CN106130125A - Electric automobile fuzzy sliding mode feedback charge controller and feedback charge control method thereof - Google Patents

Abstract

The invention discloses a kind of electric automobile fuzzy sliding mode feedback charge controller and feedback charge control method thereof.Electric automobile fuzzy sliding mode feedback charge controller of the present invention; including storage battery charge controller; the outfan of storage battery charge controller is sequentially connected with driving/isolation circuit, regenerative braking charging system, sensing acquisition circuit, and sensing acquisition circuit finally connects back to storage battery charge controller.System model stage when feedback charge control method of the present invention includes setting up DC/DC changer feedback charging, the DC/DC changer T S fuzzy model stage set up under feedback charging, the calculating predictive value stage of changer dutycycle, fuzzy sliding mode charge controller comprehensive Design stage.Control method of the present invention has the strongest robustness, it is possible to reclaim more braking energy, can be effectively improved the course continuation mileage number of electric automobile；Controller architecture of the present invention is simple, with low cost, reliability is high.

Description

Electric automobile fuzzy sliding mode feedback charge controller and feedback charge control method thereof

Technical field

The invention belongs to charging electric vehicle and control technical field, be specifically related to a kind of two-way DC/DC used for electric vehicle and become The T-S fuzzy variable structure feedback charge control method of parallel operation, and Fuzzy Sliding Model Controller.

Background technology

Pure electric automobile is the green traffic instrument of a kind of zero-emission, no pollution, in country's new-energy automobile support policy Support under, increasingly welcome by consumer.But that brings because of not enough by battery capacity in charging modes inconvenience is continuous Boat mileage is not enough, the most seriously governs the development of pure electric automobile.In the case of battery technology bottleneck is difficult to substantive breakthroughs, Feedback based on energy regeneration charging has obtained widely as the technology of a kind of effective prolongation pure electric automobile course continuation mileage number Research.

Technically, feedback charging realization it needs to be determined that power converter topologies structure, design charge control method. During feedback charging, owing to back-emf amplitude is generally below battery open-circuit voltage, the main work of DC/DC changer in electric automobile It is used as booster converter and output voltage is promoted to sufficiently high level；And during driven, by the energy in accumulator with Desired size is supplied to motor so that vehicle travels advance.At present, in the case of need not design single control circuit, greatly Most automobile-used DC/DC changers can not realize energy two-way flow between accumulator and motor；Minority can realize two-way flow Changer, owing to needing more component, cause that structure is complicated, cost increases.

And on charge control method, major part remains in needs the Traditional control of mathematical models technical.So And, rely solely on traditional control device and be difficult to obtain higher control performance.This be due to: on the one hand, include motor, change The charging system of parallel operation, accumulator and car body has the electromechanical Coupled Dynamics feature of complexity, when electric automobile runs, motor Parameter, the factors such as road conditions, load, driving model and cell voltage that travel are all changes, it is difficult to set up it and count accurately Learn model.On the other hand, most charging controls are all based on small-signal analysis method, i.e. enter at certain stable operating point Line linearityization processes, it is difficult to meet system linearity requirements under speed on a large scale, overall situation working environment.Meanwhile, electricity Electrical automobile is under complicated driving cycle, because temperature rise, vehicle parameter uncertainty, uncertain input voltage, output loading change Equivalent load resistance change when charging with battery, causes traditional controller hydraulic performance decline, and control system lacks robustness.

Therefore, design simple in construction, control accuracy is high, stability is strong and the feedback charging controlling party of high efficiency of energy utilization Method, to improving, the safety of electric automobile during traveling, stability and course continuation mileage number are significant.

Summary of the invention

First purpose of the present invention is to provide a kind of electric automobile fuzzy sliding mode feedback charge controller, this electronic vapour Car fuzzy sliding mode feedback charge controller, has control accuracy height, strong robustness, the advantage of simple in construction, thus can effectively carry The course continuation mileage number of high electric automobile.

The above-mentioned purpose of the present invention realizes by the following technical solutions: this electric automobile fuzzy sliding mode feedback is filled Electric controller, it includes storage battery charge controller, and the outfan of storage battery charge controller is sequentially connected with driving/isolation electricity Road, regenerative braking charging system, sensing acquisition circuit, sensing acquisition circuit finally connects back to storage battery charge controller.

Concrete, described storage battery charge controller is realized by digital signal processor DSP (TMS320F2812).

Concrete, described regenerative braking charging system include an accumulator, four current sensor SA1～SA4, two Voltage sensor SV1, SV2, storage capacitor C, two-way DC/DC changer and the dc motor of Y-connection；Accumulator upper End series-connected current sensors SA1 posterior end parallel connection is connected to voltage sensor SV1；Storage capacitor C is in parallel with voltage sensor SV2, Again with release can resistance R connect after in parallel with voltage sensor SV1；Two-way DC/DC changer is made up of six power tube T1～T6, Upper brachium pontis is T1, T3 and T5, and lower brachium pontis is T4, T2 and T6, and every bridging meets midpoint series-connected current sensors SA2, SA3, SA4 respectively It is connected with the phase winding of dc motor afterwards.

Concrete, described driving/isolation circuit includes photoelectric isolating device U1 and drives control circuit U2, photoisolator Part U1 uses photoelectric coupling chip 4N25, drives control circuit U2 to use chip I R2122S.

Concrete, described sensing acquisition circuit includes voltage, current sensor and two channel operation amplifier U4, U5, electricity Pressure, current sensor use HAS200-P, and operational amplifier uses LF353.

Second object of the present invention is that providing based on above-mentioned electric automobile fuzzy sliding mode feedback charge controller returns Feedback charge control method, system model stage when the method includes setting up DC/DC changer feedback charging, sets up feedback charging Under the DC/DC changer T-S fuzzy model stage, calculate the predictive value stage of changer dutycycle, fuzzy sliding mode charging control The device comprehensive Design stage；

(1) step setting up system model stage when DC/DC changer feedback is charged described in is as follows:

(1) system model of charging circuit is calculated；

Same set of DC/DC translation circuit under feedback charge condition, in the case of use and driven；Use six pipe full-bridges Pulse width modulation, it is not necessary to extra circuits unit；As a example by A phase and B phase, analyze mathematical model, open at switch, turn off In the case of two kinds, research controls the mathematical model of system when target is accumulator charging voltage, electric current, to be output as battery charging Voltage y=V_oIllustrate as a example by (t)；Writ state variable x=[i_L v_c]^T, the system model of circuit is given by following equation:

$\{\begin{array}{c}\stackrel{\·}{x}=A_{ON}x+B_{ON}{e}_{ab}+g_{ON}\\ \stackrel{\·}{x}=A_{OFF}x+B_{OFF}{e}_{ab}+g_{OFF}\end{array},\{\begin{array}{c}y=C_{ON}x+f_{ON}\\ y=C_{OFF}x+f_{OFF}\end{array};$

Formula breaker in middle opens, turn off under matrix be respectively

C_ON=[0 R_bR^-1],

C_OFF=[R_bR_cR^-1 R_bR^-1], f_ON=f_OFF=R_cR^-1v_b；Coefficient a₁=-(2R_m+R_s+R_d)/2L_m,

a₂=-1/CR, a₃=-(R_m+R_d)/L_m-R_cR_b/(2L_mR), a₄=R_b/(2L_mR), g₁=-a₂v_b,

g₂=-R_c/(2L_mR)v_b；Symbol Lm is winding inductance, R_mIt is winding resistance, R_sAnd R_dIt is that on and off switch is with continuous respectively The conducting resistance of stream diode, C and R_cIt is battery DC side capacitors electric capacity and dead resistance, R respectively_bIt is battery equivalent internal resistance, R=R_c+R_b, v_cIt is the voltage drop on capacitor, v_bRepresent cell emf, i_bFor flowing through the feedback charging current of battery, e_abFor Two phase winding counter electromotive force, v_oFor output voltage；

Output equation in circuit model is charging voltage equation, and when being output as charging current, its output equation is rewritten For i_o=-R^-1v_c+R^-1v_b；

(2) space State Average Model is calculated；

PWM duty cycle d (t) and d'(t it is multiplied by respectively on circuit model both sides)=1-d (t), and carry out handling averagely, Trying to achieve space State Average Model is:

$\left\{\begin{array}{c}{\stackrel{\·}{x}}_m=\stackrel{\‾}{A}{x}_m+\stackrel{\‾}{B}{e}_{ab}+\stackrel{\‾}{g}\\ y_m={\stackrel{\‾}{C}}_m{x}_m+\stackrel{\‾}{f}\end{array}\right.;$

Matrix in formula x_m、y_mIt is the state variable meansigma methods in single PWM cycle and output voltage average value respectively；Right In given dutycycleOrderCan try to achieveThe state variable steady-state value of place's quiescent point is:

$\left\{\begin{array}{c}{x}_m=\stackrel{\‾}{x}=-{\stackrel{\‾}{A}}^{-1}(\stackrel{\‾}{B}{e}_{ab}+\stackrel{\‾}{g})\\ y_m=\stackrel{\‾}{y}=-{\stackrel{\‾}{CA}}^{-1}(\stackrel{\‾}{B}{e}_{ab}+\stackrel{\‾}{g})\end{array}\right.;$

In formulaWithRepresent x respectively_mAnd y_mSteady-state value；Due in operating pointThere is the dry of small-signal in place Disturb, then the instantaneous value of variable can be written as:Wherein d (t), x T (), y (t) are variable instantaneous value,For small-signal disturbance；

(3) state space small-signal and integration control model are calculated；

Utilize small-signal perturbation analysis method to isolate steady-state variable and transient state variable, ignore disturbance quantity secondary and more than Higher order term, trying to achieve state space small-signal model is:

$\left\{\begin{array}{c}{\stackrel{\·}{\hat{x}}}_m\left(t\right)=\stackrel{\‾}{A}{\hat{x}}_m\left(t\right)+\stackrel{\‾}{E}\hat{d}\left(t\right)\\ {\hat{y}}_m\left(t\right)=\stackrel{\‾}{C}{\hat{x}}_m\left(t\right)\end{array}\right.;$

Matrix in formula

For State space averaging and two kinds of models of state space small-signal, all rely on dutycycle stable state at operating point ValueFor realizing zero steady track error of output voltage, introduce integrating state variable: x_e=∫ e dt=∫ (y_r-y_m) dt, Tracking error e=y in formula_r-y_m, y_rFor expectation output voltage；State space small-signal model is rewritten as following integration control mould Type:

$\left\{\begin{array}{c}{\stackrel{\·}{\hat{x}}}_a\left(t\right)={\stackrel{\‾}{A}}_a{\hat{x}}_a\left(t\right)+{\stackrel{\‾}{E}}_a\hat{d}\left(t\right)\\ {\hat{y}}_m\left(t\right)={\stackrel{\‾}{C}}_a{\hat{x}}_a\left(t\right)\end{array}\right.$

In formulaFor augmented state variable, control input for dutycycle transient valueMatrix

${\stackrel{\‾}{A}}_a=\left[\begin{array}{cc}{0}_{1\×2}& -1\\ 0_{2\×1}& \stackrel{\‾}{A}\end{array}\right],{\stackrel{\‾}{E}}_a=\left[\begin{array}{c}0\\ \stackrel{\‾}{E}\end{array}\right],{\stackrel{\‾}{C}}_a=\left[\begin{array}{c}0\\ \stackrel{\‾}{C}\end{array}\right];$

In vehicle travel process, along with the change of operating point, the dutycycle of changerChange therewith, thus cause Zero pole point and the amplitude-frequency response of state space small-signal model transmission function change, thus state space small-signal model is The nonlinear function of dutycycle, charge control system is a Nonlinear Uncertain Systems；

(2) step of the described DC-DC converter T-S fuzzy model set up under feedback charging is as follows:

(1) T-S fuzzy model；Utilize T-S fuzzy technology to approach nonlinear system, for i-th operating point, use as follows IF-THEN rule describes nonlinear state space small-signal model:

Article i-th, model rule: if z₁T () is F₁ ⁱ, and z₂T () is..., and z_nT () isSo:

In formulaFor fuzzy set, z (t)=[z₁,…,z_n] it is former piece variable,It is that state is missed Difference,It is to control input, A_iAnd B_iFor treating set matrix, rule number i=1,2 ..., r；Fuzzy weighting valueF_i ^j[z_j(t)] >=0 it is z_jT () is right under i-th fuzzy rule The degree of membership answered, and haveBased on single-point obfuscation, product inference and weighted average anti fuzzy method, overall situation mould Fuzzy model is:

(2), after the various interference being subject to during consideration charging and uncertainty, Parameter uncertainties fuzzy model is:

$\stackrel{\·}{\hat{x}}\left(t\right)=\underset{i=1}{\overset{r}{\Σ}}{h}_i\left(z\right)\[A_i+{\mathrm{\ΔA}}_i\left(t\right)\]\hat{x}\left(t\right)+\underset{i=1}{\overset{r}{\Σ}}{h}_i\left(z\right)\[B_i+{\mathrm{\ΔB}}_i\left(t\right)\]\hat{d}\left(t\right)+\underset{i=1}{\overset{r}{\Σ}}{h}_i\left(z\right)w_i(t,\hat{x});$

Δ A in formula_iWith Δ B_iFor the matching uncertainties of parameter,Represent input and load disturbance；Assume: (i) There is definitiveness function Μ_A(t), Μ_B(t) and Μ_wT () makes Δ A_i=B_iΜ_A(t), Δ B_i=B_iΜ_B(t) andAll set up；(ii) system control matrix is unsatisfactory for B₁=B₂=...=B_r；Rewrite Parameter uncertainties mould Type is as follows:

$\stackrel{\·}{\hat{x}}\left(t\right)=\underset{i=1}{\overset{r}{\Σ}}{h}_i\left(z\right)A_i\hat{x}\left(t\right)+\[B+\stackrel{\‾}{B}\stackrel{\‾}{H}\left(h\right(z\left)\right)\stackrel{\‾}{I}\].\[\hat{d}+g(t,\hat{x},\hat{d})\];$

In formula: External disturbanceThere will necessarily be one Know normal number η_BMake 0≤| | Μ_B||≤η_B< 1, and positive function η continuously_A、η_wMake Then can calculateThe upper bound of norm is

(3) determination of T-S model；[0,1] between the global work area of vehicle driving-cycle is divided into 7 sub spaces, point It is not: [0,0.2], [0.2,0.3], [0.3,0.4], [0.4,0.5], [0.5,0.6], [0.6,0.7], [0.7,1]；Each Subspace is chosen a steady operation point, determines 7 stable operating points as the following formula:

D in formula_iFor the upper bound, subspace, d_iFor subspace lower bound； Utilize T-S modeling method to obtain 7 linear submodels at above-mentioned steady operation point, be described as following T-S fuzzy rule:

Article i-th, object-rule: ifis Fⁱ, then

F in formulaⁱ(i=1～7) is fuzzy set,

(3) step in the predictive value stage of described calculating changer dutycycle is as follows:

Output voltage-input voltage transfer ratioIt is the nonlinear function of dutycycle, owing to having strong non-linear spy Property and parameter uncertainty, utilize T-S fuzzy close method to predict dutycycle；Prediction process is as follows: first, by transfer ratio district Between be divided into 12 subinterval (S₁,S₂,…,S_n), each subinterval S_iDefine an affine function；Then, this affine letter is utilized Number calculates the dutycycle predictive value on each subintervalFinally, T-S technology is utilized to incite somebody to actionJoin together, calculate Global duty cycle

T-S predictor is the affine function of a single-input single-output, inputs as α=y_r/e_abAnd meet 1≤α≤M, wherein M For transfer ratio functionMaximum；Make f_iIt is_iThe output (1≤i≤12) in individual subinterval, its form is: f_i(α)=a_i α+b_i, wherein a_i, b_iFor constant；Then desired output voltage is y_rTime, T-S predictor fuzzy rule is:

Article i-th, predictor rule: if α is is S_iSo f_i(α)=a_iα+b_i；

S in formula_iFor i-th fuzzy set, the affine function on each subinterval is exported f_iCarry out that center is average, weighting reverse Gelatinizing, then overall situation fuzzy output isμ in formula_i(α) it is that α is at fuzzy subset S_iOn Membership function, and have

(4) step in described fuzzy sliding mode charge controller comprehensive Design stage is as follows:

(1) based on sliding mode control theory and Lyapunov method design Integral Sliding Mode diverter surface:

$\mathrm{\&sigma;}(t,\hat{x})=S\&CenterDot;\hat{x}\left(t\right)+\mathrm{\&lambda;}\&Integral;S\hat{x}\left(t\right)dt---\left(17\right)$

Constant λ in formula > 0 is storage gain, sliding formwork coefficient S ∈ R^m×n, sliding mode control theory requires that the selection of coefficient S needs Guarantee the existence of Equivalent Sliding Mode controlled quentity controlled variable, i.e. matrixMust be reversible；To this end, based on Lyapunov Method provides the computational methods of coefficient S and sliding-mode surface；

Based on Lyapunov method, if following linear inequality

$\left[\begin{array}{ccc}{K}^T(A_{i}Q+{\mathrm{QA}}_i^T)K& *& *\\ \mathrm{\&mu;}{\stackrel{\&OverBar;}{B}}^{T}K& -I& *\\ A_{i}QK& \mathrm{\&eta;}\stackrel{\&OverBar;}{B}& -I\end{array}\right]<0,\&ForAll;i$

$\left[\begin{array}{cccc}Q& I& *& *\\ I& \mathrm{\&alpha;}I& *& *\\ 0& 0& \mathrm{\&beta;}I-Q& *\\ 0& 0& 0& 2\sqrt{{\mathrm{\&lambda;}}_B}\mathrm{\&mu;}-\sqrt{r}(\mathrm{\&alpha;}+\mathrm{\&beta;})\end{array}\right]>0;$

$\mathrm{\&mu;}(1+{\mathrm{\&eta;}}_B)\left|\right|\stackrel{\&OverBar;}{B}\left|\right|-(1-{\mathrm{\&eta;}}_B)<0;$

There is feasible solution (Q, α, β, μ), then design sliding formwork coefficient is S=(B^TQ^-1B)^-1B^TQ^-1, wherein invertible matrix Q ∈ R^{n ×n}, decision variable α, β, μ ∈ R, K are the orthogonal complement matrix of matrix B, λ_BIt it is matrix B^TThe minimal eigenvalue of B, and meet λ_BI≤ B^TB, the transposition of mark " * " representing matrix relevant position element；The advantage choosing Integral Sliding Mode face is to can guarantee that closed loop control system The stable state charging voltage error of system is zero；

(2) based on sliding mode control theory and Lyapunov method design control law, Lyapunov function is designed as: V₂=σ^Tσ The derivative of >=0, Lyapunov function against time is:

In order to protect CardDesign following fuzzy sliding mode tracking control regular:

Article i-th, control rule: ifisSo

${\hat{d}}_i\left(t\right)=-{\mathrm{SA}}_i\&CenterDot;\hat{x}\left(t\right)-\mathrm{\&lambda;}S\hat{x}\left(t\right)-\frac{1}{1-\mathrm{\&xi;}}{\mathrm{\&rho;}}_i(t,\hat{x})\&CenterDot;\mathrm{sgn}\left(\mathrm{\&sigma;}\right)$

In formula, ξ=η_B+τ+η_Bτ, sliding formwork handoff gain Constant ε_i> 0, sgn (σ) is sign function；The fuzzy sliding mode charge controller of the design overall situation is:

$\hat{d}\left(t\right)=-\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\&lsqb;{\mathrm{SA}}_i\&CenterDot;\hat{x}\left(t\right)-\mathrm{\&lambda;}S\hat{x}\left(t\right)-\frac{1}{1-\mathrm{\&xi;}}{\mathrm{\&rho;}}_i(t,\hat{x})\&CenterDot;\mathrm{sgn}\left(\mathrm{\&sigma;}\right)\&rsqb;;$

Based on Lyapunov stability law, nowDemonstrate when using the overall situation Fuzzy sliding mode charge controller time, control system feedback charging voltage error by asymptotic convergence in zero.

Compared with prior art and controller, control method of the present invention is embodied in as follows with the advantage of Fuzzy Sliding Model Controller Some:

(1) due to the fact that have employed T-S method fills motor in electric automobile, DC/DC changer, the feedback of battery composition Electricity system carries out obscurity model building, thus can realize the overall situation under vehicle complexity operating condition, large-scale linearization modeling, and model Accuracy is high.

(2) due to the fact that and have employed fuzzy sliding mode variable structure control, solve Traditional control side under multiple driving cycle When method is because vehicle parameter is uncertain, speed changes the uncertain input voltage, output loading change and the battery charging that cause The equivalent load resistance controller performance that causes of change decline, the feedback charge control method of the present invention has the strongest robust Property, it is possible to reclaim more braking energy.

(3) present invention need not structurally change existing electric vehicle controller, extra hard without increasing Part circuit, it is achieved that only rely under one group of six pipe full bridge PWM control, energy two-way flow between motor and battery, structure letter Single, with low cost, reliability is high.

Accompanying drawing explanation

Fig. 1 is the theory structure block diagram of embodiment of the present invention fuzzy sliding mode feedback charging method.

Fig. 2 is the circuit diagram of accumulator regenerative braking charging system 3 in Fig. 1.

Fig. 3 is Fig. 2 breaker in middle T4 equivalent circuit diagram when opening.

Equivalent circuit diagram when Fig. 4 is Fig. 2 breaker in middle T4 shutoff.

Fig. 5 is the membership function figure of Global fuzzy model in Fig. 1.

Fig. 6 is the T S linear world model illustrative view of functional configuration of regenerative braking charging system 3 in Fig. 1.

Fig. 7 is the membership function figure of fuzzy predictor in Fig. 1.

Fig. 8 is the circuit diagram that in Fig. 1, pwm signal drove/isolated circuit 4.

Fig. 9 is the circuit diagram of sensing acquisition circuit 2 in Fig. 1.

Figure 10 is the flow chart of embodiment of the present invention feedback charge control method.

Figure 11～Figure 13 is the step response curve figure that the inventive method charging controls effect.

Figure 14～Figure 16 is the sinusoidal response curve chart that the inventive method charging controls effect.

In figure, 1 is storage battery charge controller based on DSP, and 2 is sensing acquisition circuit, and 3 is regenerative braking Charging System, 4 for driving/isolation circuit.

Detailed description of the invention

Below in conjunction with Figure of description Fig. 1 to Figure 16 and specific embodiment, the inventive method and fuzzy sliding mode feedback are charged Controller is described in detail.

The present embodiment method includes: sets up system model stage when DC/DC changer feedback is charged, set up feedback charging Under the DC/DC changer T-S fuzzy model stage, calculate the predictive value stage of changer dutycycle, fuzzy sliding mode charging control The device comprehensive Design stage.

Specifically follow the steps below:

One, system model during DC/DC changer feedback charging is set up

Same set of DC/DC translation circuit under feedback charge condition, in the case of the present embodiment use and driven.I.e. adopt With the DC/DC translation circuit of six pipes full-bridge pulse width modulation (PWM), and without extra circuits unit.Divide with small-signal Analysis method is means, determines system that motor, DC/DC changer, battery the form desired electrical under energy regeneration feedback is charged Road model.During feedback charging, as a example by A phase and B phase, analyze mathematical model, now counter electromotive force e_abIt is equivalent to voltage source, switch T4 makees periodically pulsewidth modulation, and other switches are both off.When speed is constant, back-emf e_abAmplitude maintain constant.At switch T4 opens, turn off two kinds in the case of, research controls the mathematical model of system when target is accumulator charging voltage, electric current.

1, T4 opens period:

Seeing Fig. 3, now T4 turns on, other switch off, and winding inductance Lm absorbs the energy from counter electromotive force release, Winding terminal voltage is made to increase, electric current i_LFlow through switch T4 and sustained diode 2, as shown in Figure 1.At a PWM cycle T_sIn, According to Kirchhoff's law, by control target for accumulator charging voltage as a example by, circuit state equation is:

$\left\{\begin{array}{c}2L_m\frac{{\mathrm{di}}_L}{dt}=-(2R_m+R_s+R_d)i_L+e_{ab}\\ C\frac{{\mathrm{dv}}_c}{dt}=-R^{-1}{v}_c+R^{-1}{v}_b\\ v_o=R_b{R}^{-1}{v}_c+R_c{R}^{-1}{v}_b\end{array}\right.---\left(1\right)$

Wherein Lm is winding inductance, R_mIt is winding resistance, R_sAnd R_dIt is the electric conduction of on and off switch and fly-wheel diode respectively Resistance, C and R_cIt is battery DC side capacitors electric capacity and dead resistance, R respectively_bIt is battery equivalent internal resistance, R=R_c+R_b, v_cIt is electric capacity Voltage drop on device, v_bRepresent cell emf, i_bFor flowing through the feedback charging current of battery, e_abIt is that two phase windings are the most electronic Gesture, v_oFor output voltage.Output equation in formula (1) is charging voltage equation, when being output as charging current, and its output equation Can readily be rewritten as i_o=-R^-1v_c+R^-1v_b。

2, during T4 turns off:

See Fig. 4, when winding terminal voltage rises sufficiently high, turn off all switches, start battery is charged.Electric current i_LStream Cross sustained diode 1 and D2, thus brshless DC motor energy feeding telegram in reply pond, as shown in Figure 2.Same, circuit state side Cheng Wei:

$\left\{\begin{array}{c}2L_m\frac{{\mathrm{di}}_L}{dt}=-(2R_m+2R_d+R_c{R}_b{R}^{-1})i_L+R_b{R}^{-1}{v}_c+R_c{R}^{-1}{v}_b+e_{ab}\\ C\frac{{\mathrm{dv}}_c}{dt}=R_b{R}^{-1}{i}_L-R^{-1}{v}_c+R^{-1}{v}_b\\ v_o=R_b{R}_c{R}^{-1}{i}_L+R_b{R}^{-1}{v}_c+R_c{R}^{-1}{v}_b\end{array}\right.---\left(2\right)$

Writ state variable x=[i_L v_c]^T, it is output as battery charging voltage y=v_oT (), by state equation and output equation (1), (2) are converted into following form:

$\{\begin{array}{c}\stackrel{\&CenterDot;}{x}=A_{ON}x+B_{ON}{e}_{ab}+g_{ON}\\ \stackrel{\&CenterDot;}{x}=A_{OFF}x+B_{OFF}{e}_{ab}+g_{OFF}\end{array},\{\begin{array}{c}y=C_{ON}x+f_{ON}\\ y=C_{OFF}x+f_{OFF}\end{array}---\left(3\right)$

Formula breaker in middle opens, turn off under matrix be respectively C_ON=[0 R_bR^-1], C_OFF=[R_bR_cR^-1 R_bR^-1], f_ON=f_OFF=R_cR^-1v_b；Coefficient a₁ =-(2R_m+R_s+R_d)/2L_m, a₂=-1/CR, a₃=-(R_m+R_d)/L_m-R_cR_b/(2L_mR), a₄=R_b/(2L_mR), g₁=-a₂v_b, g₂ =-R_c/(2L_mR)v_b。

Dutycycle d (t) and the d'(t of PWM it is multiplied by respectively on equation (3) both sides)=1-d (t), and average Processing, trying to achieve space State Average Model is:

$\left\{\begin{array}{c}{\stackrel{\&CenterDot;}{x}}_m=\stackrel{\&OverBar;}{A}{x}_m+\stackrel{\&OverBar;}{B}{e}_{ab}+\stackrel{\&OverBar;}{g}\\ y_m={\stackrel{\&OverBar;}{C}}_m{x}_m+\stackrel{\&OverBar;}{f}\end{array}\right.---\left(4\right)$

Matrix in formula x_m、y_mIt is respectively the state variable meansigma methods in a PWM cycle and output voltage average value.For Certain given dutycycleOrderQuiescent point can be tried to achieveThe state variable steady-state value at place is

$\left\{\begin{array}{c}{x}_m=\stackrel{\&OverBar;}{x}=-{\stackrel{\&OverBar;}{A}}^{-1}(\stackrel{\&OverBar;}{B}{e}_{ab}+\stackrel{\&OverBar;}{g})\\ y_m=\stackrel{\&OverBar;}{y}=-{\stackrel{\&OverBar;}{CA}}^{-1}(\stackrel{\&OverBar;}{B}{e}_{ab}+\stackrel{\&OverBar;}{g})\end{array}\right.---\left(5\right)$

In formulaWithRepresent x respectively_mAnd y_mSteady-state value.Due in operating pointThere is small-signal interference in place, Then the instantaneous value of variable can be written as:

$d\left(t\right)=\stackrel{\&OverBar;}{d}+\hat{d}\left(t\right),x\left(t\right)=\stackrel{\&OverBar;}{x}+\hat{x}\left(t\right),y\left(t\right)=\stackrel{\&OverBar;}{y}+\hat{y}\left(t\right)---\left(6\right)$

Wherein d (t), x (t), y (t) are the instantaneous value of variable,Small-signal disturbance quantity for variable.

Utilize small-signal perturbation analysis method to isolate steady-state value and the instantaneous value of variable, ignore disturbance quantity secondary and with Upper higher order term, trying to achieve state space small-signal model is:

$\left\{\begin{array}{c}{\stackrel{\&CenterDot;}{\hat{x}}}_m\left(t\right)=\stackrel{\&OverBar;}{A}{\hat{x}}_m\left(t\right)+\stackrel{\&OverBar;}{E}\hat{d}\left(t\right)\\ {\hat{y}}_m\left(t\right)=\stackrel{\&OverBar;}{C}{\hat{x}}_m\left(t\right)\end{array}\right.---\left(7\right)$

Matrix in formulaFormula (4) and (7) two kinds of models of formula are really established a capital and are depended on dutycycle stable state Value After calculating, then calculate operating point steady-state valueFor realizing zero steady track error of output voltage, introduce Following integrating state variable:

x_e=∫ e dt=∫ (y_r-y_m)·dt (8)

Tracking error e=y in formula_r-y_m, y_rFor expectation output voltage.Then state space small-signal model (7) be rewritten as Lower integral Controlling model:

$\left\{\begin{array}{c}{\stackrel{\&CenterDot;}{\hat{x}}}_a\left(t\right)={\stackrel{\&OverBar;}{A}}_a{\hat{x}}_a\left(t\right)+{\stackrel{\&OverBar;}{E}}_a\hat{d}\left(t\right)\\ {\hat{y}}_m\left(t\right)={\stackrel{\&OverBar;}{C}}_a{\hat{x}}_a\left(t\right)\end{array}\right.---\left(9\right)$

In formulaFor augmented state variable, control input for dutycycle instantaneous valueMatrix

In vehicle travel process, along with the change of operating point, changer dutycycleValue changes therewith, thus causes Formula (7) the transmission limit of function, a Right-half-plant zero and amplitude-frequency response change, thus state space small-signal model Being the nonlinear function of dutycycle, integration control model is a uncertain nonlinear system.

Two, the DC/DC changer T-S fuzzy model under feedback charging is set up

In the case of electric automobile speed and load wide variation, the state space small-signal mould of feedback charging system Type is the nonlinear function of dutycycle, and classical control theory is difficult to effectively process this strong nonlinearity and brings control performance not Profit impact.Takagi-Sugeno (hereinafter referred to as T-S) fuzzy technology is a kind of nonlinear system to be carried out the effective of obscurity model building Means, its key Design thought is to be local linear subsystem by nonlinear system by non-linear membership function approximate description Smooth weighted sum, be proved this that be approximately consistent asymptotic expansion in theory, and its stability analysis can be by Lyapunov Method directly proves.

The present embodiment chooses N number of stable operating point, and being divided between the global work area of vehicle driving-cycle, N number of son is empty Between, utilize T-S modeling method to obtain the N number of linear submodel at equilibrium point, by Parameter Perturbation, input voltage and output loading Change, equivalent load resistance changing factor all regard as system interference, then utilize calculated with weighted average method to go out global linearization Model.

1. theoretical according to T-S fuzzy close, the integration control model of DC/DC changer feedback charging system can be by T-S Fuzzy system is infinitely approached, and for i-th operating point, uses following IF-THEN rule to describe state space small-signal model Linear submodel:

Article i-th, model rule: if z₁T () isAnd z₂T () is..., and z_nT () isSo

$\stackrel{\&CenterDot;}{\hat{x}}\left(t\right)=A_i\hat{x}\left(t\right)+B_i\hat{d}\left(t\right)---\left(10\right)$

In formulaFor fuzzy set, z (t)=[z₁,…,z_n] it is former piece variable,It is state error,It is to control input, A_iAnd B_iFor treating set matrix, rule number i=1,2 ..., r.

Ambiguity in definition weights h_i[z (t)], can be abbreviated as h_i(z):

$h_i\&lsqb;z\left(t\right)\&rsqb;=\underset{j=1}{\overset{n}{\&Sigma;}}{F}_i^j\&lsqb;z_j\left(t\right)\&rsqb;/\underset{i=1}{\overset{r}{\&Sigma;}}\underset{j=1}{\overset{n}{\&Sigma;}}{F}_i^j\&lsqb;z_j\left(t\right)\&rsqb;,---\left(11\right)$

F in formula_i ^j[z_j(t)] >=0 it is z_jT degree of membership that () is corresponding under i-th fuzzy rule, and have

Based on single-point obfuscation, product inference and weighted average anti fuzzy method, formula (10) and formula (11) can calculate entirely Office's fuzzy model is:

$\stackrel{\&CenterDot;}{\hat{x}}\left(t\right)=\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\&lsqb;A_i\hat{x}\left(t\right)+B_i\hat{d}\left(t\right)\&rsqb;---\left(12\right)$

2., after the various interference being subject to during consideration charging electric vehicle described previously and uncertainty, system (12) is further It is rewritten as:

$\stackrel{\&CenterDot;}{\hat{x}}\left(t\right)=\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\&lsqb;A_i+{\mathrm{\&Delta;A}}_i\left(t\right)\&rsqb;\hat{x}\left(t\right)+\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\&lsqb;B_i+{\mathrm{\&Delta;B}}_i\left(t\right)\&rsqb;\hat{d}\left(t\right)+\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)w_i(t,\hat{x})---\left(13\right)$

Δ A in formula_iWith Δ B_iFor coupling parameter uncertainty,Represent input and load disturbance.

Assume: (i) exists definitiveness function Μ_A(t), Μ_B(t) and Μ_wT () makes Δ A_i=B_iΜ_A(t), Δ B_i=B_i Μ_B(t) andAll set up；(ii) system control matrix is unsatisfactory for B₁=B₂=...=B_r, for i=1, 2,…,r.Then system (9) is rewritten into:

$\stackrel{\&CenterDot;}{\hat{x}}\left(t\right)=\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)A_i\hat{x}\left(t\right)+\&lsqb;B+\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{H}\left(h\right(z\left)\right)\stackrel{\&OverBar;}{I}\&rsqb;.\&lsqb;\hat{d}+g(t,\hat{x},\hat{d})\&rsqb;---\left(14\right)$

Matrix in formulaThen MeetExternal disturbance functionThere will necessarily be one known just Constant η_BMake 0≤| | Μ_B||≤η_B< 1, and two continuous positive functions η_A, η_wMake Then can calculate functionThe norm upper bound is

The determination of 3.T-S model.The present embodiment makes N=7, [0,1] between the global work area of vehicle driving-cycle is drawn It is divided into 7 sub spaces, is respectively as follows: [0,0.2], [0.2,0.3], [0.3,0.4], [0.4,0.5], [0.5,0.6], [0.6, 0.7], [0.7,1].Corresponding chooses a steady operation point in every sub spaces, determines 7 steady operations the most as the following formula Point:D in formula_iFor the upper bound, subspace, d_iFor subspace lower bound.Utilize T-S modeling side Method obtains 7 linear submodels at above-mentioned steady operation point, is described as following T-S fuzzy rule:

Article i-th, object-rule: ifis Fⁱ, then

F in formulaⁱ(i=1～7) is fuzzy set,

In the present invention, the membership function of Global fuzzy model is as it is shown in figure 5, Fig. 6 is the T S of regenerative braking charging system Linear world model functional structure.

Three, the predictive value of changer dutycycle is calculated

Output voltage-input voltage transfer ratioIt is the nonlinear function of dutycycle, it may be assumed that

Then given expectation charging voltage y_r, the dutycycle of PWM can be calculated by the equationDue to transfer ratio letter NumberHaving strong nonlinear characteristic, and there is parameter uncertainty, it is pre-that the present embodiment utilizes T-S fuzzy close method Survey dutycycle.Prediction process is as follows: first, is 12 subintervals by transfer ratio interval division, and each subinterval defines one and imitates Penetrate function；Then, this affine function is utilized to calculate the dutycycle predictive value on each subintervalFinally, T-S skill is utilized Art is by all subintervalsJoin together, calculate the dutycycle of the overall situation

A T-S predictor substantially single-input single-output process, output function is affine function, makes f_iFor i-th The output function (1≤i≤12) in subinterval, its form is as follows:

f_i(α)=a_iα+b_i (15)

A in formula_i, b_iFor constant, α=y_r/e_abFor input and the satisfied 1≤α≤M of T-S predictor, wherein M is transfer ratio letter NumberMaximum.Transfer ratio interval [1, M] is divided into 12 subintervals: (S₁,S₂,…,S_n)。

Then desired output voltage is y_rTime, T-S predictor fuzzy rule is:

Article i-th, predictor rule: if α is is S_i

So f_i(α)=a_iα+b_i

S in formula_iFor i-th fuzzy set, i=1,2 ... 12, the affine function on each subinterval is exported in carrying out The heart is average, weighting anti fuzzy method, then overall situation fuzzy output is

$D=\underset{i=1}{\overset{12}{\&Sigma;}}{\mathrm{\&mu;}}_i\left(\mathrm{\&alpha;}\right)f_i\left(\mathrm{\&alpha;}\right)=\underset{i=1}{\overset{12}{\&Sigma;}}{\mathrm{\&mu;}}_i\left(\mathrm{\&alpha;}\right)\&lsqb;a_i\mathrm{\&mu;}+b_i\&rsqb;---\left(16\right)$

In formula, α is former piece variable, μ_i(α) it is that α is at fuzzy subset S_iOn membership function, and haveIn the present invention, the degree of membership letter of fuzzy predictor Number is as shown in Figure 7.

Four, fuzzy sliding mode feedback charge controller comprehensive Design

Electric automobile speed and load wide variation in the case of, feedback charging system exists polytype not Definitiveness, thus the control performance of charging system is brought adverse effect.Firstly, since winding back-emf and electric automobile generating Machine rotating speed is proportional, and the change of automobile torque and speed is the severe jamming in feedback charge control system；Secondly, the parameter of electric machine Under different temperature rises, certain uncertainty is all there is with winding electrical variable；Finally, although the load resistance of battery is being cut Can be considered constant in changing interval, but the equivalent load resistance of battery will be along with charging voltage and the change of battery charge (SOC) And change, thus there is bigger load resistance disturbance.In the present embodiment, design the T-S Fuzzy Sliding Model Controller of robust, with Overcome above-mentioned probabilistic impact, it is achieved high performance constant voltage, current feedback charge.

Use sliding moding structure technology, choose the sliding-mode surface of the present embodiment design, utilize LMI approach to ask Solve sliding formwork coefficient and controller gain, it is achieved T-S Design of Fuzzy sliding mode controller.The design of sliding mode controller includes that sliding-mode surface sets Meter, two steps of design of control law.

1. design integral form obscures sliding-mode surface:

$\mathrm{\&sigma;}(t,\hat{x})=S\&CenterDot;\hat{x}\left(t\right)+\mathrm{\&lambda;}\&Integral;S\hat{x}\left(t\right)dt---\left(17\right)$

In formulaFor sliding-mode surface, constant λ > 0 is storage gain, sliding formwork coefficient S ∈ R^m×n, Sliding mode variable structure control is managed Opinion requires that the selection of S is necessary to ensure that the existence of Equivalent Sliding Mode controlled quentity controlled variable, i.e. matrixMust be reversible.For This, theorem 1 gives S and the computational methods of sliding-mode surface and existence proof.The present embodiment chooses the reason in Integral Sliding Mode face, is The stable state charging voltage error of closed-loop control system is the advantage of zero to utilize it can guarantee that.

Theorem 1: for the fuzzy system (14) of transducer status space small-signal model, design is such as the integral form of formula (17) Fuzzy sliding mode face.Making (18) if there is feasible solution (Q, α, β, μ), (19) become with the linear inequality (LMIs) in (20) formula Vertical, then design sliding formwork coefficient is S=(B^TQ^-1B)^-1B^TQ^-1。

$\left[\begin{array}{ccc}{K}^T(A_{i}Q+{\mathrm{QA}}_i^T)K& *& *\\ \mathrm{\&mu;}{\stackrel{\&OverBar;}{B}}^{T}K& -I& *\\ A_{i}QK& \mathrm{\&eta;}\stackrel{\&OverBar;}{B}& -I\end{array}\right]<0,\&ForAll;i---\left(18\right)$

$\left[\begin{array}{cccc}Q& I& *& *\\ I& \mathrm{\&alpha;}I& *& *\\ 0& 0& \mathrm{\&beta;}I-Q& *\\ 0& 0& 0& 2\sqrt{{\mathrm{\&lambda;}}_B}\mathrm{\&mu;}-\sqrt{r}(\mathrm{\&alpha;}+\mathrm{\&beta;})\end{array}\right]>0.---\left(19\right)$

$\mathrm{\&mu;}(1+{\mathrm{\&eta;}}_B)\left|\right|\stackrel{\&OverBar;}{B}\left|\right|-(1-{\mathrm{\&eta;}}_B)<0---\left(20\right)$

Invertible matrix Q ∈ R in formula^n×n, decision variable α, β, μ ∈ R, K are the orthogonal complement matrix of matrix B, λ_BIt it is matrix B^TB Minimal eigenvalue, and meet λ_BI≤B^TB, the transposition of mark " * " representing matrix relevant position element.

Prove: first prove the existence of sliding formwork coefficient S.According to Schur theorem, by linear inequality (18), (19) and (20) can release

$\mathrm{\&alpha;}>0,\mathrm{\&beta;}>0,\mathrm{\&mu;}>0,{\stackrel{\&OverBar;}{B}}^T\stackrel{\&OverBar;}{B}<{\mathrm{\&mu;}}^{-2}I---\left(21\right)$

Order matrix G=B^TQ^-1, S=(GB)^-1GQ,Easily derive τ ∈ R^1×1；By formula (12), haveWithSet up.Then can release following inequality to set up

$\left|\right|\left|\mathrm{\&tau;}\right|{|}^2\&le;rs\&CenterDot;\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{B}\&CenterDot;S^T\&le;{\mathrm{r\&mu;}}^{-2}{\mathrm{SS}}^T---\left(22\right)$

Again depending on Schur theorem, by linear inequality (18), (19) are derived

0<β^-1I < Q < α I, 0 < α^-1I<Q^-1<βI (23)

Then have:

${\mathrm{r\&mu;}}^{-2}{\mathrm{SS}}^T={\mathrm{r\&mu;}}^{-2}{\left(GB\right)}^{-1}{B}^{T}Q{\&lsqb;{\left(GB\right)}^{-1}{B}^{T}Q\&rsqb;}^T<{\mathrm{r\&mu;}}^{-2}\mathrm{\&alpha;}\mathrm{\&beta;}{\left(B^{T}B\right)}^{-1}<{\mathrm{r\&lambda;}}_B^{-1}{\mathrm{\&mu;}}^{-2}\mathrm{\&alpha;}\mathrm{\&beta;}I---\left(24\right)$

By formula (24) and inequality (19), and notice inequality(in formula, a, b are nonnegative number), can obtain

$\left|\right|\mathrm{\&tau;}|{|}^2\&le;\frac{r}{4{\mathrm{\&lambda;}}_B{\mathrm{\&mu;}}^2}{(\mathrm{\&alpha;}+\mathrm{\&beta;})}^{2}I<I---\left(25\right)$

| | τ | | < 1 can be obtained by equation (25), thus demonstrate matrixNonsingular, matrix Reversible.This shows, when choosing sliding formwork coefficient by theorem 1, sliding-mode surface and equivalent control amount all exist, and once system mode enters Sliding-mode surfaceIn, the dynamic of fuzzy system (14) is equivalent to the sliding formwork motion of depression of order, is The state trajectory of system also goes to zero asymptotic.

It is defined as follows linear transformation T, the depression of order shape of sliding formwork state He (n-m) rank system mode x to be decomposed into m rank State:

$T=\left[\begin{array}{c}{T}_1\\ T_2\end{array}\right]=\left[\begin{array}{c}{\left(K^{T}QK\right)}^{-1}{K}^T\\ {\left(GB\right)}^{-1}G\end{array}\right]---\left(26\right)$

T in formula₁∈R^(n-m)×n, T₂∈R^m×n, matrix K meets B^TK=0, K^TK=I, easily calculating the inverse of linear transformation is T^-1=[Q^-1K B].After linear transformation, calculating new state is:

$z=T\hat{x}=\left[\begin{array}{c}{z}_1\\ z_2\end{array}\right]---\left(27\right)$

By formula (27), it is known thatFor sliding formwork state, z₁For reduced order state, and the dynamical equation of state z For:

$\begin{array}{c}\stackrel{\&CenterDot;}{z}=T\&CenterDot;\stackrel{\&CenterDot;}{\hat{x}}\left(t\right)=\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\left[\begin{array}{c}{T}_1\\ T_2\end{array}\right]A_i\&CenterDot;(\left[\begin{array}{cc}{Q}^{-1}K& B\end{array}\right]\&CenterDot;T)\hat{x}\left(t\right)+\left[\begin{array}{c}{T}_1\\ T_2\end{array}\right]\&CenterDot;\&lsqb;B+\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}\&rsqb;\&CenterDot;\&lsqb;\hat{d}+g(t,\hat{x},\hat{d})\&rsqb;\\ =\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\left[\begin{array}{cc}{T}_1{A}_i{Q}^{-1}K& T_1{A}_{i}B\\ T_2{A}_i{Q}^{-1}K& T_2{A}_{i}B\end{array}\right]\&CenterDot;\left[\begin{array}{c}{z}_1\\ z_2\end{array}\right]+\left[\begin{array}{c}{T}_{1}B+T_1\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}\\ T_{2}B+T_2\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}\end{array}\right]\&CenterDot;\&lsqb;\hat{d}+g(t,\hat{x},\hat{d})\&rsqb;\end{array}---\left(28\right)$

T can be calculated by formula (26)₁B=(K^TQ^-1K)^-1K^TB=0 and T₂B=I, according to sliding mode control theory, orderThen calculating equivalent control amount is:

${\hat{d}}_{eq}\left(t\right)=-\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right){\&lsqb;I+S\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}\&rsqb;}^{-1}{\mathrm{SA}}_i\hat{x}\left(t\right)-\mathrm{\&lambda;}S\hat{x}\left(t\right)-g(t,\hat{x},\hat{d})---\left(29\right)$

By the equivalent control amount in formula (29) in people's formula (28), can obtain:

$\left[\begin{array}{c}{\stackrel{\&CenterDot;}{z}}_1\\ {\stackrel{\&CenterDot;}{z}}_2\end{array}\right]=\left[\begin{array}{cc}{\mathrm{\&Xi;}}_1\left(t\right)& {\mathrm{\&Xi;}}_2\left(t\right)\\ 0& -\mathrm{\&lambda;}I\end{array}\right]\left[\begin{array}{c}{z}_1\\ z_2\end{array}\right]---\left(30\right)$

In formula:

${\mathrm{\&Xi;}}_2\left(t\right)=\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right){\left(K^{T}QK\right)}^{-1}{K}^T{\mathrm{\&Theta;A}}_{i}B-\mathrm{\&lambda;}{\left(K^{T}QK\right)}^{-1}{K}^T\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}{\&lsqb;I+S\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}\&rsqb;}^{-1}.$

Carry out sliding formwork motion in analysis mode (30) belowStability.Definition Lyapunov function Wherein matrix P=K^TQK>0.To function V₁Derivation obtains:

${\stackrel{\&CenterDot;}{V}}_1=\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)z_1^T\{K^T\&lsqb;I-\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}{\&lsqb;I+S\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}\&rsqb;}^{-1}S\&rsqb;A_i{Q}^{-1}K+(*)\}{z}_1---\left(31\right)$

Due toSubstituted into formula (31), can be obtained:

${\stackrel{\&CenterDot;}{V}}_1=\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)w^T\{(A_{i}QK+*)-(\stackrel{\&OverBar;}{B}{\&lsqb;I+\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}S\stackrel{\&OverBar;}{B}\&rsqb;}^{-1}\&CenterDot;\stackrel{\&OverBar;}{H}(z)\stackrel{\&OverBar;}{I}S\&CenterDot;A_{i}QK+*)\}w---\left(32\right)$

W=Kz in formula₁∈R^n×1.By inequality (22), (24) andCan obtain:

$\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}S\&CenterDot;S^T{\stackrel{\&OverBar;}{I}}^T{\stackrel{\&OverBar;}{H}}^T\left(z\right)\&le;{\mathrm{\&mu;}}^{2}I---\left(33\right)$

Definition vectorV ∈ R in formula^(n-m)×nFor arbitrarily vector, then q_iCan It is rewritten as

$q_i=-\stackrel{\&OverBar;}{H}\left(z\right)\stackrel{\&OverBar;}{I}S\&CenterDot;\&lsqb;A_{i}QKv+\stackrel{\&OverBar;}{B}{q}_i\&rsqb;---\left(34\right)$

By formula (34), can obtain:

In formulaFor any vector v, orderB=q_i, W=μ²Ω^-1, can obtain:

$2v^T\&lsqb;\stackrel{\&OverBar;}{B}+K^T{\mathrm{QA}}_i^T\stackrel{\&OverBar;}{B}\&rsqb;q_i\&le;{\mathrm{\&mu;}}^{-2}{q}_i^T{\mathrm{\&Omega;q}}_i+{\mathrm{\&mu;}}^2{v}^T\&lsqb;I+K^T{\mathrm{QA}}_i^T\&rsqb;\stackrel{\&OverBar;}{B}\&CenterDot;{\mathrm{\&Omega;}}^{-1}\&CenterDot;{\&lsqb;\stackrel{\&OverBar;}{B}+K^T{\mathrm{QA}}_i^T\stackrel{\&OverBar;}{B}\&rsqb;}^{T}v---\left(35\right)$

By inequality (25), (26), can be calculated

$A_{i}QK+K^T{\mathrm{QA}}_i^T+K^T{\mathrm{QA}}_i^T{A}_{i}QK+{\mathrm{\&mu;}}^2\&lsqb;\stackrel{\&OverBar;}{B}+K^T{\mathrm{QA}}_i^T\stackrel{\&OverBar;}{B}\&rsqb;{\mathrm{\&Omega;}}^{-1}{\&lsqb;\stackrel{\&OverBar;}{B}+K^T{\mathrm{QA}}_i^T\stackrel{\&OverBar;}{B}\&rsqb;}^T<0---\left(36\right)$

Again depending on Schur theorem, inequality (36) is equivalent to linear inequality (37):

$\left[\begin{array}{ccc}{K}^T(A_{i}Q+{\mathrm{QA}}_i^T)K& *& *\\ \mathrm{\&mu;}{\stackrel{\&OverBar;}{B}}^{T}K& -I& *\\ A_{i}QK& \mathrm{\&eta;}\stackrel{\&OverBar;}{B}& -I\end{array}\right]<0,\&ForAll;i---\left(37\right)$

This demonstrate that the integral form that the present embodiment defines obscures sliding-mode surface and the existence of equivalent control amount, and sliding formwork motion It is asymptotically stable.

2. design of control law.

Next step of controller design process is design control law, in the present embodiment, designs following fuzzy sliding mode tracking control rule Then: the

I bar controls rule: ifis

So

In formula, ξ=η_B+τ+η_Bτ, sliding formwork handoff gainMeet constraints:

${\mathrm{\&rho;}}_i(t,\hat{x})=\mathrm{\&xi;}\left|\right|{\mathrm{SA}}_i\hat{x}\left(t\right)\left|\right|+(1+|\left|\mathrm{\&tau;}\right|\left|\right)\&lsqb;{\mathrm{\&eta;}}_A(t,\hat{x})+{\mathrm{\&eta;}}_w(t,\hat{x})\&rsqb;+{\mathrm{\&epsiv;}}_i,\&ForAll;i---\left(40\right)$

Constant ε_i> 0, sgn (σ) is sign function.

The fuzzy sliding mode tracking control rule of the design overall situation is

$\hat{d}\left(t\right)=-\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\&lsqb;{\mathrm{SA}}_i\&CenterDot;\hat{x}\left(t\right)-\mathrm{\&lambda;}S\hat{x}\left(t\right)-\frac{1}{1-\mathrm{\&xi;}}{\mathrm{\&rho;}}_i(t,\hat{x})\&CenterDot;Sgn\left(\mathrm{\&sigma;}\right)\&rsqb;---\left(41\right)$

Based on Lyapunov stability law, theorem 2 demonstrates when using controller (41), and control system feedback is charged Voltage error by asymptotic convergence in zero.

Theorem 2: for the fuzzy system (14) of transducer status space small-signal model, use the integration that formula (17) designs Fuzzy sliding-mode surface, when using controller (41), system trajectory will be driven on sliding-mode surface, and system will be asymptotic problem 's.

Prove: proving based on sliding mode control theory and Lyapunov method, Lyapunov function is designed as: V₂=σ^Tσ >=0, The derivative of Lyapunov function against time is:By inequalityWith controller formula (41) substitute into, can obtain after being computed:

$\begin{array}{c}{\stackrel{\&CenterDot;}{V}}_2={\mathrm{\&sigma;}}^T\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\&lsqb;{\mathrm{SA}}_i\hat{x}\left(t\right)+SB({\tilde{d}}_i+g(t,\hat{x},\hat{d})+\mathrm{\&lambda;}S\hat{x}\left(t\right)\&rsqb;\\ ={\mathrm{\&sigma;}}^T\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\left({\mathrm{SA}}_i\hat{x}\right(t)+SB\&lsqb;-{\mathrm{SA}}_i\hat{x}(t)-\mathrm{\&lambda;}S\hat{x}(t)-\frac{1}{1-\mathrm{\&xi;}}{\mathrm{\&rho;}}_i(t,\hat{x})\&CenterDot;\mathrm{sgn}(\mathrm{\&sigma;})+g(t,\hat{x},\hat{d})\&rsqb;+\mathrm{\&lambda;}S\hat{x}(t\left)\right)\\ \&le;\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right){\mathrm{\&sigma;}}^T\left({\mathrm{SA}}_i\hat{x}\right(t)+\mathrm{\&lambda;}S\hat{x}(t)+{\hat{d}}_i)+\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\left|\right|\mathrm{\&tau;}\left|\right|\&CenterDot;\left(\mathrm{\&tau;}\right|\left|{\hat{d}}_i\right||+(1+\left|\right|\mathrm{\&tau;}\left|\right|\left)\right({\mathrm{\&eta;}}_B\left|\right|{\hat{d}}_i\left|\right|+{\mathrm{\&eta;}}_A+{\mathrm{\&eta;}}_w\left)\right)\&CenterDot;\left|\right|\mathrm{\&sigma;}\left|\right|\\ \&le;-\frac{1}{1-\mathrm{\&xi;}}\left(\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\right(z\left){\mathrm{\&rho;}}_i\right(t,\hat{x})+\mathrm{\&xi;}\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i(z_i\left)\right|\left|{\mathrm{SA}}_i\hat{x}\right(t\left)\right||+\frac{\mathrm{\&xi;}}{1-\mathrm{\&xi;}}\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i(z_i\left){\mathrm{\&rho;}}_i\right(t,\hat{x})+(1+\left|\right|\mathrm{\&tau;}\left|\right|\left)\right({\mathrm{\&eta;}}_A+{\mathrm{\&eta;}}_w\left)\right)\&CenterDot;\left|\right|\mathrm{\&sigma;}\left|\right|\\ =-\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right){\mathrm{\&epsiv;}}_i\left|\right|\mathrm{\&sigma;}\left(t\right)\left|\right|\end{array}---\left(42\right)$

Due to ε_i> 0, formula (42) obtainThis shows when σ (t) ≠ 0, under the effect of controller (41), and system shape The integral form being driven into design is obscured on sliding-mode surface by state, i.e. enters sliding formwork motion, and closed loop system is asymptotically stable, Feedback charging voltage error will level off to zero.

Fig. 1 is the theory structure block diagram of the present embodiment electric automobile fuzzy sliding mode feedback charge control method.The present embodiment The structure of fuzzy sliding mode feedback charge controller, including storage battery charge controller 1, the outfan of storage battery charge controller 1 Being sequentially connected with driving/isolation circuit 4, regenerative braking charging system 3, sensing acquisition circuit 2, sensing acquisition circuit 2 finally connects Return storage battery charge controller 1.

Accumulator, as power source, on the one hand provides for electric automobile during traveling and drives energy, on the other hand store electronic vapour Feedback energy during car regenerative braking；Storage battery charge controller and accumulator, two-way DC/DC changer, sensing circuit and electricity Cable bus electric connection is used between machine；By controlling the power switch pipe of two-way DC/DC changer, it is achieved accumulator charged electrical Pressure, the tracing control of electric current, make feedback energy keep relative constancy and smooth, reclaim the braking energy of electric automobile；Accumulator Charge controller, using accumulator charging voltage and electric current as controlling target, has higher charging security energy and electrical energy Efficiency of transmission, burst feedback energy during well-tolerated electric vehicle brake, thus extend the distance travelled of electric automobile.

Storage battery charge controller 1 is constituted by dsp processor control circuit plate is integrated, and dsp processor passes through sensing acquisition Circuit 2 gathers voltage and current signal and pedal signal；Dsp processor is modulated from PWM1～PWM6 pin output pulse width (PWM) signal, controls the switch motion of six power tubes of two-way DC converter (DC/DC) through circuit 4 of overdriving/isolate； DSP external interrupt pin 1 (XINT1) response brake pedal step on action, 2/3 stroke before pedal, it is desirable to charged electrical Pressure, electric current are linear with pedal-displacement；At rear 1/3 stroke of pedal, do not overcharged for protection accumulator, it is desirable to charging Voltage, electric current no longer increase with pedal-displacement and increase, and are set to maximum charging voltage, electric current, and keep constant.

Fuzzy predictor, space State Average Model, overall situation mould is also included in the functional structure of storage battery charge controller 1 The modules such as fuzzy model and fuzzy sliding mode charge controller, are completed by dsp software algorithm；DSP obtains pedal via interrupt mode The displacement of braking and desired charging voltage, electric current, utilize fuzzy predictor to estimate the steady-state value of PWM ripple dutycycleThen, on the one hand combining accumulator voltage calculates state-space model, obtains state steady-state valueOn the other hand combine State variable measured value x_mWith tracking error x_e, calculate Global fuzzy model, obtain state variable small signal valueFinally, by The control method of the present embodiment calculates dutycycle instantaneous valueWith steady-state valueIt is added, defeated from PWM1～the PWM6 pin of DSP Go out modulation waveform, control six pipe switch motions of two-way DC/DC changer through overdriving and isolating circuit 4, carry out constant voltage, constant current Charging.

The circuit structure of the regenerative braking charging system that the present embodiment relates to is shown in Fig. 2, including an accumulator, four electric currents Sensor SA1～SA4, two voltage sensor SV1～SV2, storage capacitor C, two-way DC/DC changer and Y-connection straight Stream motor；The upper end series-connected current sensors SA1 posterior end parallel connection of accumulator is connected to voltage sensor SV1；Storage capacitor C with Voltage sensor SV2 is in parallel, more in parallel with voltage sensor SV1 with releasing after energy resistance R connects；Two-way DC/DC changer is by six Individual power tube T1～T6 forms, and upper brachium pontis is T1, T3 and T5, and lower brachium pontis is T4, T2 and T6, and every bridging connects midpoint series electrical respectively After flow sensor SA2～SA4, the phase winding with motor is connected.

When electric automobile works in regenerative braking operating mode, and accumulator is charged, and two-way DC/DC changer works in reversely boosting State, now goes up brachium pontis power tube S1, S3 and S5 chopping modulation, and flowing to accumulator after feeding braking energy back is boosted is that it fills Electricity；Each bridge of two-way DC/DC changer is staggered work within a cycle, 120 degree of phase places of each conducting, i.e. S1 Yu S4 be operated in 0～ 120 degree, S3 Yu S2 is operated in 120～240 degree, S5 Yu S6 is operated in 240～360 degree；High performance perseverance is realized during for making charging Pressure charging, takes Fuzzy Sliding Model Controller, and six switching tubes controlling two-way DC/DC changer carry out independent pulse width tune System.

Fig. 8 is the circuit diagram driving/isolate circuit 4, drive/isolation circuit 4 is by photoelectric isolating device U1 and drives control Circuit is constituted, and photoelectric isolating device U1 uses photoelectric coupling chip 4N25, drives control circuit U2 to use chip I R2122S, DSP The modulated signal of processor pin PWM1～PWM6 output controls the two-way each merit of DC/DC changer after driving/isolate circuit The switch motion of rate pipe.

Fig. 9 is sensing acquisition circuit diagram, by voltage, current sensor and two channel operation amplifier U4, U5 groups on circuit Becoming, voltage, current sensor use HAS200-P to realize, and operational amplifier uses LF353 to realize；Operational amplifier U5B gathers Hall voltage signal, current signal, and send in dsp processor via external interrupt ADCINT0～ADCINT5 pin.

The electric automobile fuzzy sliding mode feedback charge controller that the present embodiment relates to realizes accumulator constant voltage, constant-current charge The method controlled as shown in Figure 10, flow process comprises the following steps:

(1) connect storage battery charge controller power supply, automatic initial survey, and enter state to be charged；

(2) initialize storage battery charge controller, input parameter matrix A_i、B_i、E_iAnd D_i, and calculate matrix

(3) by DSP with interrupt mode response brake pedal signal, it is judged that accumulator is the need of charging；If desired charge, Then calculate desired charging voltage, current value y according to brake pedal displacement meter_r；Otherwise, (2nd) step is turned；

(4) being judged whether the feasible solution of LMI (18)～(20) exists by theorem 1, entering next step if existing；Otherwise, turn (2nd) step；

(5) steady-state value of PWM ripple dutycycle is estimated by fuzzy predictor

(6) byCalculate state-space model with accumulator voltage, obtain state steady-state value

(7) bonding state variable measurements x_mWith tracking error x_e, calculate Global fuzzy model, obtain state variable little Signal value

(8) dutycycle instantaneous value is calculated by the control method of the present inventionWith steady-state valueIt is added, from the PWM mouth of DSP Output modulation waveform, controls six pipe switch motions of two-way DC/DC changer through overdriving and isolating circuit, carries out constant voltage, perseverance Current charge.

Figure 11～Figure 13 is the step response curve figure that the charging of the present embodiment method controls effect.Wherein accumulator internal resistance R_b Doubled by nominal value at 0.015 second, within 0.025 second, return to nominal value；Back-emf value e that speed is corresponding_abRaised at 0.03 second+ 20%, reduced by 20% at 0.035 second；The expected value of charging voltage and y_rFor under step signal, Figure 11, Figure 12, Figure 13 give respectively Charging voltage aircraft pursuit course, winding current curve and the dutycycle curve of controller output are gone out；As seen from the figure, the rank of this method The response time that jumps is less than 0.015 second, and steady track error is less than 0.76%, and can well overcome accumulator parameter perturbation to bring Hydraulic performance decline.

Figure 14～Figure 16 is the sinusoidal response curve chart that the charging of the present embodiment method controls effect.Now, at consecutive variations Desired output voltage under, the steady operation point of DC/DC changer is also in respective change.Wherein accumulator internal resistance was at 0.015 second Doubled by nominal value, within 0.025 second, return to nominal value；Back-emf value corresponding to speed raised+20% at 0.03 second, Within 0.035 second, reduce by 20%；The expected value of charging voltage is that under step signal, Figure 14, Figure 15, Figure 16 sets forth charged electrical Pressure aircraft pursuit course, winding current curve and the dutycycle curve of controller output；As seen from the figure, this method can well overcome car Speed change, desired signal convert the hydraulic performance decline that brings, and are respectively provided with at different steady operation points that steady track error is little, Shandong The feature that rod is strong.

Through emulation and actual tests, the charging performance of the present embodiment method and controller meets demand, and charging effect makes People is satisfied, can well reclaim the braking energy of electric automobile, thus effectively extend battery-operated life-span and garage Sail mileage number.

Claims (5)

1. an electric automobile fuzzy sliding mode feedback charge controller, it is characterised in that: it includes storage battery charge controller (1), the outfan of storage battery charge controller (1) be sequentially connected with driving/isolation circuit (4), regenerative braking charging system (3), Sensing acquisition circuit (2), sensing acquisition circuit (2) finally connects back to storage battery charge controller (1).

Electric automobile fuzzy sliding mode feedback charge controller the most according to claim 1, it is characterised in that: described feedback system Dynamic charging system (3) includes an accumulator, four current sensor SA1～SA4, two voltage sensor SV1, SV2, energy storage Electric capacity C, two-way DC/DC changer and the dc motor of Y-connection；Under after the upper end series-connected current sensors SA1 of accumulator End parallel connection is connected to voltage sensor SV1；Storage capacitor C is in parallel with voltage sensor SV2, then with release can resistance R connect after with electricity Pressure sensor SV1 is in parallel；Two-way DC/DC changer is made up of six power tube T1～T6, and upper brachium pontis is T1, T3 and T5, lower bridge Arm is T4, T2 and T6, and every bridging connects the phase winding after midpoint respectively series-connected current sensors SA2, SA3, SA4 with dc motor Connect.

Electric automobile fuzzy sliding mode feedback charge controller the most according to claim 1, it is characterised in that: described driving/ Isolation circuit (4) includes photoelectric isolating device U1 and drives control circuit U2, and photoelectric isolating device U1 uses photoelectric coupling chip 4N25, drives control circuit U2 to use chip I R2122S.

Electric automobile Fuzzy Sliding Model Controller the most according to claim 1, it is characterised in that: described sensing acquisition circuit (2) including voltage, current sensor and two channel operation amplifier U4, U5, voltage, current sensor use HAS200-P, fortune Calculate amplifier and use LF353.

5. based on a feedback charge control method for electric automobile fuzzy sliding mode feedback charge controller described in claim 1, It is characterized in that: system model stage when it includes setting up DC/DC changer feedback charging, the DC/ set up under feedback is charged DC changer T-S fuzzy model stage, the calculating predictive value stage of changer dutycycle, fuzzy sliding mode charge controller comprehensively set The meter stage；

(1) step setting up system model stage when DC/DC changer feedback is charged described in is as follows:

(1) system model of charging circuit is calculated；

Same set of DC/DC translation circuit under feedback charge condition, in the case of use and driven；Use six pipe full-bridge pulses Width modulated, it is not necessary to extra circuits unit；As a example by A phase and B phase, analyze mathematical model, open at switch, turn off two kinds In the case of, research controls the mathematical model of system when target is accumulator charging voltage, electric current, to be output as battery charging voltage Y=v_oIllustrate as a example by (t)；Writ state variable x=[i_L v_c]^T, the system model of circuit is given by following equation:

$\left\{\begin{array}{c}\stackrel{\&CenterDot;}{x}=A_{ON}x+B_{ON}{e}_{ab}+g_{ON}\\ \stackrel{\&CenterDot;}{x}=A_{OFF}x+B_{OFF}{e}_{ab}+g_{OFF}\end{array}\right.,\left\{\begin{array}{c}y=C_{ON}x+f_{ON}\\ y=C_{OFF}x+f_{OFF}\end{array}\right.;$

Formula breaker in middle opens, turn off under matrix be respectively C_ON=[0 R_bR^-1], C_OFF=[R_bR_cR^-1 R_bR^-1], f_ON=f_OFF=R_cR^-1v_b；Coefficient a₁=-(2R_m+R_s+R_d)/2L_m, a₂=-1/CR, a₃=-(R_m+R_d)/L_m-R_cR_b/(2L_mR), a₄=R_b/(2L_mR), g₁=-a₂v_b, g₂=-R_c/(2L_mR)v_b；Symbol Lm is winding inductance, R_mIt is winding resistance, R_sAnd R_dIt is on and off switch and fly-wheel diode respectively Conducting resistance, C and R_cIt is battery DC side capacitors electric capacity and dead resistance, R respectively_bIt is battery equivalent internal resistance, R=R_c+ R_b, v_cIt is the voltage drop on capacitor, v_bRepresent cell emf, i_bFor flowing through the feedback charging current of battery, e_abFor biphase around Group counter electromotive force, v_oFor output voltage；

Output equation in circuit model is charging voltage equation, and when being output as charging current, its output equation is rewritten as i_o =-R^-1v_c+R^-1v_b；

(2) space State Average Model is calculated；

PWM duty cycle d (t) and d'(t it is multiplied by respectively on circuit model both sides)=1-d (t), and carry out handling averagely, try to achieve Space State Average Model is:

$\left\{\begin{array}{c}{\stackrel{\&CenterDot;}{x}}_m=\stackrel{\&OverBar;}{A}{x}_m+\stackrel{\&OverBar;}{B}{e}_{ab}+\stackrel{\&OverBar;}{g}\\ y_m={\stackrel{\&OverBar;}{C}}_m{x}_m+\stackrel{\&OverBar;}{f}\end{array}\right.;$

Matrix in formula x_m、y_mIt is the state variable meansigma methods in single PWM cycle and output voltage average value respectively；Right In given dutycycleOrderCan try to achieveThe state variable steady-state value of place's quiescent point is:

$\left\{\begin{array}{c}{x}_m=\stackrel{\&OverBar;}{x}=-{\stackrel{\&OverBar;}{A}}^{-1}(\stackrel{\&OverBar;}{B}{e}_{ab}+\stackrel{\&OverBar;}{g})\\ y_m=\stackrel{\&OverBar;}{y}=-{\stackrel{\&OverBar;}{CA}}^{-1}(\stackrel{\&OverBar;}{B}{e}_{ab}+\stackrel{\&OverBar;}{g})\end{array}\right.;$

In formulaWithRepresent x respectively_mAnd y_mSteady-state value；Due in operating pointThere is the interference of small-signal in place, then becomes The instantaneous value of amount can be written as:Wherein d (t), x (t), y (t) For variable instantaneous value,For small-signal disturbance；

(3) state space small-signal and integration control model are calculated；

Utilize small-signal perturbation analysis method to isolate steady-state variable and transient state variable, ignore the secondary of disturbance quantity and above high-order , trying to achieve state space small-signal model is:

$\left\{\begin{array}{c}{\stackrel{\&CenterDot;}{\hat{x}}}_m\left(t\right)=\stackrel{\&OverBar;}{A}{\hat{x}}_m\left(t\right)+\stackrel{\&OverBar;}{E}\hat{d}\left(t\right)\\ {\hat{y}}_m\left(t\right)=\stackrel{\&OverBar;}{C}{\hat{x}}_m\left(t\right)\end{array}\right.;$

Matrix in formula

For State space averaging and two kinds of models of state space small-signal, all rely on dutycycle steady-state value at operating point For realizing zero steady track error of output voltage, introduce integrating state variable: x_e=∫ e dt=∫ (y_r-y_m) dt, in formula Tracking error e=y_r-y_m, y_rFor expectation output voltage；State space small-signal model is rewritten as following integration control model:

$\left\{\begin{array}{c}{\stackrel{\&CenterDot;}{\hat{x}}}_a\left(t\right)={\stackrel{\&OverBar;}{A}}_a{\hat{x}}_a\left(t\right)+{\stackrel{\&OverBar;}{E}}_a\hat{d}\left(t\right)\\ {\hat{y}}_m\left(t\right)={\stackrel{\&OverBar;}{C}}_a{\hat{x}}_a\left(t\right)\end{array}\right.$

In formulaFor augmented state variable, control input for dutycycle transient valueMatrix

${\stackrel{\&OverBar;}{A}}_a=\left[\begin{array}{cc}{0}_{1\&times;2}& -1\\ 0_{2\&times;1}& \stackrel{\&OverBar;}{A}\end{array}\right],{\stackrel{\&OverBar;}{E}}_a=\left[\begin{array}{c}0\\ \stackrel{\&OverBar;}{E}\end{array}\right],{\stackrel{\&OverBar;}{C}}_a=\left[\begin{array}{c}0\\ \stackrel{\&OverBar;}{C}\end{array}\right];$

In vehicle travel process, along with the change of operating point, the dutycycle of changerChange therewith, thus cause state Zero pole point and the amplitude-frequency response of space small-signal model transmission function change, thus state space small-signal model is duty The nonlinear function of ratio, charge control system is a Nonlinear Uncertain Systems；

(2) step of the described DC-DC converter T-S fuzzy model set up under feedback charging is as follows:

(1) T-S fuzzy model；Utilize T-S fuzzy technology to approach nonlinear system, for i-th operating point, use following IF- THEN rule describes nonlinear state space small-signal model:

Article i-th, model rule: if z₁T () isAnd z₂T () is..., and z_nT () isSo:

In formulaFor fuzzy set, z (t)=[z₁,…,z_n] it is former piece variable,It is state error,It is to control input, A_iAnd B_iFor treating set matrix, rule number i=1,2 ..., r；Fuzzy weighting valueF_i ^j[z_j(t)] >=0 it is z_jT () is right under i-th fuzzy rule The degree of membership answered, and haveBased on single-point obfuscation, product inference and weighted average anti fuzzy method, overall situation mould Fuzzy model is:

(2), after the various interference being subject to during consideration charging and uncertainty, Parameter uncertainties fuzzy model is:

$\stackrel{\&CenterDot;}{\hat{x}}\left(t\right)=\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\&lsqb;A_i+{\mathrm{\&Delta;A}}_i\left(t\right)\&rsqb;\hat{x}\left(t\right)+\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\&lsqb;B_i+{\mathrm{\&Delta;B}}_i\left(t\right)\&rsqb;\hat{d}\left(t\right)+\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)w_i(t,\hat{x});$

△ A in formula_iWith △ B_iFor the matching uncertainties of parameter,Represent input and load disturbance；Assume: (i) exists Definitiveness function Μ_A(t), Μ_B(t) and Μ_wT () makes △ A_i=B_iΜ_A(t), △ B_i=B_iΜ_B(t) andAll set up；(ii) system control matrix is unsatisfactory for B₁=B₂=...=B_r；Rewrite Parameter uncertainties mould Type is as follows:

$\stackrel{\&CenterDot;}{\hat{x}}\left(t\right)=\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)A_i\hat{x}\left(t\right)+\&lsqb;B+\stackrel{\&OverBar;}{B}\stackrel{\&OverBar;}{H}\left(h\right(z\left)\right)\stackrel{\&OverBar;}{I}\&rsqb;\&CenterDot;\&lsqb;\hat{d}+g(t,\hat{x},\hat{d})\&rsqb;;$

In formula: External disturbanceThere will necessarily be one Know normal number η_BMake 0≤| | Μ_B||≤η_B< 1, and positive function η continuously_A、η_wMake Then can calculateThe upper bound of norm is

(3) determination of T-S model；[0,1] between the global work area of vehicle driving-cycle is divided into 7 sub spaces, is respectively as follows: [0,0.2], [0.2,0.3], [0.3,0.4], [0.4,0.5], [0.5,0.6], [0.6,0.7], [0.7,1]；Empty at every height Choose a steady operation point between, determine 7 stable operating points as the following formula:

D in formula_iFor the upper bound, subspace, d_iFor subspace lower bound；Utilize T-S modeling method obtains 7 linear submodels at above-mentioned steady operation point, is described as following T-S fuzzy rule:

Article i-th, object-rule: ifSo

F in formulaⁱ(i=1～7) is fuzzy set,

(3) step in the predictive value stage of described calculating changer dutycycle is as follows:

Output voltage-input voltage transfer ratioThe nonlinear function of dutycycle, due to have strong nonlinear characteristic and Parameter uncertainty, utilizes T-S fuzzy close method to predict dutycycle；Prediction process is as follows: first, is drawn in transfer ratio interval It is divided into 12 subinterval (S₁,S₂,…,S_n), each subinterval S_iDefine an affine function；Then, this affine function meter is utilized Calculate the dutycycle predictive value on each subintervalFinally, T-S technology is utilized to incite somebody to actionJoin together, calculate the overall situation Dutycycle

T-S predictor is the affine function of a single-input single-output, inputs as α=y_r/e_abAnd meet 1≤α≤M, wherein M is for passing Defeated compare functionMaximum；Make f_iFor the output (1≤i≤12) in i-th subinterval, its form is: f_i(α)=a_iα+ b_i, wherein a_i, b_iFor constant；Then desired output voltage is y_rTime, T-S predictor fuzzy rule is:

Article i-th, predictor rule: if α is is S_iSo f_i(α)=a_iα+b_i；

S in formula_iFor i-th fuzzy set, the affine function on each subinterval is exported f_iCarry out that center is average, weighting Anti-fuzzy Change, then overall situation fuzzy output isμ in formula_i(α) it is that α is at fuzzy subset S_i On membership function, and have

(4) step in described fuzzy sliding mode charge controller comprehensive Design stage is as follows:

(1) based on sliding mode control theory and Lyapunov method design Integral Sliding Mode diverter surface:

$\mathrm{\&sigma;}(t,\hat{x})=S\&CenterDot;\hat{x}\left(t\right)+\mathrm{\&lambda;}\&Integral;S\hat{x}\left(t\right)dt---\left(17\right)$

Constant λ in formula > 0 is storage gain, sliding formwork coefficient S ∈ R^m×n, sliding mode control theory requires that the selection of coefficient S is necessary to ensure that The existence of Equivalent Sliding Mode controlled quentity controlled variable, i.e. matrixMust be reversible；To this end, give based on Lyapunov method Go out the computational methods of coefficient S and sliding-mode surface；

Based on Lyapunov method, if following linear inequality

$\left[\begin{array}{ccc}{K}^T(A_{i}Q+{\mathrm{QA}}_i^T)& *& *\\ \mathrm{\&mu;}{\stackrel{\&OverBar;}{B}}^{T}K& -I& *\\ A_{i}QK& \mathrm{\&eta;}\stackrel{\&OverBar;}{B}& -I\end{array}\right]<0,\&ForAll;i$

$\left[\begin{array}{cccc}Q& I& *& *\\ I& \mathrm{\&alpha;}I& *& *\\ 0& 0& \mathrm{\&beta;}I-Q& *\\ 0& 0& 0& 2\sqrt{{\mathrm{\&lambda;}}_B}\mathrm{\&mu;}-\sqrt{r}(\mathrm{\&alpha;}+\mathrm{\&beta;})\end{array}\right]>0;$

$\mathrm{\&mu;}(1+{\mathrm{\&eta;}}_B)\left|\right|\stackrel{\&OverBar;}{B}\left|\right|-(1-{\mathrm{\&eta;}}_B)<0;$

There is feasible solution (Q, α, β, μ), then design sliding formwork coefficient is S=(B^TQ^-1B)^-1B^TQ^-1, wherein invertible matrix Q ∈ R^n×n, certainly Plan variable α, β, μ ∈ R, K are the orthogonal complement matrix of matrix B, λ_BIt it is matrix B^TThe minimal eigenvalue of B, and meet λ_BI≤B^TB, note The transposition of number " * " representing matrix relevant position element；Choosing the advantage in Integral Sliding Mode face is can guarantee that closed-loop control system steady State charging voltage error is zero；

(2) based on sliding mode control theory and Lyapunov method design control law, Lyapunov function is designed as:

V₂=σ^Tσ >=0, the derivative of Lyapunov function against time is:

In order to ensure Design following fuzzy sliding mode tracking control regular:

Article i-th, control rule: ifSo

${\hat{d}}_i\left(t\right)=-{\mathrm{SA}}_i\&CenterDot;\hat{x}\left(t\right)-\mathrm{\&lambda;}S\hat{x}\left(t\right)-\frac{1}{1-\mathrm{\&xi;}}{\mathrm{\&rho;}}_i(t,\hat{x})\&CenterDot;\mathrm{sgn}\left(\mathrm{\&sigma;}\right)$

In formula, ξ=η_B+τ+η_Bτ, sliding formwork handoff gain Constant ε_i> 0, sgn (σ) is sign function；The fuzzy sliding mode charge controller of the design overall situation is:

$\hat{d}\left(t\right)=-\underset{i=1}{\overset{r}{\&Sigma;}}{h}_i\left(z\right)\&lsqb;{\mathrm{SA}}_i\&CenterDot;\hat{x}\left(t\right)-\mathrm{\&lambda;}S\hat{x}\left(t\right)-\frac{1}{1-\mathrm{\&xi;}}{\mathrm{\&rho;}}_i(t,\hat{x})\&CenterDot;\mathrm{sgn}\left(\mathrm{\&sigma;}\right)\&rsqb;;$

Based on Lyapunov stability law, nowDemonstrate when using the fuzzy of the overall situation During sliding formwork charge controller, control system feedback charging voltage error by asymptotic convergence in zero.