determines the theoretical control moment of the roll; the roll control moment is determined by adjusting the roll theoretical control moment by the roll output variable stretch factor.
Example 3
In this embodiment, an electronic device is disclosed that includes a memory and a processor, and computer instructions stored on the memory and running on the processor, which when executed by the processor, perform the steps described in the full vehicle semi-active suspension multiple depth neural fuzzy network control method disclosed in embodiment 1.
Example 4
In this embodiment, a computer readable storage medium is disclosed for storing computer instructions that, when executed by a processor, perform the steps described in the full vehicle semi-active suspension multiple depth neural fuzzy network control method disclosed in embodiment 1.
Finally, it should be noted that: the above embodiments are only for illustrating the technical aspects of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the above embodiments, it should be understood by those of ordinary skill in the art that: modifications and equivalents may be made to the specific embodiments of the invention without departing from the spirit and scope of the invention, which is intended to be covered by the claims.
Claims (10)
1. The control method of the multi-depth neural fuzzy network of the semi-active suspension of the whole vehicle is characterized by comprising the following steps:
acquiring the vertical speed, the pitch angle speed and the roll angle speed of the vehicle body;
calculating the obtained difference value between each direction speed and the corresponding direction reference speed and the difference value change rate thereof;
fuzzy reasoning is carried out on the speed difference value and the difference value change rate of each direction through a variable domain module of each direction, and the input variable expansion factor and the output variable expansion factor of each direction are determined;
the speed difference value and the difference value change rate of the speed difference value in each direction are adjusted through the input variable expansion factors in each direction, and the speed difference value and the difference value change rate of the speed difference value after adjustment in each direction are obtained;
carrying out fuzzy reasoning on the speed difference value and the difference value change rate thereof after each direction is adjusted through a main controller module of each direction, and determining theoretical control force or theoretical control moment of each direction;
the theoretical control force or the theoretical control moment in each direction is adjusted through the output variable expansion factors in each direction, and the control force or the control moment in each direction is determined;
and controlling the whole vehicle according to the control force or the control moment in each direction.
2. The method for controlling the multi-depth neural fuzzy network of the semi-active suspension of the whole vehicle according to claim 1, wherein the variable domain module of each direction takes a speed difference value and a difference value change rate thereof in the corresponding direction as input, takes two input variable expansion factors and two output variable expansion factors as output, and is obtained by constructing an adaptive neural fuzzy network.
3. The method for controlling the multi-depth neural fuzzy network of the semi-active suspension of the whole vehicle according to claim 1, wherein the main controller module in each direction takes the speed difference value and the change rate of the difference value after the adjustment in the corresponding direction as input, takes the theoretical control force or the theoretical control moment in the corresponding direction as output, and is obtained by constructing an adaptive neural fuzzy network.
4. The method for controlling the multi-depth neural fuzzy network of the semi-active suspension of the whole vehicle according to claim 1, wherein the membership functions of the main controller module and the variable domain module are all Gaussian membership functions.
5. The control method of the full-vehicle semi-active suspension multi-depth neural fuzzy network according to claim 1, wherein the speed difference value and the difference value change rate of each direction are divided by the input variable expansion factors of the corresponding directions to obtain the speed difference value and the difference value change rate of each direction after adjustment;
and multiplying the theoretical control force or the theoretical control moment of each direction by the output variable expansion factor of the corresponding direction to obtain the control force or the control moment of each direction.
6. The method for controlling the multi-depth neural fuzzy network of the semi-active suspension of the whole vehicle according to claim 1, wherein pitch angle and roll angle are also obtained;
determining an ideal control force of each vehicle shock absorber according to the pitch angle, the roll angle and the control forces or control moments in all directions;
determining an actual control current for each shock absorber according to the desired control force for each vehicle shock absorber;
inputting actual control current into the shock absorber to obtain actual control force;
the vehicle is controlled by the actual control force.
7. The control method of the full vehicle semi-active suspension multi-depth neural fuzzy network according to claim 6, wherein the actual control current of the shock absorber is determined according to a shock absorber inverse model of the vehicle;
the inverse model of the shock absorber takes ideal control force and the relative speed of a piston rod of the shock absorber as input, takes actual control current as output, and is obtained by constructing a self-adaptive neural fuzzy network.
8. The utility model provides a full vehicle semi-active suspension multiple degree of depth neural fuzzy network control system which characterized in that includes:
the speed acquisition module is used for acquiring the vertical speed, the pitch angle speed and the roll angle speed of the vehicle body;
the input module is used for calculating the difference value between each acquired direction speed and the corresponding direction reference speed and the difference value change rate thereof;
the neural fuzzy network control module is used for carrying out fuzzy reasoning on the speed difference value and the difference value change rate of each direction through the variable domain module of each direction, and determining the input variable expansion factor and the output variable expansion factor of each direction; the speed difference value and the difference value change rate of the speed difference value in each direction are adjusted through the input variable expansion factors in each direction, and the speed difference value and the difference value change rate of the speed difference value after adjustment in each direction are obtained; carrying out fuzzy reasoning on the speed difference value and the difference value change rate thereof after each direction is adjusted through a main controller module of each direction, and determining theoretical control force or theoretical control moment of each direction; the theoretical control force or the theoretical control moment in each direction is adjusted through the output variable expansion factors in each direction, and the control force or the control moment in each direction is determined;
and the whole vehicle control module is used for controlling the whole vehicle according to the vertical control force, the pitching control moment and the rolling control moment.
9. An electronic device comprising a memory and a processor, and computer instructions stored on the memory and running on the processor, which when executed by the processor, perform the steps of the whole vehicle semi-active suspension multiple depth neural fuzzy network control method of any one of claims 1-7.
10. A computer readable storage medium storing computer instructions which, when executed by a processor, perform the steps of the whole vehicle semi-active suspension multiple depth neural fuzzy network control method of any one of claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311243373.7A CN117331312A (en) | 2023-09-22 | 2023-09-22 | Multi-depth neural fuzzy network control method and system for semi-active suspension of whole vehicle |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311243373.7A CN117331312A (en) | 2023-09-22 | 2023-09-22 | Multi-depth neural fuzzy network control method and system for semi-active suspension of whole vehicle |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN117331312A true CN117331312A (en) | 2024-01-02 |
Family
ID=89294478
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311243373.7A Pending CN117331312A (en) | 2023-09-22 | 2023-09-22 | Multi-depth neural fuzzy network control method and system for semi-active suspension of whole vehicle |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117331312A (en) |
-
2023
- 2023-09-22 CN CN202311243373.7A patent/CN117331312A/en active Pending
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108891221A (en) | A kind of active suspension system and its working method based on mode energy distribution method | |
| CN105539052B (en) | A kind of controllable suspension sliding formwork tracking controller using vehicle plateau as reference | |
| CN103754081B (en) | The optimum fuzzy compound controller method of Vehicle Suspension with Non-linear Spring | |
| CN101916113B (en) | Automotive body gesture decoupling control method based on active suspension evaluation indicator | |
| CN108859648B (en) | Suspension shock absorber damping control switching weighting coefficient determination method | |
| WO2023174327A1 (en) | Cooperative control module, adaptive cruise control system as well as control method thereof, and vehicle | |
| CN106926660B (en) | A kind of electromagnetic suspension system and its control method based on wheel rim driven motor vehicle | |
| Tianjun et al. | Model reference adaptive control of semi-active suspension model based on AdaBoost algorithm for rollover prediction | |
| CN102975587B (en) | Vehicle semiactive suspension based on double controllable dampers and control method thereof | |
| CN105676694A (en) | Intelligent sampling data output feedback control method for automobile suspension system | |
| CN112906133A (en) | Vertical vibration negative effect suppression method for movable inertial suspension controlled by ground shed | |
| CN115099035B (en) | Suspension vibration reduction design method with negative stiffness and inertial capacity cooperation under random displacement excitation | |
| WO2023108699A1 (en) | Novel lqg control method | |
| CN117331312A (en) | Multi-depth neural fuzzy network control method and system for semi-active suspension of whole vehicle | |
| yin Tang et al. | Research on suspension system based on genetic algorithm and neural network control | |
| jie Yue et al. | Research on vehicle suspension systems based on fuzzy logic control | |
| Du et al. | Control Strategy for Semi-Active Suspension Based on Suspension Parameter Estimation | |
| CN111338208B (en) | Vehicle side-tipping and smoothness coordination control method | |
| Chen et al. | Study on control of an two hub-motor electric vehicle | |
| CN110765554A (en) | Intelligent control method of automobile semi-active suspension system based on TS model | |
| Chen et al. | Vehicle chassis decoupling control based on neural network inverse method | |
| Liu et al. | Multi-Objective Control of Dynamic Chassis Considering Road Roughness Class Recognition | |
| Bagchi et al. | Designing and optimizing an active car suspension with genetic algorithm | |
| Chen et al. | Research on optimal control of active suspension for intelligent vehicles based on simulink | |
| CN118358311B (en) | Magnetorheological semi-active suspension control method based on model reference self-adaptive control |
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 |