CN106787874A - Clean energy resource electricity generation grid-connecting inverter Finite State Model forecast Control Algorithm - Google Patents

Abstract

The invention discloses a kind of clean energy resource electricity generation grid-connecting inverter Finite State Model forecast Control Algorithm, step is as follows, step S1, defines on off state S_i；S2, obtains output voltage U_jWith on off state S_iExpression formula；S3, constructs power prediction model；S4, construction cost function g；S5, initialization；S6, gathers line voltage and output current；S7, calculates output voltage U under current switch states_j；S8, calculates first time power prediction value；S9, calculates second power prediction value；S10, calculates cost function g；The size of S11, relative value function g and comparison variable m, and minimum value is assigned to comparison variable m；S12, judges and exports.The present invention carries out two-staged prediction by power output, and optimal voltage vector is calculated in advance, and effective compensation is carried out to algorithm time delay, reduces the influence that delay on system performance is produced.After adding compensation of delay, when sample frequency is higher, control strategy of the present invention can be substantially reduced power swing and reduce grid-connected current harmonic distortion.

Description

Clean energy resource electricity generation grid-connecting inverter Finite State Model forecast Control Algorithm

Technical field

The invention belongs to intelligent power grid technology field, and in particular to a kind of clean energy resource electricity generation grid-connecting inverter finite state Model predictive control method.

Background technology

With becoming increasingly conspicuous for environmental pollution and the contradiction such as energy resource structure is unbalance, clean energy resource generates electricity and is increasingly closed Note.Combining inverter is the Core equipment that clean energy resource generates electricity, and its control system performance is directly influenced and network electric energy quality.

Finite state mould PREDICTIVE CONTROL has a good adaptivity and robustness, and do not need inner ring current control and PWM, the advantages of control program is easily realized, progressively as the focus of combining inverter control strategy research.Document " three-phase Combining inverter model current Prediction and Control Technology electrotechnics journals, 2011,26 (6):153-159 " devises one kind to be used for The model current predictive control strategy of three-phase grid-connected inverter.Document " Model-predictive control of grid- tied four-level diode-clamped inverters for high-power wind energy conversion systems.IEEE Transactions onPower Electronics,2014,29(6):2861-2873 " proposes one kind For three level clamping combining inverter model current forecast Control Algorithms of wind-power electricity generation.Document " three-phase voltage type PWM rectifications The limited on off sequence model prediction current control electrotechnics journals of device, 2013,28 (12)：182-190 " is by model prediction control System is applied to the design of Three-Phase PWM Rectifier control strategy.Because the amount of calculation that MPC strategies need is larger, control algolithm time lag meeting Devices switch state delay is caused to update.Simultaneously as switching frequency is higher, cause switching loss higher, output-power fluctuation It is larger so that energy conversion efficiency is relatively low.

The content of the invention

The present invention is to solve the control strategy of existing combining inverter is computationally intensive, exists when performing control strategy and prolong When, cause devices switch state delay to update so that poor system performance.And existing control strategy breaker in middle frequency is higher, make Higher into switching loss, output-power fluctuation is larger so that the more low technical problem of energy conversion efficiency, so as to provide a kind of logical Compensation of delay is realized in overpower prediction, while the model prediction grid-connected control method of switching frequency can be reduced.

In order to solve the above technical problems, the technical solution adopted in the present invention is as follows：A kind of clean energy resource electricity generation grid-connecting is inverse Become device Finite State Model forecast Control Algorithm, step is as follows,

Step S1, defines the on off state S of combining inverter_i；

Wherein, i is the phase of AC network, and i ∈ (a, b, c)；

S2, obtains the output voltage U of combining inverter under dq rotational coordinates_jWith on off state S_iExpression formula；Specifically such as Under：

Wherein, U_dcIt is DC bus-bar voltage, θ is line voltage space angle；S_aIt is the on off state value of a phases；S_bIt is b phases On off state value；S_cIt is the on off state value of c phases；u_dIt is the d components of output voltage；u_qIt is the q components of output voltage；

S3, construction combining inverter and output voltage vector U_jRelevant power prediction model；

Concretely comprise the following steps, S3.1, according to Kirchhoff's law, obtain shape of the combining inverter under three-phase static coordinate system State equation；

Wherein, u_anIt is a phase output voltages of combining inverter；u_bnIt is the b phase output voltages of combining inverter；u_cnFor simultaneously The c phase output voltages of net inverter；i_aIt is a phase output currents of combining inverter；i_bIt is the b phase output currents of combining inverter； i_cIt is the c phase output currents of combining inverter；e_aIt is power network a phase voltages；e_bIt is power network b phase voltages；e_cIt is power network c phase voltages；L It is inductance；R is resistance；u_nNIt is the voltage between dc bus and three phase network neutral point；

S3.2, Park conversion is carried out to the formula 3 in step S3.1, obtains the state equation under dq rotational coordinates；Specifically It is as follows：

Wherein, L is inductance；R is resistance；ω is electrical network angular frequency；e_dIt is the d components of line voltage；e_qIt is line voltage Q components；i_dIt is the d components of the output current of combining inverter；i_qIt is the q components of the output current of combining inverter；u_dIt is output The d components of voltage；u_qIt is the q components of output voltage；

S3.3, because resistance R is smaller, the influence of negligible resistance R obtains the discretization Final finishing of formula 5 in step S3.2：

Wherein, i_d(k+1) it is t_k+1The d components of moment output current predicted value；i_q(k+1) it is t_k+1Moment output current is pre- The q components of measured value；i_dK () is t_kThe d components of moment output current；i_qK () is t_kThe q components of moment output current；e_dIt is power network The d components of voltage；e_qIt is the q components of line voltage；u_dK () is t_kThe d components of the output voltage of moment combining inverter；u_q(k) It is t_kThe q components of the output voltage of moment combining inverter；L is inductance；T is sample frequency；

S3.4, according to instantaneous power theory, under dq coordinate systems, combining inverter output instantaneous active power P and idle Power Q is：

S3.5, t is obtained by the discretization of formula 7 in step S3.4_k+1The power prediction model of moment combining inverter：

Wherein, i_d(k+1) it is t_k+1The d components of moment output current predicted value；i_q(k+1) it is t_k+1Moment output current is pre- The q components of measured value；i_dK () is t_kThe d components of moment output current；i_qK () is t_kThe q components of moment output current；e_dIt is power network The d components of voltage；e_qIt is the q components of line voltage；P (k+1) is t_k+1Moment active power predicted value；Q (k+1) is t_k+1Moment Reactive power predicted value；P (k) is t_kMoment active power predicted value；Q (k) is t_kMoment reactive power predicted value；

Formula 6 in step S3.3 is brought into formula 8, combining inverter and output voltage U is obtained_jRelevant power is pre- Survey model：

Wherein, i_dIt is the d components of combining inverter output current；i_qIt is the q components of combining inverter output current；e_dFor The d components of line voltage；e_qIt is the q components of line voltage；P (k+1) is t_k+1Moment active power predicted value；Q (k+1) is t_k+1 Moment reactive power predicted value；P (k) is t_kMoment active power predicted value；Q (k) is t_kMoment reactive power predicted value；u_d(k) It is t_kThe d components of moment combining inverter output voltage；u_qK () is t_kThe q components of moment combining inverter output voltage；L is electricity Sense；T is sample frequency；ω is electrical network angular frequency；

S3.6, the formula 9 in step S3.5 obtains t_k+2Moment combining inverter and output voltage U_jRelevant power Forecast model；Specially：

Wherein, i_dIt is the d components of combining inverter output current；i_qIt is the q components of combining inverter output current；e_dFor The d components of line voltage；e_qIt is the q components of line voltage；P (k+1) is t_k+1Moment active power predicted value；Q (k+1) is t_k+1 Moment reactive power predicted value；P (k+2) is t_k+2Moment active power predicted value；Q (k+2) is t_k+2Moment reactive power is predicted Value；u_d(k+1) it is t_k+1The d components of moment combining inverter output voltage；u_q(k+1) it is t_k+1Moment combining inverter output electricity The q components of pressure；L is inductance；T is sample frequency；ω is electrical network angular frequency.

S4, construction cost function g；

G=| P^*-P(k+2)|+|Q^*-Q(k+2)|+λS (11)；

Wherein, P^*It is active power reference value, Q^*It is reactive power reference qref, λ is weight coefficient；S is from current switch shape State S_iK () arrives subsequent time on off state S_i(k+1) in renewal process, the total degree that on off state changes；

The computing formula of on off state change frequency S is：

Wherein, S_iIt is the on off state of combining inverter；S_aK () is t_kThe on off state of moment combining inverter a phase；S_b K () is t_kThe on off state of moment combining inverter b phase；S_cK () is t_kThe on off state of moment combining inverter c phase；S_a(k+ 1) it is t_+1kThe on off state of moment combining inverter a phase；S_b(k+) it is t_k+1The on off state of moment combining inverter b phase；S_c(k + 1) it is t_k+1The on off state of moment combining inverter c phase；

S5, initialization gives the comparison variable m of cost function g, and to comparison variable m and on off state S_iAssign initial value；

S6, collection line voltage e_a、e_b、e_c, carry out the d components e that dq conversion obtains line voltage_dWith q components e_q；Collection is simultaneously The output current i of net inverter_a、i_b、i_cAnd carry out the d components i that dq conversion obtains combining inverter output current_dWith q components i_q；

S7, the output voltage U of the combining inverter under current switch states is calculated with reference to step S2 and S6_j；

S8, the first time power prediction value of combining inverter is calculated with reference to step S3 and step S7；

S9, second power prediction value of combining inverter is calculated with reference to step S3, step S7 and step S8；

S10, cost function g is calculated with reference to step S4 and step S9；

The size of S11, relative value function g and comparison variable m, and minimum value is assigned to comparison variable m；

S12, judges whether cycle-index reaches setting value, when cycle-index is less than setting value, changes on off state value, Repeat step S6-S11；When cycle-index is equal to setting value, the output voltage vector corresponding to minimum value function g is exported U_j；Output voltage vector U_jCorresponding on off state is applied to subsequent time, realizes direct Power Control.

The present invention carries out two-staged prediction by power output, and optimal voltage vector is calculated in advance, and algorithm time delay is entered Row effective compensation, reduces the influence that delay on system performance is produced.After adding compensation of delay, when sample frequency is higher, this hair Bright control strategy can be substantially reduced power swing and reduce grid-connected current harmonic distortion.

In clean energy resource grid-connected system, power device switching frequency is lower, and power attenuation is just smaller.In the present invention By to additional control target item is introduced in cost function, realizing corresponding control performance optimization.To reduce switching frequency, design Switching frequency addition Item λ S.After adding switching frequency addition Item, suitable weight coefficient is selected, can when sample frequency is higher The switching frequency of system is significantly reduced, system grid-connected current quality while switching frequency is reduced also is met requirement.System System can fast and flexible regulation power output, with good dynamic property.

Brief description of the drawings

Fig. 1 is the circuit diagram of clean energy resource generating three-phase combining inverter of the present invention.

Fig. 2 is clean energy resource generating three-phase combining inverter voltage vector-diagram of the present invention.

Fig. 3 is that Finite State Model of the present invention predicts coordinated control system structure chart.

Fig. 4 is for ideally in the absence of time delay, algorithm performs process.

Fig. 5 is the presence of time delay, but does not carry out the algorithm performs process of compensation of delay

Fig. 6 carries out the algorithm calculating process after compensation of delay for system.

Specific embodiment

As Figure 1-3, a kind of clean energy resource electricity generation grid-connecting inverter Finite State Model forecast Control Algorithm, step is such as Under,

Step S1, defines the on off state S of combining inverter_i；

Wherein, i is the phase of AC network, and i ∈ (a, b, c).

S2, obtains the output voltage U of combining inverter under dq rotational coordinates_jWith on off state S_iExpression formula；Specifically such as Under：

Wherein, U_dcIt is DC bus-bar voltage, θ is line voltage space angle；S_aIt is the on off state value of a phases；S_bIt is b phases On off state value；S_cIt is the on off state value of c phases；u_dIt is the d components of output voltage；u_qIt is the q components of output voltage.

S3, construction combining inverter and output voltage vector U_jRelevant power prediction model.

Concretely comprise the following steps：S3.1, according to Kirchhoff's law, obtains shape of the combining inverter under three-phase static coordinate system State equation；

Wherein, u_anIt is a phase output voltages of combining inverter；u_bnIt is the b phase output voltages of combining inverter；u_cnFor simultaneously The c phase output voltages of net inverter；i_aIt is a phase output currents of combining inverter；i_bIt is the b phase output currents of combining inverter； i_cIt is the c phase output currents of combining inverter；e_aIt is power network a phase voltages；e_bIt is power network b phase voltages；e_cIt is power network c phase voltages；L It is inductance；R is resistance；u_nNIt is the voltage between dc bus and three phase network neutral point；

S3.2, Park conversion is carried out to the formula 3 in step S3.1, obtains the state equation under dq rotational coordinates；Specifically It is as follows：

Wherein, L is inductance；R is resistance；ω is electrical network angular frequency；e_dIt is the d components of line voltage；e_qIt is line voltage Q components；i_dIt is the d components of the output current of combining inverter；i_qIt is the q components of the output current of combining inverter；u_dIt is output The d components of voltage；u_qIt is the q components of output voltage；

S3.3, because resistance R is smaller, the influence of negligible resistance R obtains the discretization Final finishing of formula 5 in step S3.2：

Wherein, i_d(k+1) it is t_k+1The d components of moment output current predicted value；i_q(k+1) it is t_k+1Moment output current is pre- The q components of measured value；i_dK () is t_kThe d components of moment output current；i_qK () is t_kThe q components of moment output current；e_dIt is power network The d components of voltage；e_qIt is the q components of line voltage；u_dK () is t_kThe d components of the output voltage of moment combining inverter；u_q(k) It is t_kThe q components of the output voltage of moment combining inverter；L is inductance；T is sample frequency；

S3.4, according to instantaneous power theory, under dq coordinate systems, combining inverter output instantaneous active power P and idle Power Q is：

S3.5, t is obtained by the discretization of formula 7 in step S3.4_k+1The power prediction model of moment combining inverter：

Wherein, i_d(k+1) it is t_k+1The d components of moment output current predicted value；i_q(k+1) it is t_k+1Moment output current is pre- The q components of measured value；i_dK () is t_kThe d components of moment output current；i_qK () is t_kThe q components of moment output current；e_dIt is power network The d components of voltage；e_qIt is the q components of line voltage；P (k+1) is t_k+1Moment active power predicted value；Q (k+1) is t_k+1Moment Reactive power predicted value；P (k) is t_kMoment active power predicted value；Q (k) is t_kMoment reactive power predicted value；

Formula 6 in step S3.3 is brought into formula 8, combining inverter and output voltage U is obtained_jRelevant power is pre- Survey model：

Wherein, i_dIt is the d components of combining inverter output current；i_qIt is the q components of combining inverter output current；e_dFor The d components of line voltage；e_qIt is the q components of line voltage；P (k+1) is t_k+1Moment active power predicted value；Q (k+1) is t_k+1 Moment reactive power predicted value；P (k) is t_kMoment active power predicted value；Q (k) is t_kMoment reactive power predicted value；u_d(k) It is t_kThe d components of moment combining inverter output voltage；u_qK () is t_kThe q components of moment combining inverter output voltage；L is electricity Sense；T is sample frequency；ω is electrical network angular frequency；

S3.6, the formula 9 in step S3.5 obtains t_k+2Moment combining inverter and output voltage U_jRelevant power Forecast model；Specially：

Wherein, i_dIt is the d components of combining inverter output current；i_qIt is the q components of combining inverter output current；e_dFor The d components of line voltage；e_qIt is the q components of line voltage；P (k+1) is t_k+1Moment active power predicted value；Q (k+1) is t_k+1 Moment reactive power predicted value；P (k+2) is t_k+2Moment active power predicted value；Q (k+2) is t_k+2Moment reactive power is predicted Value；u_d(k+1) it is t_k+1The d components of moment combining inverter output voltage；u_q(k+1) it is t_k+1Moment combining inverter output electricity The q components of pressure；L is inductance；T is sample frequency；ω is electrical network angular frequency.

S4, construction cost function g；

G=| P^*-P(k+2)|+|Q^*-Q(k+2)|+λS (11)；

Wherein, P^*It is active power reference value, Q^*It is reactive power reference qref, λ is weight coefficient；S is from current switch shape State S_iK () arrives subsequent time on off state S_i(k+1) in renewal process, the total degree that on off state changes.

The computing formula of on off state change frequency S is：

Wherein, S_iIt is the on off state of combining inverter；S_aK () is t_kThe on off state of moment combining inverter a phase；S_b K () is t_kThe on off state of moment combining inverter b phase；S_cK () is t_kThe on off state of moment combining inverter c phase；S_a(k+ 1) it is t_+1kThe on off state of moment combining inverter a phase；S_b(k+) it is t_k+1The on off state of moment combining inverter b phase；S_c(k + 1) it is t_k+1The on off state of moment combining inverter c phase.

S5, initialization gives the comparison variable m of cost function g, and to comparison variable m and on off state S_iAssign initial value.

S6, collection line voltage e_a、e_b、e_c, carry out the d components e that dq conversion obtains line voltage_dWith q components e_q；Collection is simultaneously The output current i of net inverter_a、i_b、i_cAnd carry out the d components i that dq conversion obtains combining inverter output current_dWith q components i_q。

S7, the output voltage U of the combining inverter under current switch states is calculated with reference to step S2 and S6_j。

S8, the first time power prediction value of combining inverter is calculated with reference to step S3 and step S7.

S9, second power prediction value of combining inverter is calculated with reference to step S3, step S7 and step S8.

S10, cost function g is calculated with reference to step S4 and step S9.

The size of S11, relative value function g and comparison variable m, and minimum value is assigned to comparison variable m.

S12, judges whether cycle-index reaches setting value, when cycle-index is less than setting value, changes on off state value, Repeat step S6-S11；When cycle-index is equal to setting value, the output voltage vector corresponding to minimum value function g is exported U_j；Output voltage vector U_jCorresponding on off state is applied to subsequent time, realizes direct Power Control.

In the present invention, there are 7 different voltage vectors in three-phase grid-connected inverter, and the space vector of voltage that it is produced is such as Fig. 2.

After Finite State Model prediction direct Power Control algorithm is applied in grid-connection control system, control algolithm is performed Time delay can be produced.In limited time period, processor needs to complete the analog-to-digital conversion of sampled value, algorithm calculating and updates most Good on off state.Although the time delay performed produced by these algorithms is very short, when sample frequency is higher, if do not carried out effectively Compensation of delay, can also make poor system performance.In addition, to reduce switching loss, improving energy conversion efficiency, grid-connected wanting is being met On the basis of asking, expect to reduce switching frequency as far as possible.Based on considerations above, design is optimized to control system, realize coordinating Control.

Finite State Model prediction coordinated control system structure chart such as Fig. 3.Collection line voltage e_a、e_b、e_cAnd output current i_a、i_b、i_cConverted by dq, be converted to e_d、e_qAnd i_d、i_q.DC voltage U_dcVoltage vector is obtained by formula 2.

Voltage vector is assessed by cost function, the voltage vector for making cost function minimum will be applied to next sampling In the cycle, the corresponding on off state S of the voltage vector is obtained in optimizing link_a、S_b、S_c, so that controlling switch pipe turn-on and turn-off. In Optimal Control Strategy, power prediction function obtains P (k+2), Q (k+2) by two-staged prediction, used as the defeated of cost function Enter, carry out compensation of delay, by improving cost function reduction switching frequency.

To improve control system performance, time delay is compensated using two-staged prediction method.Fig. 4 is do not exist ideally Time delay, algorithm performs process.In t_kAfter instance sample, t is updated immediately_k+1The optimized switch state that moment calculates.

Algorithm is performed in real system can produce time delay, and Fig. 5 is the presence of time delay, but does not carry out the algorithm performs of compensation of delay Process.First sample and calculate optimized switching state S_iK (), updates on off state afterwards, time delay can lead to not to upgrade in time optimal On off state.In t_kAfter instance sample, will be in t₁Continue to apply t in time period_kThe on off state at moment, it is optimal until calculating After on off state, pwm control signal can be just updated.Therefore, there is t₁After time period time delay, t can be influenceed_k+1The switch shape at moment State is selected.

Fig. 6 carries out the algorithm calculating process after compensation of delay for system.t_kInstance sample and application current time switch shape State, t_k+1Moment performance number is estimated using formula 9.Then, the beginning predicted as all on off states, to t_k+2When The power at quarter is predicted, and selects the on off state for making cost function minimum, treats t_k+1Moment updates.Although increased t_k+2Moment Power prediction, but every time sampling after can immediately update on off state.Both contrasts, have preferably in real time after compensation of delay Control performance.After two-staged prediction, t_k+2Moment power prediction function is：

In clean energy resource grid-connected system, power device switching frequency is lower, and power attenuation is just smaller.Model prediction control The cost function of system allows, comprising multiple control targes, variable and constraints, to realize coordinating control.By in cost function Additional control target item is introduced, corresponding control performance optimization is realized.To reduce switching frequency, design switching frequency addition Item λ S.Wherein, S is from current switch states S_iK () arrives subsequent time on off state S_i(k+1) in renewal process, on off state occurs The total degree of change, expression formula is：

S_iEqual to 0 or 1, wherein 0 represents bridge arm shut-off on switching tube, the conducting of lower bridge arm, 1 then in contrast.S_a(k)、S_b (k)、S_cK () is t_kThe on off state at moment, S_a(k+1)、S_b(k+1)、S_c(k+1) it is t_k+1The on off state at moment.

8 on off states as shown in Figure 2, it is assumed that current time U₁(100) it is employed, and U₅(001) it is applied to lower a period of time Carve, then obtain S_a(k)=1, S_b(k)=0, S_c(k)=0；S_a(k+1)=0, S_b(k+1)=0, S_c(k+1)=1.According to formula 12 Obtain S=2.

After the addition Item adds cost function, calculate in optimal vector process, on off state change frequency turns into constraint bar One of part.According to different weight coefficients, it is considered to the constraints, selection swears one group of minimum optimal voltage of on-off times S Amount, reduces switching frequency.

According to power prediction function formula 9, in t_kEstimation obtains t on the basis of the on off state of moment application_k+1Moment has Work(power P (k+1) and reactive power Q (k+1).In order to predict t_k+2The instantaneous active power P (k+2) and reactive power Q (k at moment + 2), it is necessary to detect the power network current i at current time_d、i_q, line voltage e_d、e_q, inverter output voltage u_d(k)、u_qK (), opens Off status S_iAnd DC bus-bar voltage U_dc。

In order to pick out optimal on off state, it is necessary to set up cost function g, the institute that will be predicted by cost function is active Rate value is compared, and selects the minimum voltage vector of cost function of sening as an envoy to and is applied to t_k+1Moment.Simultaneously to reduce switching frequency, Increasing in cost function makes the addition Item S of switching frequency reduction.

The present invention carries out two-staged prediction by power output, and optimal voltage vector is calculated in advance, and algorithm time delay is entered Row effective compensation, reduces the influence that delay on system performance is produced.After adding compensation of delay, when sample frequency is higher, this hair Bright control strategy can be substantially reduced power swing and reduce grid-connected current harmonic distortion.

In clean energy resource grid-connected system, power device switching frequency is lower, and power attenuation is just smaller.In the present invention By to additional control target item is introduced in cost function, realizing corresponding control performance optimization.To reduce switching frequency, design Switching frequency addition Item λ S.After adding switching frequency addition Item, suitable weight coefficient is selected, can when sample frequency is higher The switching frequency of system is significantly reduced, system grid-connected current quality while switching frequency is reduced also is met requirement.System System can fast and flexible regulation power output, with good dynamic property.

Claims (2)

1. a kind of clean energy resource electricity generation grid-connecting inverter Finite State Model forecast Control Algorithm, it is characterised in that：Step is as follows,

Step S1, defines the on off state S of combining inverter_i；

Wherein, i is the phase of AC network, and i ∈ (a, b, c)；

S2, obtains the output voltage U of combining inverter under dq rotational coordinates_jWith on off state S_iExpression formula；It is specific as follows：

$U_j=\left[\begin{array}{c}{u}_d\\ u_q\end{array}\right]=\frac{2}{3}\left[\begin{array}{cc}\cos \mathrm{\θ}& \sin \mathrm{\θ}\\ -\sin \mathrm{\θ}& \cos \mathrm{\θ}\end{array}\right]\left[\begin{array}{ccc}1& -\frac{1}{2}& -\frac{1}{2}\\ 0& \frac{\sqrt{3}}{2}& -\frac{\sqrt{3}}{2}\end{array}\right]\left[\begin{array}{c}{S}_a{U}_{dc}\\ S_b{U}_{dc}\\ S_c{U}_{dc}\end{array}\right]---\left(2\right);$

Wherein, U_dcIt is DC bus-bar voltage, θ is line voltage space angle；S_aIt is the on off state value of a phases；S_bIt is opening for b phases Off status value；S_cIt is the on off state value of c phases；u_dIt is the d components of output voltage；u_qIt is the q components of output voltage；

S3, construction combining inverter and output voltage vector U_jRelevant power prediction model；

Power prediction model is specific as follows：

$\left[\begin{array}{c}P(k+1)\\ Q(k+1)\end{array}\right]=\frac{3T}{2L}\left[\begin{array}{cc}{e}_d& e_q\\ e_q& -e_d\end{array}\right]\left[\begin{array}{c}{u}_d\left(k\right)-e_d+{\mathrm{\ωLi}}_q\\ u_q\left(k\right)-e_q-{\mathrm{\ωLi}}_d\end{array}\right]+\left[\begin{array}{c}P\left(k\right)\\ Q\left(k\right)\end{array}\right]---\left(9\right);$

Wherein, i_dIt is the d components of combining inverter output current；i_qIt is the q components of combining inverter output current；e_dIt is power network The d components of voltage；e_qIt is the q components of line voltage；P (k+1) is t_k+1Moment active power predicted value；Q (k+1) is t_k+1Moment Reactive power predicted value；P (k) is t_kMoment active power predicted value；Q (k) is t_kMoment reactive power predicted value；u_dK () is t_k The d components of moment combining inverter output voltage；u_qK () is t_kThe q components of moment combining inverter output voltage；L is inductance；T It is sample frequency；ω is electrical network angular frequency；

S4, construction cost function g；

G=| P^*-P(k+2)|+|Q^*-Q(k+2)|+λS (11)；

Wherein, P^*It is active power reference value, Q^*It is reactive power reference qref, λ is weight coefficient；S is from current switch states S_i K () arrives subsequent time on off state S_i(k+1) in renewal process, the total degree that on off state changes；

The computing formula of on off state change frequency S is：

$\begin{array}{c}S=\underset{i=a,b,c}{\Σ}|S_i(k+1)-S_i\left(k\right)|=|S_a(k+1)-S_a\left(k\right)|\\ +|S_b(k+1)-S_b\left(k\right)|+|S_c(k+1)-S_c\left(k\right)|\end{array}---\left(12\right);$

Wherein, S_iIt is the on off state of combining inverter；S_aK () is t_kThe on off state of moment combining inverter a phase；S_bK () is t_kThe on off state of moment combining inverter b phase；S_cK () is t_kThe on off state of moment combining inverter c phase；S_a(k+1) it is t_+1kThe on off state of moment combining inverter a phase；S_b(k+1) it is t_k+1The on off state of moment combining inverter b phase；S_c(k+1) It is t_k+1The on off state of moment combining inverter c phase；

S5, initialization gives the comparison variable m of cost function g, and to comparison variable m and on off state S_iAssign initial value；

S6, collection line voltage e_a、e_b、e_c, carry out the d components e that dq conversion obtains line voltage_dWith q components e_q；Gather grid-connected inverse Become the output current i of device_a、i_b、i_cAnd carry out the d components i that dq conversion obtains combining inverter output current_dWith q components i_q；

S7, the output voltage U of the combining inverter under current switch states is calculated with reference to step S2 and S6_j；

S8, the first time power prediction value of combining inverter is calculated with reference to step S3 and step S7；

S9, second power prediction value of combining inverter is calculated with reference to step S3, step S7 and step S8；

S10, cost function g is calculated with reference to step S4 and step S9；

The size of S11, relative value function g and comparison variable m, and minimum value is assigned to comparison variable m；

S12, judges whether cycle-index reaches setting value, when cycle-index is less than setting value, changes on off state value, repeats Step S6-S11；When cycle-index is equal to setting value, the output voltage vector U corresponding to minimum value function g is exported_j；It is defeated Go out voltage vector U_jCorresponding on off state is applied to subsequent time, realizes direct Power Control.

2. clean energy resource electricity generation grid-connecting inverter Finite State Model forecast Control Algorithm according to claim 1, it is special Levy and be：In step s3, concretely comprise the following steps,

S3.1, according to Kirchhoff's law, obtains state equation of the combining inverter under three-phase static coordinate system；

$L\frac{d}{dt}\left[\begin{array}{c}{i}_a\\ i_b\\ i_c\end{array}\right]+R\left[\begin{array}{c}{i}_a\\ i_b\\ i_c\end{array}\right]+\left[\begin{array}{c}{e}_a\\ e_b\\ e_c\end{array}\right]=\left[\begin{array}{c}{u}_{an}\\ u_{bn}\\ u_{cn}\end{array}\right]---\left(3\right);$

$\left[\begin{array}{c}{u}_{an}\\ u_{bn}\\ u_{cn}\end{array}\right]=\left[\begin{array}{c}{u}_{aN}\\ u_{bN}\\ u_{cN}\end{array}\right]-\left[\begin{array}{c}{u}_{nN}\\ u_{nN}\\ u_{nN}\end{array}\right]---\left(4\right);$

Wherein, u_anIt is a phase output voltages of combining inverter；u_bnIt is the b phase output voltages of combining inverter；u_cnFor grid-connected inverse Become the c phase output voltages of device；i_aIt is a phase output currents of combining inverter；i_bIt is the b phase output currents of combining inverter；i_cFor The c phase output currents of combining inverter；e_aIt is power network a phase voltages；e_bIt is power network b phase voltages；e_cIt is power network c phase voltages；L is electricity Sense；R is resistance；u_nNIt is the voltage between dc bus and three phase network neutral point；

S3.2, Park conversion is carried out to the formula 3 in step S3.1, obtains the state equation under dq rotational coordinates；It is specific as follows：

$L\frac{d}{dt}\left[\begin{array}{c}{i}_d\\ i_q\end{array}\right]+\left[\begin{array}{cc}R& -\mathrm{\ω}L\\ \mathrm{\ω}L& R\end{array}\right]\left[\begin{array}{c}{i}_d\\ i_q\end{array}\right]+\left[\begin{array}{c}{e}_d\\ e_q\end{array}\right]=\left[\begin{array}{c}{u}_d\\ u_q\end{array}\right]---\left(5\right);$

Wherein, L is inductance；R is resistance；ω is electrical network angular frequency；e_dIt is the d components of line voltage；e_qIt is q points of line voltage Amount；i_dIt is the d components of the output current of combining inverter；i_qIt is the q components of the output current of combining inverter；u_dIt is output electricity The d components of pressure；u_qIt is the q components of output voltage；

S3.3, because resistance R is smaller, the influence of negligible resistance R obtains the discretization Final finishing of formula 5 in step S3.2：

$\frac{L}{T}\left[\begin{array}{c}{i}_d(k+1)-i_d\left(k\right)\\ i_q(k+1)-i_q\left(k\right)\end{array}\right]=\left[\begin{array}{c}{u}_d\left(k\right)\\ u_q\left(k\right)\end{array}\right]+\left[\begin{array}{c}\mathrm{\ω}L{i}_q\left(k\right)\\ -\mathrm{\ω}L{i}_d\left(k\right)\end{array}\right]-\left[\begin{array}{c}{e}_d\\ e_q\end{array}\right]---\left(6\right);$

Wherein, i_d(k+1) it is t_k+1The d components of moment output current predicted value；i_q(k+1) it is t_k+1Moment output current predicted value Q components；i_dK () is t_kThe d components of moment output current；i_qK () is t_kThe q components of moment output current；e_dIt is line voltage D components；e_qIt is the q components of line voltage；u_dK () is t_kThe d components of the output voltage of moment combining inverter；u_qK () is t_k The q components of the output voltage of moment combining inverter；L is inductance；T is sample frequency；

S3.4, according to instantaneous power theory, under dq coordinate systems, combining inverter output instantaneous active power P and reactive power Q For：

$\left[\begin{array}{c}P\\ Q\end{array}\right]=\frac{3}{2}\left[\begin{array}{cc}{e}_d& e_q\\ e_q& -e_d\end{array}\right]\left[\begin{array}{c}{i}_d\\ i_q\end{array}\right]---\left(7\right);$

S3.5, t is obtained by the discretization of formula 7 in step S3.4_k+1The power prediction model of moment combining inverter：

$\left[\begin{array}{c}P(k+1)-P\left(k\right)\\ Q(k+1)-Q\left(k\right)\end{array}\right]=\frac{3}{2}\left[\begin{array}{cc}{e}_d& e_q\\ e_q& -e_d\end{array}\right]\left[\begin{array}{c}{i}_d(k+1)-i_d\left(k\right)\\ i_q(k+1)-i_q\left(k\right)\end{array}\right]---\left(8\right);$

Wherein, i_d(k+1) it is t_k+1The d components of moment output current predicted value；i_q(k+1) it is t_k+1Moment output current predicted value Q components；i_dK () is t_kThe d components of moment output current；i_qK () is t_kThe q components of moment output current；e_dIt is line voltage D components；e_qIt is the q components of line voltage；P (k+1) is t_k+1Moment active power predicted value；Q (k+1) is t_k+1Moment is idle Power prediction value；P (k) is t_kMoment active power predicted value；Q (k) is t_kMoment reactive power predicted value；

Formula 6 in step S3.3 is brought into formula 8, combining inverter and output voltage U is obtained_jRelevant power prediction mould Type：

$\left[\begin{array}{c}P(k+1)\\ Q(k+1)\end{array}\right]=\frac{3T}{2L}\left[\begin{array}{cc}{e}_d& e_q\\ e_q& -e_d\end{array}\right]\left[\begin{array}{c}{u}_d\left(k\right)-e_d+{\mathrm{\ωLi}}_q\\ u_q\left(k\right)-e_q-{\mathrm{\ωLi}}_d\end{array}\right]+\left[\begin{array}{c}P\left(k\right)\\ Q\left(k\right)\end{array}\right]---\left(9\right);$

Wherein, i_dIt is the d components of combining inverter output current；i_qIt is the q components of combining inverter output current；e_dIt is power network The d components of voltage；e_qIt is the q components of line voltage；P (k+1) is t_k+1Moment active power predicted value；Q (k+1) is t_k+1Moment Reactive power predicted value；P (k) is t_kMoment active power predicted value；Q (k) is t_kMoment reactive power predicted value；u_dK () is t_k The d components of moment combining inverter output voltage；u_qK () is t_kThe q components of moment combining inverter output voltage；L is inductance；T It is sample frequency；ω is electrical network angular frequency；

S3.6, the formula 9 in step S3.5 obtains t_k+2Moment combining inverter and output voltage U_jRelevant power prediction Model；Specially：

$\left[\begin{array}{c}P(k+2)\\ Q(k+2)\end{array}\right]=\frac{3T}{2L}\left[\begin{array}{cc}{e}_d& e_q\\ e_q& -e_d\end{array}\right]\left[\begin{array}{c}{u}_d(k+1)-e_d+{\mathrm{\ωLi}}_q\\ u_q(k+1)-e_q-{\mathrm{\ωLi}}_d\end{array}\right]+\left[\begin{array}{c}P(k+1)\\ Q(k+1)\end{array}\right]---\left(10\right);$

Wherein, i_dIt is the d components of combining inverter output current；i_qIt is the q components of combining inverter output current；e_dIt is power network The d components of voltage；e_qIt is the q components of line voltage；P (k+1) is t_k+1Moment active power predicted value；Q (k+1) is t_k+1Moment Reactive power predicted value；P (k+2) is t_k+2Moment active power predicted value；Q (k+2) is t_k+2Moment reactive power predicted value；u_d (k+1) it is t_k+1The d components of moment combining inverter output voltage；u_q(k+1) it is t_k+1The q of moment combining inverter output voltage Component；L is inductance；T is sample frequency；ω is electrical network angular frequency.