CN104401232A  Electric vehicle torque optimization method based on data driven predictive control  Google Patents
Electric vehicle torque optimization method based on data driven predictive control Download PDFInfo
 Publication number
 CN104401232A CN104401232A CN201410793965.0A CN201410793965A CN104401232A CN 104401232 A CN104401232 A CN 104401232A CN 201410793965 A CN201410793965 A CN 201410793965A CN 104401232 A CN104401232 A CN 104401232A
 Authority
 CN
 China
 Prior art keywords
 data
 predictive
 torque
 output
 control
 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.)
 Granted
Links
 238000005457 optimization Methods 0.000 title claims abstract description 30
 238000004088 simulation Methods 0.000 claims abstract description 21
 230000005284 excitation Effects 0.000 claims abstract description 12
 239000011159 matrix material Substances 0.000 claims description 33
 230000004044 response Effects 0.000 claims description 7
 230000005540 biological transmission Effects 0.000 claims description 5
 238000010276 construction Methods 0.000 claims description 3
 238000005259 measurement Methods 0.000 claims 1
 230000000875 corresponding Effects 0.000 abstract description 4
 238000009826 distribution Methods 0.000 abstract description 3
 230000000694 effects Effects 0.000 description 13
 238000000034 methods Methods 0.000 description 8
 230000002493 climbing Effects 0.000 description 4
 238000010586 diagrams Methods 0.000 description 4
 238000010606 normalization Methods 0.000 description 4
 239000007787 solids Substances 0.000 description 4
 230000035852 Tmax Effects 0.000 description 3
 125000004122 cyclic group Chemical group 0.000 description 3
 238000005516 engineering processes Methods 0.000 description 3
 230000001276 controlling effects Effects 0.000 description 2
 WHXSMMKQMYFTQSUHFFFAOYSAN lithium Chemical compound data:image/svg+xml;base64,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 data:image/svg+xml;base64,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 [Li] WHXSMMKQMYFTQSUHFFFAOYSAN 0.000 description 2
 229910052744 lithium Inorganic materials 0.000 description 2
 230000000452 restraining Effects 0.000 description 2
 238000005096 rolling process Methods 0.000 description 2
 241001269238 Data Species 0.000 description 1
 239000003570 air Substances 0.000 description 1
 238000004891 communication Methods 0.000 description 1
 239000008358 core components Substances 0.000 description 1
 238000003912 environmental pollution Methods 0.000 description 1
 230000004301 light adaptation Effects 0.000 description 1
 238000004519 manufacturing process Methods 0.000 description 1
 239000000203 mixtures Substances 0.000 description 1
 230000001141 propulsive Effects 0.000 description 1
 230000001172 regenerating Effects 0.000 description 1
 230000000717 retained Effects 0.000 description 1
 230000035939 shock Effects 0.000 description 1
 230000001360 synchronised Effects 0.000 description 1
Classifications

 Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSSSECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSSREFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
 Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
 Y02T—CLIMATE CHANGE MITIGATION TECHNOLOGIES RELATED TO TRANSPORTATION
 Y02T10/00—Road transport of goods or passengers
 Y02T10/60—Other road transportation technologies with climate change mitigation effect
 Y02T10/72—Electric energy management in electromobility
Abstract
The invention provides an electric vehicle torque optimization method based on data driven predictive control, belongs to the technical field of electric vehicles, and aims to effectively perform optimized distribution of driving and braking torque of an electric vehicle so as to realize a tracking control method for the longitudinal velocity of the vehicle. The method comprises the steps as follows: firstly, appropriate excitation data are designed according to dynamic characteristics of a system, so that sufficient excitation for the system is guaranteed; secondly, a predictive output equation of the system is constructed by input/ output data obtained by excitation; then actuator rigid constraints of a motor, a battery pack and a brake are considered, and a cost function for torque optimization control is constructed with a model predictive control algorithm; finally, an optimization problem corresponding to the cost function is solved to obtain control input which then acts on the system, so that the system is controlled. For the control algorithm, all that is required is offline simulation experiments when the excitation data are acquired, so that the development cost is lower.
Description
Technical field
The invention belongs to electric vehicle engineering field.
Background technology
Along with day by day increasing the weight of of environmental pollution and energy shock, pure electric automobile is because high energy efficiency, antiemission carburetor, low noise, the great direction that the advantages such as energy regenerating become current automobile industry development can be carried out, country has also put into effect a large amount of preferential policies and has supported business research electronlmobil, Tenth FiveYear Plan Period, National 863 plan " electronlmobil major scientific and technological project " continuous two phases obtain country and support energetically, through a lot of enterprise, the effort research of institution of higher learning and research institute personnel, there has been very big breakthrough in China in the research of pure electric automobile, the research and development of its core component and Full Vehicle System design aspect have had many technology innovations and breakthrough.Be the problem that orthodox car or electronlmobil all relate to moment of torsion optimization, how CONTROLLER DESIGN reasonably produces driving and lock torque signal, realizes the super performance of vehicle, is one of major issue of auto trade research always.For electronlmobil, because it is relatively simple for structure, control problem can suitably be simplified, and meanwhile, the fast response time of motor, moment of torsion and the rotating speed of motor easily obtain, and this is that the moment of torsion optimal control of electronlmobil provides good basic condition.Be directed to centralized electric automobile torque optimal control, mainly contain following problem:
1. because research object is centralized pure electric automobile, only has a motor, and be fwd, so there is not allocation optimization problems, and then torque distribution optimal design can not be carried out as fourwheel Inwheel motor driving electronlmobil, can only drive from entire car controller/the generation aspect of brake torque is considered.
2. electric automobile torque Optimal Control System mechanism model is complicated, by force nonlinear, modeling difficulty.Meanwhile, the output driving of controller and lock torque signal, meet the firm constraints of performer motor and drg, and namely the output driving of controller and lock torque signal can not exceed the actual maximum output torque of motor and the maximum braking moment of drg.
3. electronlmobil needs electrical source of power, conventional is lithium cell group now, come to feeding electric motors, the power supply voltage of motor also affects the maximum output torque of motor, therefore also must consider the impact of battery pack output voltage when considering the maximum output torque of motor.
Summary of the invention
The object of the invention is effectively to carry out electronlmobil to drive and lock torque optimization distribution, to realize the electric automobile torque optimization method based on datadriven predictive control of vehicular longitudinal velocity tracking and controlling method.
Step of the present invention is:
A, centralized electric vehicle simulation model buildings: adopt modular approach, the whole vehicle model of centralized electronlmobil is built in AMESim, whole electric vehicle simulation model mainly comprises Electric drive module, transmission module, tire module and longitudinal direction of car dynam, whole vehicle model parameter is as table one:
Table one electronlmobil parameter list
；
B, moment of torsion optimization data drive Design of Predictive: the datadriven predictive control algorithm of employing is combined by Subspace Identification and Model Predictive Control to form, and concrete step is as follows:
1. the choosing of excited data: choose drive torque order and lock torque order as excitation incoming signal, in AMESim whole vehicle model, by design drive torque and lock torque data to motor and drg, open loop operation model, just the actual vehicle speed measured is obtained, the speed of a motor vehicle that drive torque order, lock torque order and measurement obtain is the final system incentive data needed; Obtain input Hankel matrix U p, the Uf of system according to these data, export Hankel matrix Yp, Yf:
；
2. predictive equation:
The prediction output equation being gone out system by input and output Hankel matrix construction is:
，
(1)
being the following output valve of the system estimated by subspace state space system identification, by solving least square problem, obtaining two prediction matrix coefficient Lw and Lu in predictive equation
(2)
The solution of two prediction matrix coefficient Lw and Lu is tried to achieve by following formula:
(3)
At the Hankel matrix of the input/output information of known system, and after obtaining Lw and Lu, utilize subspace predictive equation (1), the following output valve of the system that namely obtains;
Final prediction of output equation is as formula (4),
it is Hankel matrix
first row
，
(4)
3. datadriven Design of Predictive: obtain the drive torque after optimizing and lock torque signal to solving of objective cost function, the mathematical expression form of cost function is as follows:
(5)
The predictive equation (5) that system exports is rewritten as increment pattern:
；
(6)
Export to carry out adding up to the prediction of newadded item type and obtain following model prediction output equation:
(7)
Wherein,
；
F item is system free response, and correspondingly S uf item is called as the control response of system;
Output model constructs the cost function of following form to utilize incremental to predict:
(8)
Wherein,
，
；
(9)
Wherein, Γ u and Γ y is the weighting matrix of input item and output item respectively; Control time domain is Nu, and predicts that time domain is Np; Coefficient of weight Λ=Γ u/ Γ y, selected Λ value is:
Motor maximum torque constraint in moment of torsion optimization system, the maximum braking moment of drg retrain and the constraint of actual vehicle speed is described by mathematic(al) representation below:
(10)。
The present invention, mainly for the moment of torsion optimization problem of centralized frontdriving electric automobile, utilizes and carrys out CONTROLLER DESIGN based on datadriven predictive control algorithm, realizes the optimization of electronlmobil driving and lock torque.Based on the perfect adaptation that datadriven predictive control algorithm is datadriven Subspace Identification and Model Predictive Control, the inputoutput data that it is directly gathered by offline and the characteristic obtained through data processing carry out CONTROLLER DESIGN, efficiently avoid complicated modelling by mechanism.By the inputoutput data gathered, the predictive equation of moment of torsion can be obtained, then utilize model predictive control method to construct cost function, consider constraint condition, the driving after optimizing solves and is optimized and lock torque command signal simultaneously.The present invention can solve the three point problem of prior art well, complicated system mechanism modeling effectively can be avoided based on datadriven predictive control algorithm, the moment of torsion optimal controller of the present invention's design effectively can avoid complicated system mechanism modeling, considers the actr firm constraints of motor, battery pack and drg simultaneously.Beneficial effect is:
1. traditional control algorithm majority is all the mechanism model based on control system, but in actual torque optimizing process, modelling by mechanism is complicated and work capacity is large.The inputoutput data that datadriven predictive control algorithm is directly gathered by offline and the characteristic obtained through data processing carry out CONTROLLER DESIGN, efficiently avoid complicated modelling by mechanism, eliminate solving of the middle state space equation to model, such one side decreases calculated amount, the model error problem introduced when avoiding from data identification to statespace model on the other hand.
2. the moment of torsion optimal controller system designed in the present invention is a nonlinear system, and take into account the actr firm constraints of motor, battery pack and drg, traditional control algorithm can not the constraint of actv. disposal system, and datadriven predictive control algorithm can process the control problem of multiinputmultioutput belt restraining effectively, directly constraint condition is compiled into line solver in the S_function in simulink when solving.
3. the development process of orthodox car ECU (Electrical Control Unit) needs a large amount of experiences and rating test.Only need in control algorithm of the present invention to need some offline simulations to test when obtaining excited data, cost of development is lower.
Accompanying drawing explanation
Fig. 1 is the electric automobile torque optimal control block diagram based on datadriven predictive control of the present invention;
Fig. 2 is centralized electronlmobil AMESim whole vehicle model of the present invention;
Fig. 3 is the desired speed under NEDC operating mode of the present invention, and unit is m/s, and abscissa is the time, unit s;
Fig. 4 is that datadriven predictive control algorithm used in the present invention forms block diagram;
Fig. 5 is the excited data that in the present invention, design torque optimal controller is chosen, and excitation is input as driving and lock torque order, unit Nm; The output data that offline obtains are actual vehicle speed, and unit is m/s;
Fig. 6 be the present invention in order to verify the identification result of forecast model one section of excited data of arbitrary extracting from excited data, abscissa is the time, unit s;
Fig. 7 is the present invention to verify that forecast model has redesigned one group and encouraged the identification result obtained, and abscissa is the time, unit s;
Fig. 8 is that the moment of torsion optimization data that the present invention builds in simulink drives predictive controller, mainly realizes datadriven predictive control algorithm by the M file compiler in Matlab;
Fig. 9 is the contrast of actual vehicle speed under initial parameter of the present invention after controller action and desired speed, and wherein dotted line represents actual vehicle speed, and solid line represents desired speed, and unit is m/s, and abscissa is the time, unit s;
Figure 10 is of the present inventionly only changing the gradient, and other parameters remain unchanged under condition, the actual vehicle speed after controller action and the contrast of desired speed, wherein dotted line represents actual vehicle speed, and solid line represents desired speed, and unit is m/s, abscissa is the time, unit s;
Figure 11 is of the present inventionly only changing car load load, and other parameters remain unchanged under condition, the actual vehicle speed after controller action and the contrast of desired speed, wherein dotted line represents actual vehicle speed, and solid line represents desired speed, and unit is m/s, abscissa is the time, unit s;
Figure 12 is of the present inventionly only changing wind speed, and other parameters remain unchanged under condition, the actual vehicle speed after controller action and the contrast of desired speed, wherein dotted line represents actual vehicle speed, and solid line represents desired speed, and unit is m/s, abscissa is the time, unit s.
Detailed description of the invention
Research technique of the present invention is the Model Predictive Control based on datadriven, comprises the following steps:
First for the excited data that the dynamic properties design of system is suitable, to ensure the abundant excitation to system; Secondly, the prediction output equation will the inputoutput data obtained being encouraged to construct system; Then, consider the actr firm constraints of motor, battery pack and drg, utilize Model Predictive Control Algorithm to construct the cost function of moment of torsion optimal control; Finally, by solving optimal problem corresponding to cost function, obtaining control inputs and acting on system, thus realizing the control to system.
For solving the problems of the technologies described above, the present invention adopts following technical scheme to realize:
Electric automobile torque optimization based on datadriven predictive control of the present invention is realized by the associative simulation of software system.Software system are made up of the senior simulation software of Matlab/Simulink and AMESim.Wherein AMESim software is a commercial modeling and simulating of complex system platform, its Main Function is in the present invention to provide the concentrated electric motor type pure electric automobile model of highfidelity, and in emulation experiment, this model replaces real vehicles as the objective for implementation of control method; MATLAB/Simulink software is then build for the realistic model of controller, namely have been programmed the building of moment of torsion optimal controller based on datadriven predictive control by Simulink.
Functionally the present invention can comprise following a few part: controller simulation module and centralized pure electric automobile model.Effect the following detailed description of each several part:
The effect main in the design of controller of centralized pure electric automobile model is the input and output excited data being obtained the dynamic characteristics that can embody system by offline simulation, and controller goes out the prediction output equation of system according to these data configurations.
The Main Function of controller module gathers various status informations such as the speed of a motor vehicle, cell output voltage, the motor output torque etc. of electronlmobil model, then carry out tabling lookup and computing, production control signalthe drive torque signal of motor and mechanical braking moment signal and this signal is sent to the actuating unit of electronlmobilmotor and traditional mechanical braking arrangement;
Electronlmobil model comprises several parts such as power drive system model, transmission shaft model, vehicle dynamic model, tire model.Its effect is that the control signal provided according to moment of torsion optimal controller produces corresponding driving or lock torque, changes the dynamic of vehiclemotor output torque, mechanical braking moment, the speed of a motor vehicle etc., realize the super performance of vehicle.
The control block diagram implemented based on the electric automobile torque optimization method of datadriven predictive control in the present invention as shown in Figure 1, in figure, moment of torsion optimal controller is built in Simulink, the input of controller is the speed of a motor vehicle expected, and actual vehicle speed is Realtime Feedback returns controller, Tmax is motor maximum drive moment, it is jointly determined by the mechanical specialities of motor and the output voltage of battery, both embodied the actr firm constraints condition of motor itself, embodied again along with battery discharge time increases the impact of voltage reduction on vehicle performance.The drive torque that entire car controller obtains must be less than or equal to Tmax, and therefore Tmax is given to controller as constraint.In figure, centralized electronlmobil model is built in AMESim, is used for simulating the operation of real vehicle.The driving that controller optimization obtains and lock torque signal are given to motor and brake module respectively, control the operation of vehicle, and the actual vehicle speed of vehicle as feedback signal back to controller.
Control objectives of the present invention is, the actual vehicle speed that moment of torsion optimal controller is returned according to Realtime Feedback and desired speed signal contrast, meeting under constraint condition prerequisite, optimize and obtain driving and lock torque signal Tm, Tb, and the motor be given in whole vehicle model and brake module, control the operation of vehicle, finally allow actual vehicle speed follow the tracks of upper desired speed.
The invention provides a set of device based on above operation logic and operational process.Namely based on the offline electric automobile torque optimal design test platform of PC.Build and operational process as follows:
1, software is selected
The controlled object of this control system and the realistic model of controller are built respectively by software Matlab/Simulink and AMESim, and software version is respectively Matlab R2009a and AMESim R10, and solver is selected to be respectively ode3 and Euler.Simulation step length is fixed step size, and steplength is chosen as 0.1s.
2, associative simulation is arranged
Realize the associative simulation of Matlab/Simulink and AMESim, first the environmental variance of PC computer must be set as required, make both interrelated.Then in AMESim interface, add the interface module communicated with simulink, the variable needing between Matlab/Simulink with AMESim to communicate is connected to this module; After eventually passing System build, the model information in AMESim is retained in Simulink with the form of Sfunction, thus the associative simulation of both realizations and communication.When running Simulink realistic model, AMESim model also carries out calculating and solving at the same time.The exchange of data is constantly carried out between the two in simulation process.If revised the model structure in AMESim or optimum configurations, then need to recompilate.It should be noted that both simulation step length must be consistent.
Detailed step of the present invention is:
A, centralized electric vehicle simulation model buildings: adopt modular approach, in AMESim, build the whole vehicle model of centralized electronlmobil, as shown in Figure 2.Whole electric vehicle simulation model mainly comprises Electric drive module, transmission module, tire module and longitudinal direction of car dynam, and whole vehicle model parameter is as table one:
Table one electronlmobil parameter list
；
Power drive system provides power for vehicle, comprises battery compartment and motor part.Battery is mainly feeding electric motors, and the battery compartment of pure electric automobile is made up of multiple cell connection in seriesparallel, and conventional is lithium cell group.The terminal voltage that battery pack externally exports be single battery output voltage and, battery system A terminal voltage and battery pack are supplied to the voltage of motor, and the specified terminal voltage of battery system is 320V in this model; What adopt in the present invention is permagnetic synchronous motor, and the control method of motor is vector control method, and the maximum output torque of motor is 300Nm.
Transmission module mainly comprises changespeed box and axle drive shaft.The moment that motor exports produces different speed ratio thus reduction rotating speed increase output torque via changespeed box by different tooth radius, here transmitting ratio is 5:1, and namely rotating speed becomes original 1/5, and moment of torsion is increased to original 5 times, just in time make up the rotating speed of motor own higher, the problem of moment of torsion deficiency.The linear running of primary study vehicle of the present invention, the output torque of left and right wheels is identical with rotating speed without difference, does not thus have diff.
Have four tire models in figure, tire used be Magic Formula magic formula model.
Longitudinal direction of car kinetic part, it considers vehicle and is subject to propulsive effort in the process of moving, the effect of braking force and resistance to motion, and wherein resistance to motion comprises air resistance, rolling resistance and friction drag.The parameter such as total quality, the gradient, wind speed of vehicle can be set in this module.
The state of cyclic operation of Model Selection is the state of cyclic operation NEDC (New European Driving Cycle) of European Lightduty Vehicle, parking containing the several times long period in this operating mode and uniform movement, and starting and braking curve are all comparatively level and smooth, can embody the steadystate behaviour of vehicle preferably.This Work condition analogue motoring condition of automobile under the road conditions of urban district, has stronger practicality.1200 seconds consuming time of EUDC driving cycle, under this operating mode, the desired speed of vehicle as shown in Figure 3, and it is exactly the desired speed of moment of torsion optimal controller.
The control objectives of b, middle controller of the present invention is the rate request met carrying out moment of torsion optimization while NEDC operating mode, simultaneously drive torque and lock torque should meet the constraint of motor maximum torque in reality and battery, and these constraints drive during predictive controller at design information and embody in Restriction condition treat.Moment of torsion optimization data of the present invention drives Design of Predictive: the datadriven predictive control algorithm of employing is combined by Subspace Identification and Model Predictive Control to form, and it forms schematic diagram as shown in Figure 4.The key distinction of the method and other control methods does not need to carry out CONTROLLER DESIGN by traditional transfer function or statespace model, but the following output valve of application subspace linear prediction prediction equation system, the coefficient of subspace predictive equation is directly calculated by the offline inputoutput data of system, based on predictive equation, combination model predictive control algorithm groundwork, obtains the datadriven forecast Control Algorithm of belt restraining.Concrete step is as follows:
1. the choosing of excited data: before CONTROLLER DESIGN, first need for the suitable excited data of the dynamic properties design of the whole vehicle model built above, then obtain system prediction output equation according to these excited datas.Here drive torque order and lock torque order is chosen as excitation incoming signal, in AMESim whole vehicle model, by design drive torque and lock torque data to motor and drg, open loop operation model, just obtain the actual vehicle speed measured, simulation step length is here 0.1s.The speed of a motor vehicle that drive torque order, lock torque order and measurement obtain, be the final system incentive data needed as shown in Figure 5.In the scope chosen due to drive torque of excitation be :300Nm300Nm, lock torque signal range: 01, vehicle speed range: 040m/s, in order to ensure the particularity of excitation model, normalization method has been carried out to excited data, the normalization coefficient of drive torque is 1/100, and lock torque normalization coefficient is 1, and speed of a motor vehicle normalization coefficient is 1/40.Before it should be noted that the driving after Optimization Solution obtains by controller and lock torque order are given to whole vehicle model, need renormalization.After obtaining excited data, saved as data data file, obtained input Hankel matrix U p, the Uf of system according to these data, exported Hankel matrix Yp, Yf:
P and f of the subscript of matrix represents " past (past) " and " following (future) " respectively, in order to ensure that data can comprise the enough information about state of the system, Hankel matrix column number is often needed to need enough large.For singleinput singleoutput system, the element in Hankel matrix is a numerical value, and for multiinput multioutput system, the element in Hankel matrix is the column matrix of a corresponding dimension.Of the present inventionly be input as drive torque order and lock torque order, export as the speed of a motor vehicle, therefore the element inputted in Hankel matrix U p, Uf is the column matrix of 2 dimensions, and the element exported in Hankel matrix Yp, Yf is a numerical value;
。
2. predictive equation and checking thereof:
The prediction output equation being gone out system above by the input and output Hankel matrix construction obtained is:
，
(1)
being the following output valve of the system estimated by subspace state space system identification, by solving least square problem, obtaining two prediction matrix coefficient Lw and Lu in predictive equation
(2)
The solution of two prediction matrix coefficient Lw and Lu is tried to achieve by following formula:
(3)
At the Hankel matrix of the input/output information of known system, and after obtaining Lw and Lu, utilize subspace predictive equation (1), the following output valve of the system that namely obtains; Actually during controller application, only have and export Hankel matrix
first row as the estimated valve in system future, greatly reduce calculated amount so on the one hand, on the other hand for the application of Model Predictive Control is got ready.Therefore final prediction of output equation is as formula (4),
it is Hankel matrix
first row
，
(4)
Obtain the prediction output model of system according to excited data, for datadriven predictive control algorithm, forecast model can calculate to a nicety system future export, decide the quality of controller performance.Below, the present invention carries out certain checking by several groups of emulation experiments to the predictive equation obtained.
Fig. 6 is the identification result of one section of excited data of arbitrary extracting from excited data.Can see that predicting that the system obtained exports exports closely with real system.In addition, we have redesigned one group of excitation as shown in Figure 7, and the system of prediction exports the system also well following the tracks of reality and exports.This illustrates that the precision of forecasting model of design is satisfactory.
3. datadriven Design of Predictive: after the prediction output equation that the system of acquisition exports, according to Model Predictive Control Algorithm.By obtaining the drive torque after optimizing and lock torque signal to solving of objective cost function, the mathematical expression form of cost function is as follows:
(5)
From predictive control theory, adopt the predictive controller of incremental, controller can be made like this to have integral action, thus reduce the steady state error of system, improve control effects.Therefore the predictive equation (5) that system exports is rewritten as increment pattern:
；
(6)
Export to carry out adding up to the prediction of newadded item type and obtain following model prediction output equation:
(7)
Wherein,
；
F item is system free response, and correspondingly S uf item is called as the control response of system;
Output model constructs the cost function of following form to utilize incremental to predict:
(8)
Wherein,
，
；
(9)
Wherein, Γ u and Γ y is the weighting matrix of input item and output item respectively; Control time domain is Nu, and predicts that time domain is Np; In the present invention, in the later stage after great many of experiments emulation, selected prediction time domain is 20, and controlling time domain is 4.Here reference value Ref is exactly the desired speed v_q of NEDC state of cyclic operation, is normalized rear realtime update is given in controller it.Coefficient of weight Λ=Γ u/ Γ y, after a large amount of emulation experiment adjustment, selected Λ value is:
Motor maximum torque constraint in moment of torsion optimization system, the maximum braking moment of drg retrain and the constraint of actual vehicle speed is described by mathematic(al) representation below:
(10)。
Final moment of torsion optimization data drives predictive controller model as shown in Figure 8, mainly passes through Matlab
In M file compiler realize.By AMESim and Simulink associative simulation technology, after debugging, as shown in Figure 9, actual vehicle speed overlaps with desired speed the speed of a motor vehicle simulation result figure obtained under initial parameter substantially as seen from the figure, and control effects is good.
Experimental verification
In order to verify that moment of torsion optimization data drives the controller performance of predictive controller further, in centralized electronlmobil model, changing the gradient, car load load and wind speed size respectively and under these three kinds of operating modes of other parameter constants, performance verification has been carried out to moment of torsion optimal controller.
1) different gradient
The hill climbing ability of electronlmobil requires it is generally be more than or equal to 20%(that is 9 degree), learn that the torque peak of motor directly affects grade climbing performance (grade climbing performance here refers to that can realize ENDC desired speed under this gradient follows the tracks of) by experiment, when torque peak is 100Nm, hill climbing ability is 10%, can not realize the tracking of speed after more than 10%; Be 15% during torque peak 200Nm, the tracking of speed after more than 15%, can not be realized; During torque peak 300Nm, maximum grade can reach the requirement of 20%.Setting motor maximum torque in the present invention is 300Nm, and keep other amounts constant, wind speed 0m/s, complete vehicle quality 1000kg, only changes the gradient, is set as 5%, 10%, 20% respectively, and the simulation result obtained as shown in Figure 10.
As seen from the figure, owing to there being the gradient, when just starting uphill starting, vehicle has had the trend of downslide, and the trend of the larger downslide of the gradient is larger, and in reality, uphill starting needs to step on brake, so this is rational in a model.Along with the speed tracing effect of the increase entirety of the gradient is still good, but the rigid constraint of the actr due to motor maximum torque, when the gradient is very large, the speed of a motor vehicle cannot continue to increase after reaching higher limit, thus when speed is higher, the speed of a motor vehicle can only maintain higher limit, and cannot continue to follow the tracks of, but the speed tracing effect below speed limit is fine.Wherein, be about 18m/s(64.8km/h in 20% gradient hourly velocity upper limit) left and right, this is also rational in reality.Illustrate that moment of torsion optimization data drives predictive controller in different gradient situation, still there is good controller performance.
2) different car load load
Keep other amounts constant, the setting gradient is 0, and wind speed is 0m/s, only change complete vehicle quality m, suppose that the average weight of passenger is 75kg, setting complete vehicle quality is respectively 1500kg, (1500+2*75) kg, (1500+4*75) kg, and the speed curves obtained as shown in figure 11.
Under these quality, control effects entirety is all fine as seen from the figure, but along with the increase of quality, deviation can increase, some overshoot have been had for during (1500+4*75) kg in quality, but overall tracking effect is all fine, illustrate that datadriven rolling optimization entire car controller is in different quality situation, still has good controller performance.
3) different wind speed
Keep other amounts constant, the setting gradient is 0, and complete vehicle quality is 1000kg, only changes ambient wind velocity, is set as 10m/s, 20m/s, 30m/s respectively, and the speed of a motor vehicle correlation curve obtained as shown in figure 11.As seen from the figure, after change ambient wind velocity to 30m/s, the tracking effect of the speed of a motor vehicle is still fine, illustrates that moment of torsion optimization data drives predictive controller under different wind conditions, still has good controller performance.
Present invention is directed at centralized electronlmobil and devise moment of torsion optimal controller based on datadriven predictive control, datadriven Subspace Identification and model predictive control method combine by this method cleverly, the predictive equation of system is obtained according to the inputoutput data of system, the control inputs of system is obtained again, the constraint of the simultaneously explicit system that considers by Model Predictive Control optimization.In order to verify the validity of moment of torsion optimal controller, in the senior simulation software of AMESim, build centralized frontdriving electric car model, and demonstrated the performance of controller under different gradient, different car load load, different wind speed operating mode.Simulation result shows, datadriven predictive controller entire car controller is at different quality, and different gradient, under obstructed wind conditions, still has good controller performance.
Claims (1)
1., based on an electric automobile torque optimization method for datadriven predictive control, it is characterized in that: the steps include:
A, centralized electric vehicle simulation model buildings: adopt modular approach, the whole vehicle model of centralized electronlmobil is built in AMESim, whole electric vehicle simulation model mainly comprises Electric drive module, transmission module, tire module and longitudinal direction of car dynam, whole vehicle model parameter is as table one:
Table one electronlmobil parameter list
；
B, moment of torsion optimization data drive Design of Predictive: the datadriven predictive control algorithm of employing is combined by Subspace Identification and Model Predictive Control to form, and concrete step is as follows:
1. the choosing of excited data: choose drive torque order and lock torque order as excitation incoming signal, in AMESim whole vehicle model, by design drive torque and lock torque data to motor and drg, open loop operation model, just the actual vehicle speed measured is obtained, the speed of a motor vehicle that drive torque order, lock torque order and measurement obtain is the final system incentive data needed; Obtain input Hankel matrix U p, the Uf of system according to these data, export Hankel matrix Yp, Yf:
；
2. predictive equation:
The prediction output equation being gone out system by input and output Hankel matrix construction is:
，
(1)
being the following output valve of the system estimated by subspace state space system identification, by solving least square problem, obtaining two prediction matrix coefficient Lw and Lu in predictive equation
(2)
The solution of two prediction matrix coefficient Lw and Lu is tried to achieve by following formula:
(3)
At the Hankel matrix of the input/output information of known system, and after obtaining Lw and Lu, utilize subspace predictive equation (1), the following output valve of the system that namely obtains;
Final prediction of output equation is as formula (4),
it is Hankel matrix
first row
，
(4)
3. datadriven Design of Predictive: obtain the drive torque after optimizing and lock torque signal to solving of objective cost function, the mathematical expression form of cost function is as follows:
(5)
The predictive equation (5) that system exports is rewritten as increment pattern:
；
(6)
Export to carry out adding up to the prediction of newadded item type and obtain following model prediction output equation:
(7)
Wherein,
；
F item is system free response, and correspondingly S uf item is called as the control response of system;
Output model constructs the cost function of following form to utilize incremental to predict:
(8)
Wherein,
，
；
(9)
Wherein, Γ u and Γ y is the weighting matrix of input item and output item respectively; Control time domain is Nu, and predicts that time domain is Np; Coefficient of weight Λ=Γ u/ Γ y, selected Λ value is:
Motor maximum torque constraint in moment of torsion optimization system, the maximum braking moment of drg retrain and the constraint of actual vehicle speed is described by mathematic(al) representation below:
(10)。
Priority Applications (1)
Application Number  Priority Date  Filing Date  Title 

CN201410793965.0A CN104401232B (en)  20141221  20141221  Electric automobile torque optimization method based on datadriven PREDICTIVE CONTROL 
Applications Claiming Priority (1)
Application Number  Priority Date  Filing Date  Title 

CN201410793965.0A CN104401232B (en)  20141221  20141221  Electric automobile torque optimization method based on datadriven PREDICTIVE CONTROL 
Publications (2)
Publication Number  Publication Date 

CN104401232A true CN104401232A (en)  20150311 
CN104401232B CN104401232B (en)  20160622 
Family
ID=52638925
Family Applications (1)
Application Number  Title  Priority Date  Filing Date 

CN201410793965.0A Active CN104401232B (en)  20141221  20141221  Electric automobile torque optimization method based on datadriven PREDICTIVE CONTROL 
Country Status (1)
Country  Link 

CN (1)  CN104401232B (en) 
Cited By (7)
Publication number  Priority date  Publication date  Assignee  Title 

CN106647280A (en) *  20170116  20170510  山东工商学院  HEV (hybrid electric vehicle) mode switching data driving prediction controller design method and HEV mode switching data driving prediction controller design system 
CN107168104A (en) *  20170623  20170915  吉林大学  Pure electric intelligent automobile longitudinal method for controlling driving speed based on observer 
CN107269408A (en) *  20170515  20171020  吉林大学  Diesel engine optimizes combustion controller and to simulation model control method 
CN107323308A (en) *  20170626  20171107  杭州氢途科技有限公司  A kind of fuel cell car tractive torque computational methods with predictive ability 
CN107719187A (en) *  20170930  20180223  合肥巨动力系统有限公司  A kind of method that electric drive system for electric vehicles moment of torsion control improves torque precision 
CN109035477A (en) *  20180704  20181218  浙江中控技术股份有限公司  A kind of fork truck equipment state comprehensive appraisal procedure, apparatus and system 
CN109039168A (en) *  20180713  20181218  浙江工业大学之江学院  The multiple dimensioned approximate explicit model forecast Control Algorithm of brshless DC motor 
Citations (5)
Publication number  Priority date  Publication date  Assignee  Title 

EP2426008A1 (en) *  20100906  20120307  Renault S.A.S.  Method for controlling an electric motor driving an automobile 
CN103192737A (en) *  20130325  20130710  吉林大学  Drive control method for allelectric car 
CN103303159A (en) *  20130509  20130918  常州迈控智能科技有限公司  Electric automobile controller and dualdriving control method thereof 
WO2014034333A1 (en) *  20120830  20140306  日立オートモティブシステムズ株式会社  Electric vehicle drive system 
CN104210383A (en) *  20140918  20141217  上海工程技术大学  Fourwheel independently driven electric vehicle torque distribution control method and system 

2014
 20141221 CN CN201410793965.0A patent/CN104401232B/en active Active
Patent Citations (5)
Publication number  Priority date  Publication date  Assignee  Title 

EP2426008A1 (en) *  20100906  20120307  Renault S.A.S.  Method for controlling an electric motor driving an automobile 
WO2014034333A1 (en) *  20120830  20140306  日立オートモティブシステムズ株式会社  Electric vehicle drive system 
CN103192737A (en) *  20130325  20130710  吉林大学  Drive control method for allelectric car 
CN103303159A (en) *  20130509  20130918  常州迈控智能科技有限公司  Electric automobile controller and dualdriving control method thereof 
CN104210383A (en) *  20140918  20141217  上海工程技术大学  Fourwheel independently driven electric vehicle torque distribution control method and system 
Cited By (8)
Publication number  Priority date  Publication date  Assignee  Title 

CN106647280A (en) *  20170116  20170510  山东工商学院  HEV (hybrid electric vehicle) mode switching data driving prediction controller design method and HEV mode switching data driving prediction controller design system 
CN107269408A (en) *  20170515  20171020  吉林大学  Diesel engine optimizes combustion controller and to simulation model control method 
CN107168104A (en) *  20170623  20170915  吉林大学  Pure electric intelligent automobile longitudinal method for controlling driving speed based on observer 
CN107323308A (en) *  20170626  20171107  杭州氢途科技有限公司  A kind of fuel cell car tractive torque computational methods with predictive ability 
CN107323308B (en) *  20170626  20190625  杭州氢途科技有限公司  A kind of fuel cell car tractive torque calculation method with predictive ability 
CN107719187A (en) *  20170930  20180223  合肥巨动力系统有限公司  A kind of method that electric drive system for electric vehicles moment of torsion control improves torque precision 
CN109035477A (en) *  20180704  20181218  浙江中控技术股份有限公司  A kind of fork truck equipment state comprehensive appraisal procedure, apparatus and system 
CN109039168A (en) *  20180713  20181218  浙江工业大学之江学院  The multiple dimensioned approximate explicit model forecast Control Algorithm of brshless DC motor 
Also Published As
Publication number  Publication date 

CN104401232B (en)  20160622 
Similar Documents
Publication  Publication Date  Title 

CN104401232B (en)  Electric automobile torque optimization method based on datadriven PREDICTIVE CONTROL  
Li et al.  Analysis of downshift’s improvement to energy efficiency of an electric vehicle during regenerative braking  
Chen et al.  Optimal strategies of energy management integrated with transmission control for a hybrid electric vehicle using dynamic particle swarm optimization  
CN105365592A (en)  Residual endurance mileage estimation method and system of electric vehicle and electric vehicle  
Yang et al.  A stochastic predictive energy management strategy for plugin hybrid electric vehicles based on fast rolling optimization  
CN103419588B (en)  The active energy regenerative suspension multilayer controller that threestage damping is adjustable and building method thereof  
Kamal et al.  Intelligent energy management strategy based on artificial neural fuzzy for hybrid vehicle  
CN106080584A (en)  A kind of hybrid vehicle pattern based on Model Predictive Control Algorithm switching control method for coordinating  
Zhao et al.  Parametric optimization of novel electric–hydraulic hybrid steering system based on a shuffled particle swarm optimization algorithm  
Ramakrishnan et al.  Design strategy for improving the energy efficiency in series hydraulic/electric synergy system  
CN103863087A (en)  Plugin hybrid electric vehicle energysaving predictive control method based on optimal engine operation line  
CN104175891A (en)  Pure electric vehicle energy recovery and regeneration brake control method  
Dimitrova et al.  Environomic design of vehicle energy systems for optimal mobility service  
Taghavipour et al.  A comparative analysis of route‐based energy management systems for PHEVs  
CN107168104A (en)  Pure electric intelligent automobile longitudinal method for controlling driving speed based on observer  
Ma et al.  Predictive energysaving optimization based on nonlinear model predictive control for cooperative connected vehicles platoon with V2V communication  
Sun et al.  An integrated control strategy for the composite braking system of an electric vehicle with independently driven axles  
CN104156517A (en)  Batteryelectricvehicle hybrid power system efficiency calculation model and optimization method  
Lv et al.  Cyberphysical system based optimization framework for intelligent powertrain control  
CN106647280B (en)  The datadriven Design of Predictive method and system of HEV mode switching  
Sun et al.  A novel torque coordination control strategy of a singleshaft parallel hybrid electric vehicle based on model predictive control  
Goodarzi et al.  Stability enhancement and fuel economy of the 4wheeldrive hybrid electric vehicles by optimal tyre force distribution  
Sun et al.  Hybrid modeling and predictive control of intelligent vehicle longitudinal velocity considering nonlinear tire dynamics  
Horrein et al.  Forward and backward simulations of a power propulsion system  
CN104925063A (en)  Model predictive control method for electromechanical composite transmission vehicle 
Legal Events
Date  Code  Title  Description 

C06  Publication  
PB01  Publication  
C10  Entry into substantive examination  
SE01  Entry into force of request for substantive examination  
C14  Grant of patent or utility model  
GR01  Patent grant 