CN103269176B - Inverter control method based on fractional order PI forecasting function - Google Patents

Abstract

The invention discloses an inverter control method based on a fractional order PI forecasting function and belongs to the technical field of inverter control. According to the control method, an optimized control value is obtained by optimization computing, the control value is used as an input signal of an inverting driving circuit, the purpose that inverter output currents are controlled so that output voltages of the inverter can be affected is achieved, steady-state characteristics are good, response time is short, loading adaptability is strong, anti-jamming capability is strong, and the inverter control method is suitable for a grid-connected inverter in a wind generating system.

Description

A kind of inverter control method based on fractional order PI anticipation function

Technical field

The invention discloses a kind of inverter control method based on fractional order PI anticipation function, belong to the technical field of inverter control.

Background technology

In the power-supply system of reality, sometimes need converting direct-current power into alternating-current power for load, this process direct current being become alternating current, is called inversion.In existing a variety of power supply, if storage battery, solar cell etc. are all DC power supply, when these power supplys of needs are powered to AC load, just need inversion.It is generally acknowledged, the development of inversion transformation technique can be divided into following three phases.

1956-1980 is traditional developing stage, and the feature in this stage is that switching device is based on low speed devices, the switching frequency of inverter is lower, and output voltage waveforms improves based on Multiple Superposition method, and volume weight is larger, inversion efficiency is lower, and sine wave inverter technology starts to occur.

1981-2000 is the high frequency new technology stage, and the feature in this stage is that switching device is based on high speed device, the switching frequency of inverter is higher, and waveform improves based on PWM, and volume weight is little, inversion efficiency is high, and the development of sine wave inverter technology is gradually improved.

Within 2000, be the high-efficiency low-pollution stage so far, the feature in this stage is the combination property based on inverter, low speed and high-speed switching devices use, and Multiple Superposition method and PWM method are also used, no longer deflection pursues high-speed switching devices and high switching frequency, and the inversion transformation technique of high-efficiency environment friendly starts to occur.

The research of High Performance PWM inverter in recent years more and more receives publicity, and occurs and has developed diversified adverser control technology.The control technology of inverter can be divided into two large classes on the whole: based on the control in cycle, based on instantaneous control.Control based on the cycle is by processing the output waveform in last cycle or multiple cycle, utilizes the control method that the result obtained corrects current control.Inherently see, the control based on the cycle is by compensating the periodicity of error, realize the floating effect of stable state, but dynamic responding speed is slow.Control based on the cycle can be divided into effective value FEEDBACK CONTROL, Repetitive controller etc.In order to improve the dynamic responding speed of inverter output voltage waveform, there is the control method of instantaneous values feedback.Be carry out effective controlling in real time to output waveform according to error current based on Instantaneous Control, Hysteresis control, PI control, double-loop control, track with zero error, SVPWM control, Repetitive controller, fuzzy control etc. can be divided into.Detailed content document Zhao that sees reference is, Yu Shijie, Shen Yuliang, Deng. the control method [J] of grid-connected photovoltaic system. electrotechnics, 2002, (3): 12-13.Liviu Mihalache.DSP Control Method of SinglePhase Inverters for UPS Applications [C] .Proc.IEEE APEC, 2002,590-596.

PREDICTIVE CONTROL is the novel computer control algorithm of a class that development in recent years is got up.It is applicable to not easily set up accurate digital model and the industrial processes of more complicated, so it is once occurring the attention being subject to domestic and international project circle, and is successfully applied in the control system of the industrial departments such as oil, chemical industry, metallurgy, machinery.Electric power system is difficult to set up accurate Mathematical Modeling equally, but is again a rapid system simultaneously, and traditional PREDICTIVE CONTROL on-line calculation is large, and poor real, is not suitable for the excitation con-trol of electric power system.Under this background, PFC (Prediction FunctionControl, Predictive function control) adapt to the needs of Fast Process, general principle based on PREDICTIVE CONTROL develops, its detailed content can see document [Wang Shuqing, Jin Xiaoming. Advanced Control Techniques application example [M]. Beijing, Chemical Industry Press, 2005.].Anticipation function is substantially identical with the general principle of forecast Control Algorithm: model prediction, rolling optimization, feedback compensation.The maximum difference of itself and PREDICTIVE CONTROL is the version focusing on controlled quentity controlled variable, thinks that controlled quentity controlled variable is the linear combination of one group of previously selected basic function.Abroad, PFC follows the tracks of at the quick high accuracy of industrial robot, obtains successful application in the rapid system such as the target following of military field.But not yet find the document, the report that the method that fractional order PI (ProportionalIntegral, proportional integral) and Predictive function control combine are applied to synchronous generator excited system control at present.

Summary of the invention

Technical problem to be solved by this invention is the deficiency for above-mentioned background technology, provides a kind of inverter control method based on fractional order PI anticipation function.

The present invention adopts following technical scheme for achieving the above object:

Based on an inverter control method for fractional order PI anticipation function, comprising: set up inverter control system model, determine inverter output current reference locus, comprise the steps:

Step 1, the parameter of the following inverter control system model of initialization: the equivalent series resistance R of filter inductance L, filter capacitor C, filter inductance L _l; And the second-order system that inverter dual input list exports is converted into state space equation, draw coefficient matrices A, B, C, $A=\left[\begin{array}{cc}0& \frac{1}{C}\\ -\frac{1}{L}& -\frac{r}{L}\end{array}\right],B=\left[\begin{array}{cc}0& -\frac{1}{C}\\ \frac{1}{L}& 0\end{array}\right],$ C=[1 0], r are the comprehensive equivalent damping resistance of the various damping factor of inverter;

Step 2, calculates controlled quentity controlled variable u (n) according to following formula:

u(n)＝(R _p+R _i) ^Tf _n(0)，

Wherein:

$R_p+R_i={\left\{\right[K_i{T}_s^{\mathrm{\λ}}+\underset{l=1}{\overset{k}{\mathrm{\Σ}}}(-\frac{1+\mathrm{\λ}}{l}{K}_i{T}_s^{\mathrm{\λ}}{q}_{l-1})]g{\mathrm{Qg}}^T+\mathrm{fR}{f}^T\}}^{-1}\·[K_p(1-q^{-1})+K_i{T}_s^{\mathrm{\λ}}+\underset{l=1}{\overset{k}{\mathrm{\Σ}}}(-\frac{1+\mathrm{\λ}}{l}{K}_i{T}_s^{\mathrm{\λ}}{q}_{l-1})q^{-l}]\mathrm{gQ}{g}^T,$

q ₀＝1， $q_l=(1-\frac{1+\mathrm{\λ}}{l})q_{l-1},$

f＝[f _n1(0),f _nj(0),…,f _nJ(0)] ^T，

$g=g_n={[g_{n1}\left(h_i\right),g_{n2}\left(h_i\right),...,g_{\mathrm{nj}}\left(h_i\right),...,g_{\mathrm{nJ}}\left(h_i\right)]}^T=\left(\begin{array}{cccc}{g}_{n1}\left(h_1\right)& g_{n1}\left(h_2\right)& ...& g_{n1}\left(h_{n_s}\right)\\ g_{n2}\left(h_1\right)& g_{n2}\left(h_2\right)& ...& g_{n2}\left(h_{n_s}\right)\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ g_{\mathrm{nj}}\left(h_1\right)& g_{\mathrm{nj}}\left(h_2\right)& ...& g_{\mathrm{nj}}\left(h_{n_s}\right)\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ g_{\mathrm{nJ}}\left(h_1\right)& g_{\mathrm{nJ}}\left(h_2\right)& ...& g_{\mathrm{nJ}}\left(h_{n_s}\right)\end{array}\right),$

$d=d\left(n\right)=\left(\begin{array}{c}d(n+h_1)\\ d(n+h_2)\\ .\\ .\\ .\\ d(n+h_{n_s})\end{array}\right),$ Wherein, $d(n+h_i)=(1-{\mathrm{\α}}^{h_i})[X\left(n\right)-y_P\left(n\right)]-C(A^i-I)x_m\left(n\right),$

g _nj(i)＝CA ^i-1Bf _nj(0)+CA ^i-2Bf _nj(1)+…+CBf _nj(i-1),

$\mathrm{\α}=\exp \left[\frac{-3T_o}{T_r}\right],$

The controlled quentity controlled variable that u (n) is inverter control system n-th moment exports, R _pfor proportional resistance, R _ifor integration item resistance, f _n(0) for basic function is at the initial value in the n-th moment, 1,2 ..., l ..., k is sampled signal sequence, and k is positive integer, q _l-1for binomial coefficient, Q and R represents error weighting matrix respectively and controls weighting matrix, q ^-1and q ^-lfor time delay operator, J represents the exponent number of basic function, and j is the positive number between 0 to J, K _p, K _ibe respectively broad sense proportional coefficient, integral item coefficient, T _sfor time step, λ is integral parameter, f _nl(0) be the initial value of l rank basic function in the n-th moment, f is the matrix be made up of at the initial value in the n-th moment 1 to J rank basic function, n _sfor optimizing the number of time domain match point, i is 1 to n _sbetween integer, h _ibe the numerical value in i-th match point, g _nji (), for i-th match point on jth time basic function is at the process response function in the n-th moment, g is the matrix be made up of in the value of the n-th etching process response function J basic function, the standard value that X (n) is inverter output current, y _pn () is the output current value of the n-th moment inverter, T _ofor the sampling time, T _rfor the Expected Response time of inverter output current reference locus, α is the speed degree that reference locus is tending towards inverter output current standard value, x _mn () is the state value of the n-th moment inverter control system model;

Step 3, the output current of controlled quentity controlled variable u (n) control inverter obtained according to step 2 thus affect its output voltage.

Described a kind of inverter control method based on fractional order PI anticipation function, basic function described in step 2 is unit step function, and the value of exponent number J is 1.

The present invention adopts technique scheme, has following beneficial effect: steady-state characteristic is good, the response time is shorter, and workload-adaptability is strong, and antijamming capability is strong.

Accompanying drawing explanation

Fig. 1 is the principle schematic of the single-phase inverter control method based on fractional order PI anticipation function.

Fig. 2 is the digital control experiment porch hardware designs figure of single-phase inverter.

When Fig. 3 is fractional order PI Predictive function control inverter, the oscillogram of output voltage and output current.

When Fig. 4 is fractional order PI Predictive function control inverter, the wave form varies comparison diagram of shock load instantaneous output current in output circuit.

When Fig. 5 is PI control inverter, the wave form varies comparison diagram of shock load instantaneous output current in output circuit.

Embodiment

Be described in detail below in conjunction with the technical scheme of accompanying drawing to invention.

The object of the invention is to the method that fractional order PI and Predictive function control combine to be incorporated in single-phase inverter control system and carry out alternative traditional PI and control, a kind of new control strategy is provided.

1, basic function and reference locus is chosen

Predictive function control regards the key of influential system performance as control inputs structure.And in the situation that input signal spectrum is limited in Predictive function control, control inputs only belongs to one group of specific Ball curve relevant with reference locus and object property, the importance chosen of basic function is well imagined.Especially, for linearly, the output of system will be the weighted array that above-mentioned basic function acts on object model response.Control inputs is represented as a series of known basic function { f _nlinear combination, namely

$u(n+i)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}{\mathrm{\μ}}_j\left(n\right)f_{\mathrm{nj}}\left(i\right)---\left(1\right)$

In formula (1): u (n+i) is the controlled quentity controlled variable in the n+i moment;

μ _jn () is basic function weight coefficient;

F _nji () is the value of basic function when iT;

J is the exponent number of basic function.

Basic function choose the character depending on object and desired trajectory, such as can get step, slope, exponential function etc.For optionally fixed basic function f _nji (), can export response g by the off-line object calculated under its effect _nji (), namely weighted array obtain system and export.

The same with Model Algorithmic contral, in PFC (anticipation function), gently set point is reached gradually in order to enable the output of system, avoid occurring overshoot, according to prediction output valve and the output of process value, we can specify a progressive curve trending towards following set point, are called reference locus.It is selected and depends on the requirement of designer to system completely.Common reference locus is as follows:

y _r(n+i)＝X(n+i)-α ⁱ[X(n)-y _P(n)] (2)

In formula (2): y _r(n+i) be the reference locus in (n+i) moment;

X (n+i) is the set point in (n+i) moment;

Y _pn process real output value that () is the n moment;

α is the speed degree that reference locus is tending towards set point, generally gets wherein, T _othe sampling time, T _rit is the Expected Response time of reference locus.

From rolling optimization principle, every one-step optimization is all be based upon on latest data basis that real process obtains, therefore y _r(n)=y _p(n), y _rn () is the process desired output in n moment.

For plan tracking fixed valure, usually can think:

X(n+i)＝X(n) (3)

In formula (3): X (n) is the set point in n moment.

Can obtain reference locus detailed expressions by formula (2) (3) formula is:

$y_r(n+i)=X\left(n\right)+\underset{l=1}{\overset{L_1}{\mathrm{\Σ}}}{c}_l\left(n\right)i^l-{\mathrm{\α}}^i[X\left(n\right)-y_P\left(n\right)]---\left(4\right)$

2, Strategy for Single-Phase Grid-connected Inverter closed-loop control system Mathematical Modeling is set up

The present invention adopts single-phase full-bridge inverter to be control object, and its main circuit structure figure as shown in Figure 1.In Fig. 1, E is DC bus input voltage, u ₁for inverter bridge brachium pontis output voltage, u ₀for inverter output voltage, i _lfor filter inductance electric current, i _cfor filter capacitor electric current, i ₀for load current, r is the comprehensive equivalent damping resistance of various damping factor in the inverter such as equivalent series resistance, dead time effect, switching tube conduction voltage drop, line resistance of filter inductance L.

For the such dual input of single-phase inverter, single second-order system exported, select capacitance voltage u ₀with inductive current i _las state variable, can state-space expression be obtained as follows:

$\left[\begin{array}{c}\stackrel{\·}{u_0}\\ \stackrel{\·}{i_L}\end{array}\right]=\left[\begin{array}{cc}0& \frac{1}{C}\\ -\frac{1}{L}& -\frac{r}{L}\end{array}\right]\left[\begin{array}{c}{u}_0\\ i_L\end{array}\right]+\left[\begin{array}{c}0\\ \frac{1}{L}\end{array}\right]u_1+\left[\begin{array}{c}-\frac{1}{c}\\ 0\end{array}\right]i_0---\left(5\right)$

$y=\left[\begin{array}{cc}1& 0\end{array}\right]\left[\begin{array}{c}\stackrel{\·}{u_0}\\ \stackrel{\·}{i_L}\end{array}\right]---\left(6\right)$

3, by the form that the state-space expression discretization of single-phase inverter control system can be obtained state space equation:

$\left\{\begin{array}{c}{x}_m\left(n\right)={\mathrm{Ax}}_m(n-1)+\mathrm{Bu}(n-1)\\ y_m\left(n\right)={\mathrm{Cx}}_m\left(n\right)\end{array}\right.---\left(7\right)$

In formula (7), y _mn ()---n moment model prediction exports;

X _m(n)---n moment model state value;

U (n-1)---(n-1) moment control inputs;

A, B, C---matrix equation coefficient.

$A=\left[\begin{array}{cc}0& \frac{1}{C}\\ -\frac{1}{L}& -\frac{r}{L}\end{array}\right],B=\left[\begin{array}{cc}0& -\frac{1}{C}\\ \frac{1}{L}& 0\end{array}\right],C=\left[\begin{array}{cc}1& 0\end{array}\right]---\left(8\right)$

4, the model of computational prediction model exports

For the model state value x in (n+i) moment _m(n+i), obtained by above formula (7) recursion

x _m(n)＝Ax _m(n-1)+Bu _n(n-1)

x _m(n+1)＝Ax _m(n)+Bu _n(n)

… (9)

x _m(n+i-1)＝Ax _m(n+i-2)+Bu _n(n+i-2)

x _m(n+i)＝Ax _m(n+i-1)+Bu _n(n+i-1)

Can calculate further

x _m(n+i)＝A ⁱx _m(n)+A ^i-1Bu(n)+A ^i-2Bu(n+1)+…+ABu(n+i-2)+Bu(n+i-1) (10)

It can thus be appreciated that the model prediction in (n+i) moment exports and is

y _m(n+i)＝Cx _m(n+i)＝CA ⁱx _m(n)+CA ^i-1Bu(n)+CA ^i-2Bu(n+1)+…+CABu(n+i-2)+CBu(n+i-1)

(11)

Formula (1) is substituted in formula (11) and can obtain:

In formula (12): μ (n)=[μ ₁(n), μ ₂(n) ..., μ _j(n)] ^t;

g _n(i)＝[g _n1(i),g _n2(i),…,g _nJ(i)] ^T；

, can be obtained by formula (12), the process response function g of basic function meanwhile _njn () can embody calculated off-line and go out before basic function is known:

g _nj(i)＝CA ^i-1Bf _nj(0)+CA ^i-2Bf _nj(1)+…+CBf _nj(i-1) (13)

The controlled quentity controlled variable added in model is not amount separate in time, but the linear combination of basic function used, therefore, its exporting change caused just shows as each basic function response g _njthe linear superposition of (n), but not the superposition of different time points control effect, g _njn () can calculate by off-line, unknown only has linear combination coefficient μ _j(n).

5, the model prediction after calculation compensation exports

In actual industrial process, due to the reason such as model mismatch, noise, model exports exists certain error between the output of process, that is:

e(n)＝y _P(n)-y _m(n) (14)

For the prediction of following (n+i) moment error, can think in the controls:

e(n+i)＝e(n)＝y _P(n)-y _m(n) (15)

In formula (15): e (n) is the error in n moment;

Y _pthe actual output of n process that () is the n moment;

Y _mn model prediction that () is the n moment exports.

Real process prediction exports expression formula:

y _P(n+i)＝y _m(n+i)+e(n+i) (16)

(12), (15) are substituted into (16) and can obtain:

y _P(n+i)＝CA ⁱx _m(n)+μ(n) ^Tg _n(i)+y _P(n)-y _m(n) (17)。

6, controlled quentity controlled variable is solved based on quadratic form fractional order PI target function

In order to make control system have better Control platform, fractional order PI control and PFC being controlled to combine, adopts the new target function of additional proportion, integration, make the controller of derivation have the architectural characteristic of sensu lato ratio, integration.Utilize the target function of fractional order PI algorithm to PFC algorithm to improve, the novel fractional order PI anticipation function algorithm derived not only has the advantage of fractional order PI and PFC algorithm, can also overcome their shortcoming.

$D=\underset{i=1}{\overset{p}{\mathrm{\Σ}}}q\{K_p{\left[\mathrm{\Δe}(n+1)\right]}^2+K_a{\left[e(n+1)\right]}^2+\underset{j=1}{\overset{k}{\mathrm{\Σ}}}{K}_j{\left[e(n+1-j)\right]}^2\}+r\underset{i=1}{\overset{p}{\mathrm{\Σ}}}{\left[u\left(n\right)\right]}^2---\left(18\right)$

In formula (18), p is prediction step, and q, r are controlled quentity controlled variable weighted factor.

Wherein:

$\begin{array}{c}e(n+1)=y_P(n+h_i)-y_r(n+h_i)=y_m(n+h_i)+e(n+h_i)-y_r(n+h_i)\\ ={\mathrm{CA}}^{h_i}{x}_m\left(n\right)+\underset{j=1}{\overset{J}{\mathrm{\Σ}}}{\mathrm{\μ}}_j\left(n\right)g_{\mathrm{nj}}\left(h_i\right)+y_P\left(n\right)-y_m\left(n\right)-C\left(n\right)+{\mathrm{\α}}^{h_i}[C\left(n\right)-y_P\left(n\right)]\end{array}$

$=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}{\mathrm{\μ}}_j\left(n\right)g_{\mathrm{nj}}\left(h_i\right)-\left\{(1-{\mathrm{\α}}^{h_i})\right[C\left(n\right)-y_P\left(n\right)]-C(A^i-I)x_m\left(n\right)\}$

In formula: μ (n)=[μ ₁(n), μ ₂(n) ..., μ _j(n)] ^t

g _n(h _i)＝[g _n1(h _i),g _n2(h _i),…,g _nJ(h _i)] ^T

$d(n+h_i)=(1-{\mathrm{\α}}^{h_i})[X\left(n\right)-y_P\left(n\right)]-C(A^i-I)x_m\left(n\right)$

That is: e (k+1)=μ (k+1) ^tg (i)-d (k+1+i)

In like manner can obtain:

Δe(k+1)＝e(k+1)-e(k)＝Δμ(k+1) ^Τg(i)-Δd(k+1+i)

Δ ²e(k+1)＝Δe(k+1)-Δe(k)＝Δ ²μ(k+1) ^Τg(i)-Δ ²d(k+1+i)

Order $\frac{\∂D}{\∂\mathrm{\μ}}=0,$ Can obtain:

$\mathrm{\μ}=R_p+R_i={\left\{\right[K_a+\underset{j=1}{\overset{k}{\mathrm{\Σ}}}{K}_j]g{\mathrm{Qg}}^T+f{\mathrm{Rf}}^T\}}^{-1}\·[K_p(1-q^{-1})+K_a+\underset{j=1}{\overset{k}{\mathrm{\Σ}}}{K}_j{q}^{-j}]\mathrm{gQ}{d}^T$

$K_a=K_i{T}_s^{\mathrm{\λ}},K_j=-\frac{1+\mathrm{\λ}}{j}{K}_i{T}_s^{\mathrm{\λ}}{q}_{j-1}$

f＝[f _n1(0),f _n2(0),…,f _nJ(0)] ^T

$g=g_n={[g_n\left(h_i\right),g_n\left(h_i\right),...,g_n\left(h_i\right)]}^T=\left(\begin{array}{cccc}{g}_{n1}\left(h_1\right)& g_{n1}\left(h_2\right)& ...& g_{n1}\left(h_{n_s}\right)\\ g_{n2}\left(h_1\right)& g_{n2}\left(h_2\right)& ...& g_{n2}\left(h_{n_s}\right)\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ g_{\mathrm{nJ}}\left(h_1\right)& g_{\mathrm{nJ}}\left(h_2\right)& ...& g_{\mathrm{nJ}}\left(h_{n_s}\right)\end{array}\right)$

$d=d\left(n\right)=\left(\begin{array}{c}d(n+h_1)\\ d(n+h_2)\\ .\\ .\\ .\\ d(n+h_{n_s})\end{array}\right)$

$d(n+h_i)=(1-{\mathrm{\α}}^{h_i})[X\left(n\right)-y_P\left(n\right)]-C(A^i-I)x_m\left(n\right)$

g _nj(i)＝CA ^i-1Bf _nj(0)+CA ^i-2Bf _nj(1)+…+CBf _nj(i-1)

Wherein, the controlled quentity controlled variable that u (n) is inverter control system n-th moment exports, R _pfor proportional resistance, R _ifor integration item resistance, f _n(0) for basic function is at the initial value in the n-th moment, 1,2 ..., l ..., k is sampled signal sequence, and k is positive integer, q _j-1for binomial coefficient, Q and R represents error weighting matrix respectively and controls weighting matrix, q ^-1and q ^-2for time delay operator, J represents the exponent number of basic function, and j is the positive number between 0 to J, K _p, K _ibe respectively broad sense proportional coefficient, integral item coefficient, T _sfor time step, λ is integral parameter, f _nl(0) be the initial value of l rank basic function in the n-th moment, f is the matrix be made up of at the initial value in the n-th moment 1 to J rank basic function, n _sfor optimizing the number of time domain match point, i is 1 to n _sbetween integer, h _ibe the numerical value in i-th match point, g _nji (), for i-th match point on jth time basic function is at the process response function in the n-th moment, g is the matrix be made up of in the value of the n-th etching process response function J basic function, the standard value that X (n) is inverter output current, y _pn () is the current value that the n-th moment inverter exports, T _ofor the sampling time, T _rfor the Expected Response time of inverter output current reference locus, α is the speed degree that reference locus is tending towards inverter output current standard value, x _mn () is the state value of the n-th moment inverter control system model;

Because controlled quentity controlled variable equation $\begin{array}{c}u(n+i)=\underset{j=1}{\overset{J}{\mathrm{\Σ}}}{\mathrm{\μ}}_j\left(n\right)f_{\mathrm{nj}}\left(i\right)\\ =\mathrm{\μ}{\left(n\right)}^T{f}_n\left(0\right)\end{array}$

Final controlled quentity controlled variable can be obtained: u (n)=(R _p+ R _i) ^tf _n(0)

According to the inventive method structure digital control hardware experiment platform of single-phase inverter as shown in Figure 2, experiment porch is according to compatible and modular design principle, and experiment porch is made up of host computer (PC), simulator, digital signal processor (DSP) control circuit module, main circuit power and drive circuit module thereof, load, auxiliary power module and a logical news modular circuit.

Host computer (PC) major function one operates the normal experiment designed and developed: the connection of system, the startup of system and stopping, the parameters such as SPWM modulation system, modulating frequency, modulation ratio, carrier wave ratio are set, the control strategy of electric current and voltage double-loop control is selected, adjusting of closed loop control parameters, receives slave computer return data and Dynamic Announce SPWM modulating wave, output voltage and current waveform.Another major function is exactly all algorithms of inverter control, by DSP composing software editor relative program, carries out in-circuit emulation and debugging operations by simulator to slave computer DSP.

Digital signal processor (DSP) control circuit module mainly comprises: digital signal processor (DSP), feedback of voltage and current signal conditioning circuit, voltage zero-crossing detection circuit, current foldback circuit, level shifting circuit, condition indication circuit, communication interface circuit and jtag interface circuit.

Main circuit power and driver module thereof mainly comprise: EMI input filter circuit, AC/DC rectification circuit, DC/AC full bridge inverter, power switch tube drives and buffer circuit, LC output filter circuit, inductive current and output voltage sampling circuit.

Auxiliary power module provides low-tension supply for control circuit normally works, and mainly comprises: four tunnels are power switch driving power, inductive current and output voltage sampling circuit and signal conditioning circuit power supply, digital signal processor (DSP) working power independently.Need way many by analyzing known accessory power supply above, require voltage stabilization, ripple is little.By reference to the accompanying drawings 1, reference current i _l ^*with actual grid-connected current instantaneous feedback value i _lthe process of difference via controller after, compare as modulating wave and triangular carrier, export SPWM control signal, through amplifying rear drive power switch pipe, then obtain grid-connected current after inductor filter, thus form whole reversals.Specifically in accordance with the following methods:

The parameter of step 1, the following inverter control system of initialization: the equivalent series resistance R of filter inductance L, filter capacitor C, filter inductance L _l; And single-phase inverter dual input, single second-order system exported are converted into state space equation, draw coefficient matrices A, B, C;

Step 2, according to following formula calculate controlled quentity controlled variable u (n):

u(n)＝(R _p+R _i) ^Tf _n(0)

Wherein:

$R_p+R_i={\left\{\right[K_i{T}_s^{\mathrm{\λ}}+\underset{l=1}{\overset{k}{\mathrm{\Σ}}}(-\frac{1+\mathrm{\λ}}{l}{K}_i{T}_s^{\mathrm{\λ}}{q}_{j-1})]g{\mathrm{Qg}}^T+\mathrm{fR}{f}^T\}}^{-1}\·[K_p(1-q^{-1})+K_i{T}_s^{\mathrm{\λ}}+\underset{l=1}{\overset{k}{\mathrm{\Σ}}}(-\frac{1+\mathrm{\λ}}{l}{K}_i{T}_s^{\mathrm{\λ}}{q}_{j-1})q^{-l}]\mathrm{gQ}{d}^T$

q ₀＝1， $q_j=(1-\frac{1+\mathrm{\λ}}{j})q_{j-1},$

f＝[f _n1(0),f _nj(0),…,f _nJ(0)] ^T，

$g=g_n={[g_{n1}\left(h_i\right),g_{n2}\left(h_i\right),...,g_{\mathrm{nj}}\left(h_i\right),...,g_{\mathrm{nJ}}\left(h_i\right)]}^T=\left(\begin{array}{cccc}{g}_{n1}\left(h_1\right)& g_{n1}\left(h_2\right)& ...& g_{n1}\left(h_{n_s}\right)\\ g_{n2}\left(h_1\right)& g_{n2}\left(h_2\right)& ...& g_{n2}\left(h_{n_s}\right)\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ g_{\mathrm{nj}}\left(h_1\right)& g_{\mathrm{nj}}\left(h_2\right)& ...& g_{\mathrm{nj}}\left(h_{n_s}\right)\\ .& .& .& .\\ .& .& .& .\\ .& .& .& .\\ g_{\mathrm{nJ}}\left(h_1\right)& g_{\mathrm{nJ}}\left(h_2\right)& ...& g_{\mathrm{nJ}}\left(h_{n_s}\right)\end{array}\right)$

$d=d\left(n\right)=\left(\begin{array}{c}d(n+h_1)\\ d(n+h_2)\\ .\\ .\\ .\\ d(n+h_{n_s})\end{array}\right),$

$d(n+h_i)=(1-{\mathrm{\α}}^{h_i})[X\left(n\right)-y_P\left(n\right)]-C(A^i-I)x_m\left(n\right),$

Wherein, the controlled quentity controlled variable that u (n) is inverter control system n-th moment exports, R _pfor proportional resistance, R _ifor integration item resistance, f _n(0) for basic function is at the initial value in the n-th moment, 1,2 ..., l ..., k is sampled signal sequence, and k is positive integer, q _j-1for binomial coefficient, Q and R represents error weighting matrix respectively and controls weighting matrix, q ^-1and q ^-2for time delay operator, J represents the exponent number of basic function, and j is the positive number between 0 to J, K _p, K _ibe respectively broad sense proportional coefficient, integral item coefficient, T _sfor time step, λ is integral parameter, f _nl(0) be the initial value of l rank basic function in the n-th moment, f is the matrix be made up of at the initial value in the n-th moment 1 to J rank basic function, n _sfor optimizing the number of time domain match point, i is 1 to n _sbetween integer, h _ibe the numerical value in i-th match point, g _nji (), for i-th match point on jth time basic function is at the process response function in the n-th moment, g is the matrix be made up of in the value of the n-th etching process response function J basic function, the standard value that X (n) is inverter output current, y _pn () is the current value that the n-th moment inverter exports, T _ofor the sampling time, T _rfor the Expected Response time of inverter output current reference locus, α is the speed degree that reference locus is tending towards inverter output current standard value, x _mn () is the state value of the n-th moment inverter control system model.

Step 3, the controlled quentity controlled variable u (n) obtained according to step 2 control output current thus affect its output voltage, realize reversals.

The controlled quentity controlled variable that prediction function controller exports is converted into sinusoidal signal and inputs inverse changing driving circuit after amplifying by DSP microprocessor, change inverter output voltage, after the output voltage of inverter changes, output current changes accordingly, thus control output voltage, just can carry out tracing control to inverter output voltage by such cyclic process, realize the same frequency homophase of output current output voltage.

In order to verify the effect of the inventive method, carry out following experiment: relevant parameter chosen by the topological structure according to single-phase full-bridge inverter, filter inductance L=0.003H, filter capacitor C=0.0005F, the equivalent series resistance R of filter inductance L _l=0.01 Ω, load R=5 Ω, K _p=5, K _i=1500, λ=0.005, Q=0.3, R=0..3, T _r=0.05, T=0.0045.Utilize MATLAB simulated environment to build single-phase full-bridge inverter simulation model, adopt fractional order PI Predictive function control to carry out emulation experiment, the basic parameter chosen is consistent.Experiment is from aspect analyses such as stability, steady-state error, anti-interferences.Fig. 3 shows that fractional order PI prediction function controller achieves the same frequency homophase of output voltage and output current, and output current reference waveform can be good at following the tracks of line voltage, and output waveform is level and smooth, and steady-state characteristic is good.Fig. 4 and Fig. 5 is engraved in shock load in output circuit when being respectively 0.04S, the wave form varies figure of output current.Known by the little block diagram in the comparative analysis upper right corner (0.04s moment current break partial enlargement oscillogram), fractional order PI Predictive function control lower system response time is shorter, and workload-adaptability is strong, and antijamming capability is strong.

In sum, the inverter control method steady-state characteristic based on fractional order PI anticipation function of the present invention is good, the response time is shorter, and workload-adaptability is strong, and antijamming capability is strong, is applicable to combining inverter in wind generator system.

Claims (2)

1. based on an inverter control method for fractional order PI anticipation function, comprising: set up inverter control system model, determine inverter output current reference locus, it is characterized in that, comprise the steps:

Step 1, the parameter of the following inverter control system model of initialization: the equivalent series resistance R of filter inductance L, filter capacitor C, filter inductance L _l; And the second-order system that inverter dual input list exports is converted into state space equation, draw coefficient matrix Α, Β, C, $A=\left[\begin{array}{cc}0& \frac{1}{C}\\ -\frac{1}{L}& -\frac{r}{L}\end{array}\right],$ $B=\left[\begin{array}{cc}0& -\frac{1}{C}\\ \frac{1}{L}& 0\end{array}\right],$ C=[1 0], r are the comprehensive equivalent damping resistance of the various damping factor of inverter;

Step 2, calculates controlled quentity controlled variable u (n) according to following formula:

u(n)＝(R _p+R _i) ^Tf _n(0)，

Wherein:

$R_p+R_i={\left\{\right[K_i{T}_s^{\mathrm{\λ}}+\underset{l=1}{\overset{k}{\mathrm{\Σ}}}(-\frac{1+\mathrm{\λ}}{l}{K}_i{T}_s^{\mathrm{\λ}}{q}_{l-1})]\mathrm{gQ}{g}^T+\mathrm{fR}{f}^T\}}^{-1}\·[K_p(1-q^{-1})+K_i{T}_s^{\mathrm{\λ}}+\underset{l=1}{\overset{k}{\mathrm{\Σ}}}(-\frac{1+\mathrm{\λ}}{l}{K}_i{T}_s^{\mathrm{\λ}}{q}_{l-1})q^{-l}]\mathrm{gQ}{g}^T,$

q ₀＝1, $q_l=(1-\frac{1+\mathrm{\λ}}{l})q_{l-1},$

f＝[f _n1(0),f _nj(0),…,f _nJ(0)] ^T，

$g=g_n=[g_{n1}\left(h_i\right),g_{n2}\left(h_i\right),\·\·\·,g_{\mathrm{nj}}\left(h_i\right),\·\·\·,g_{\mathrm{nJ}}\left(h_i\right){]}^T=\left(\begin{array}{cccc}{g}_{n1}\left(h_1\right)& g_{n1}\left(h_2\right)& \·\·\·& g_{n1}\left(h_{n_s}\right)\\ g_{n2}\left(h_1\right)& g_{n2}\left(h_2\right)& \·\·\·& g_{n2}\left(h_{n_s}\right)\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ g_{\mathrm{nj}}\left(h_1\right)& g_{\mathrm{nj}}\left(h_2\right)& \·\·\·& g_{\mathrm{nj}}\left(h_{n_s}\right)\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ \·& \·& \·& \·\\ g_{\mathrm{nJ}}\left(h_1\right)& g_{\mathrm{nJ}}\left(h_2\right)& \·\·\·& g_{\mathrm{nJ}}\left(h_{n_s}\right)\end{array}\right),$

$d=d\left(n\right)=\left(\begin{array}{c}d(n+h_1)\\ d(n+h_2)\\ \·\\ \·\\ \·\\ d(n+h_{n_s})\end{array}\right),$ Wherein, $d(n+h_i)=(1-{\mathrm{\α}}^{h_i})[X\left(n\right)-y_P\left(n\right)]-C(A^i-I)x_m\left(n\right),$

g _nj(i)＝CA ^i-1Bf _nj(0)+CA ^i-2Bf _nj(1)+…+CBf _nj(i-1),

$\mathrm{\α}=\exp \left[\frac{-3T_o}{T_r}\right],$

The controlled quentity controlled variable that u (n) is inverter control system n-th moment exports, R _pfor proportional resistance, R _ifor integration item resistance, f _n(0) for basic function is at the initial value in the n-th moment, 1,2 ..., l ..., k is sampled signal sequence, and k is positive integer, q _l-1for binomial coefficient, Q and R represents error weighting matrix respectively and controls weighting matrix, q ^-1and q ^-lfor time delay operator, J represents the exponent number of basic function, and j is the positive number between 0 to J, K _p, K _ibe respectively broad sense proportional coefficient, integral item coefficient, T _sfor time step, λ is integral parameter, f _nl(0) be the initial value of l rank basic function in the n-th moment, f is the matrix be made up of at the initial value in the n-th moment 1 to J rank basic function, n _sfor optimizing the number of time domain match point, i is 1 to n _sbetween integer, h _ibe the numerical value in i-th match point, g _nji (), for i-th match point on jth time basic function is at the process response function in the n-th moment, g is the matrix be made up of in the value of the n-th etching process response function J basic function, the standard value that X (n) is inverter output current, y _pn () is the output current value of the n-th moment inverter, T _ofor the sampling time, T _rfor the Expected Response time of inverter output current reference locus, α is the speed degree that reference locus is tending towards inverter output current standard value, x _mn () is the state value of the n-th moment inverter control system model;

Step 3, the output current of controlled quentity controlled variable u (n) control inverter obtained according to step 2 thus affect its output voltage.

2. a kind of inverter control method based on fractional order PI anticipation function as claimed in claim 1, it is characterized in that, basic function described in step 2 is unit step function, and the value of exponent number J is 1.