CN115241912A - A Model Predictive Control Method for Model Mismatch Compensation of Three-phase Grid-connected Inverters - Google Patents

Abstract

The invention provides a model mismatch compensation method for a model predictive control three-phase grid-connected inverter. It is then observed by the model mismatch module. And finally, compensating the observed disturbance to a predicted inverter output voltage initial value, and realizing the control of the inverter through a model prediction controller. The method solves the problem that the performance of the three-phase grid-connected inverter based on model predictive control is reduced due to inaccurate control of the mathematical model of the controlled object on the premise of not improving the complexity of a predictive model. The control efficiency and quality of the MPC are effectively improved, and the reliability of the three-phase grid-connected inverter is improved.

Description

A Model Predictive Control Method for Model Mismatch Compensation of Three-phase Grid-connected Inverters

technical field

The invention relates to the field of predictive control of distributed power generation systems, in particular to a model mismatch compensation method of a three-phase grid-connected inverter for model predictive control.

Background technique

With the increasing computing power of microprocessors, Model Predictive Control (MPC) has been widely studied in the field of power electronic conversion, which enables distributed generation systems, AC motor control, uninterruptible power supplies, etc. to achieve good static and dynamic performance. MPC can decouple the interdependent control loops, thereby improving the dynamic response of the system, so the response speed of the three-phase grid-connected inverter based on the MPC algorithm is much faster than that of the classical controller.

Three-phase grid-connected inverters usually use LCL filtering. Compared with the L filter, the LCL filter has the characteristics of a third-order low-pass filter. Under the same harmonic standard and lower switching frequency, the total inductance value of the LCL filter is smaller than the single inductance value of the L filter.

However, if the three-phase grid-connected inverter based on the MPC algorithm chooses the LCL filter for filtering, there will be some problems. In the control algorithm, the mathematical model of the three-phase grid-connected inverter will become more complicated, which leads to the increase of the difficulty of designing the MPC controller. In terms of hardware design, an additional sampling circuit needs to be configured to obtain the current value flowing into the capacitor, which increases the equipment cost. Therefore, most of the current application process of three-phase grid-connected inverter based on MPC algorithm ignores the inflowing capacitor current, and establishes the dynamic model of the inverter based on the traditional inductance L filter, resulting in the existence of the mathematical model of the controlled object. Unmodeled part, resulting in control error; in addition, the ferromagnetic material of the filter inductor changes according to the magnetization curve, which causes the actual value of the inductor to vary in time under different operating conditions. At the same time, the resistance in the model is not only the stray resistance of the inductor, but also includes other unknown resistances such as line resistance that cannot be effectively measured in the inverter system. Therefore, the inductance and resistance parameters in the model often deviate from the actual object, which reduces the control performance of the MPC algorithm. How to compensate for the above model errors is very important to improve the MPC performance.

In order to improve the control performance of MPC, some parameter identification methods are used in some projects to improve the accuracy of the mathematical model. However, this control strategy has the following disadvantages: 1) it cannot compensate the unmodeled part of the model; 2) it does not observe the resistance parameters; 3) it increases the calculation amount of the controller, thereby increasing the requirements of the inverter system on the control chip.

SUMMARY OF THE INVENTION

In view of this, the purpose of the present invention is to provide a three-phase grid-connected inverter model mismatch compensation method based on model predictive control, which solves the problem that the three-phase grid-connected inverter based on model predictive control is not modeled and parameterized. The inaccurate prediction model caused by time-varying can effectively improve the efficiency and quality of MPC control, and improve the reliability of the three-phase grid-connected inverter.

To achieve the above object, the present invention adopts the following technical solutions:

A model-predictive-controlled three-phase grid-connected inverter model mismatch compensation method, comprising the following steps:

Step S1: Establish a discrete mathematical model of the three-phase grid-connected inverter based on the L filter, and substitute the sampled signal and the given current value into the discrete mathematical model of the three-phase grid-connected inverter based on the L filter, and calculate the prediction. Initial values of inverter output voltage u _anref0 , u _bnref0 , u _cnref0 ;

Step S2: establish the actual mathematical model of the three-phase grid-connected inverter based on the LCL filter, define the voltage generated by the current flowing into the capacitor as the model disturbance, and define the sum of the voltage generated by the resistance and inductance deviation value and the model disturbance as the system. Unknown disturbance; define the collected grid voltage as known disturbance;

Step S3: define the known inductance as the control gain, take the inverter output voltages u _an , _ubn , u _cn as the input of the model mismatch compensation, and the inverter side currents i _a , _ib , _ic as the model mismatch Compensation output, establish a model mismatch compensation module, and based on the model mismatch compensation module, after tuning β _a1 , β _a2 , β _b1 , β _b2 , β _c1 , β _c2 , the unknown disturbances z _a2 , z _b2 , z of the system are observed. _c2 ;

Step S4: compensating the unknown disturbance of the system to the initial values u _anref0 , u _bnref0 , and u _cnref0 of the predicted inverter output voltage to obtain the three-phase predicted inverter output voltages u _anref , u _bnref , and u _cnref ;

Step S5 : Design a cost function according to u _anref , u _bnref , and u _cnref , send the three-phase predicted inverter output voltage to the model prediction controller for traversal, and select a voltage vector for the inverter to control at the next moment.

Further, the step S1 is specifically:

The mathematical model of the grid-connected inverter based on the LCL filter in the static abc coordinate system is:

Among them, u _rx =-L _x2 di _cx /dt-R _x2 i _cx is the model disturbance term; L _x and R _x are the sum of the inductance on both sides of the capacitor and its parasitic resistance, that is, L _x =L _x1 +L _x2 , R _x =R _x1 +R _x2 ;

Ignoring the model disturbance term, an L filter is established to simplify the inverter dynamic model:

Because the sampling period T _s is short enough, the forward Euler method can be used to discretize equation (2) to obtain the discrete mathematical model of the inverter:

When the sampling time is less than the preset time, we get:

The given currents _idref and i _qref in the two-phase rotating coordinate system are transformed as follows:

Substitute equation (5) into equation (3) to obtain the initial value of the predicted inverter output voltage in the abc stationary coordinate system at time k:

Further, the step S2 is specifically:

Assuming that there are inductance-resistance deviations ΔL _x and ΔR _x between the phases of the actual system and the prediction model, the actual mathematical model of the three-phase grid-connected inverter based on the LCL filter is as follows:

Let the voltage errors generated by parameter mismatch be the internal unknown disturbances, respectively

f _a ( i _a )=(u _ra +ΔL _a di _a /dt+ΔR _a i _a )

f _b (i _b )=( _urb +ΔL _b di _b /dt+ΔR _b i _b )

f _c ( _ic )=(u _rc +ΔL _{c dic /dt+ΔR c i c ) ,}

The known external disturbances are the three-phase grid side phase voltages w _a (t)=-e _a , w _b (t)=-e _b and w _c (t)=-e _c , respectively, and the control gains are b _a =1 /L _a , _{bb =1/L b and bc} =1/L _c

but:

Let u _an , _ubn , u _cn be the system inputs u _as , _ubs , u _cs , i _a , ib , ic are the system outputs _{y a1} , y _b1 , _{y c1} ; _ia , _ib , _ic are The state variables x _a1 , x _b1 , and x _c1 construct a first-order system as follows:

Let the state variables x _a2 = f _a ( i _a ), x _b2 = f _b ( i _b ), and x _c2 = f _c ( i _c ), the new system is constructed as:

Further, the step S3 is specifically:

Substitute the inverter output voltages u _an , _ubn , u _cn , the three-phase phase currents i _a , _{ib , ic and predicted inverter output voltage initial values u anref0 , u bnref0 , u cnref0} into the model. The compensation module is equipped with the following calculation:

A-phase model mismatch compensation: e _a1 =z _a1 -i _a ,

b-phase model mismatch compensation: e _b1 =z _b1 -i _b ,

其中,z_a1、z_b1、z_c1是三相电流的估计值,

分别是z_a1、z_b1、z_c1的导数，

分别是z_a2、z_b2、z_c2的导数，e_a1、e_b1、e_c1逆变侧三相电流估计值和三相电流采样值的误差值。C-phase model mismatch compensation: e _c1 =z _{c1 -ic} ,

Among them, z _a1 , z _b1 , z _c1 are the estimated values of the three-phase current,

are the derivatives of z _a1 , z _b1 , and z _c1 , respectively,

They are the derivatives of z _a2 , z _b2 , and z _c2 , respectively, and the error values of the estimated three-phase current at the inverter side of e _a1 , e _b1 , and e _c1 and the sampling value of the three-phase current.

Further, the step S4 is specifically:

Compensate the unknown disturbance of the system to the initial values of the predicted inverter output voltage u _anref0 , u _bnref0 , u _cnref0 to obtain the three-phase predicted inverter output voltage, and obtain the compensated grid-side reference voltage at time k:

Further, the cost function g(k) is specifically:

g(k)=(u _anref (k)-u _a (k)) ² +(u _bnref (k)-u _b (k)) ² +(u _cnref (k)-u _c (k)) ²

Among them, u _a (k), u _b (k), and u _c (k) are the inverter output voltages estimated by each voltage vector.

Compared with the prior art, the present invention has the following beneficial effects:

The invention solves the problem that the mathematical model of the controlled object of the three-phase grid-connected inverter based on model prediction control is inaccurate, effectively improves the MPC control efficiency and quality, and improves the reliability of the three-phase grid-connected inverter.

Description of drawings

Fig. 1 is the method structure block diagram of the present invention;

FIG. 2 is a structural block diagram of predicting the initial value of the inverter output voltage in an embodiment of the present invention;

3 is a block diagram of a model mismatch compensation structure in an embodiment of the present invention;

4 is a hardware structure of a three-phase grid-connected inverter system in an embodiment of the present invention;

FIG. 5 is a coordinate system definition in an embodiment of the present invention.

Detailed ways

The present invention will be further described below with reference to the accompanying drawings and embodiments.

Referring to FIG. 1 , in this embodiment, the system is powered by a DC power supply V _dc , and any inverter topology is connected to the LCL filter and then merged into the power grid. Among them, L _a1 , L _b1 , L _c1 and R _a1 , R _b1 , and R _c1 are the inverter-side inductance and the parasitic resistance of the inverter-side inductance, respectively; L _a2 , L _b2 , L _c2 and R _a2 , R _b2 , R _c2 are grid-side inductance and parasitic resistance of grid-side inductance respectively; C _a , C _b , C _c are filter capacitors; R _a , R _b , R _c are passive damping resistors; filter capacitors _ia , _ib , _ic is the inverter output current; _ica , _icb , _icc are the currents flowing into the filter capacitor; i _ga , _igb , _igc are the three-phase currents on the grid side; u _an , _ubn , u _cn are the inverse The inverter side outputs the phase voltage; e _a , e _b , and e _c are the phase voltages on the grid side.

Figure 2 is a block diagram showing the structure of the initial value of the predicted inverter output voltage. A discrete mathematical model of three-phase grid-connected inverter based on L filter is established, and the three-phase phase current on the inverter side, the phase voltage on the grid side and the three-phase given current i _aref , i _bref , i _cref are used as the predicted inverter output voltage Input the initial value, output the initial value u _anref0 , u _bnref0 , u _cnref0 of the output voltage of the three-phase predicted inverter in the static coordinate system.

Figure 3 shows a block diagram of the model mismatch compensation structure. Among them, z _a1 , z _b1 , and z _c1 are the estimated values of the state variables i _a , _ib , and _ic , respectively; z _a2 , z _b2 , and z _c2 are the estimated values of the model mismatch error. β _a1 , β _a2 , β _b1 , β _b2 , β _c1 , and β _c2 are adjustable correction coefficients. The three-phase phase currents i _a , _{ib , ic , the initial value predictions of the three-phase output voltages on the inverter side u anref0 , u bnref0 , u cnref0} and the inverter output voltages u _an , _{ubn , and u cn are} used as the model mismatch. Input and output three-phase predicted inverter output voltages u _anref , u _bnref , u _cnref .

Referring to FIG. 4 , the entire control system in this embodiment includes: a DC power supply, a three-phase grid-connected inverter, a filter circuit, a relay switch, a DC bus sampling, a three-phase grid voltage acquisition circuit, and a three-phase current acquisition circuit on the inverter side , controller, isolation driver, human-computer interaction part, etc.

The three-phase inverter is provided by a suitable DC power supply. In the inverter, the power tube adopts IGBT or MOSFET with diodes in parallel, and the controller adopts DSP or single-chip microcomputer. The three-phase current acquisition circuit on the inverter side is composed of a combination of a Hall current sensor and an operational amplifier, or a combination of a winding string power resistor followed by a differential operational amplifier. The use of the Hall sensor scheme can effectively realize the electrical isolation of the control loop and the main loop, and the use of the winding string power resistance scheme can reduce the system cost. The grid voltage acquisition circuit is composed of a combination of a precision voltage sensor and an operational amplifier, or a combination of a voltage follower composed of an operational amplifier after dividing the voltage by a parallel resistor. The current detection, voltage sampling circuit, and the output weak current signal are sent to the A/D conversion module of the controller. When entering the grid-connected mode or entering the island mode, the GPIO module of the controller outputs the relay action signal, and then the relay is actuated by the isolation driver after passing through the level conversion chip.

The definition of the coordinate system is shown in Figure 5. The abc axis is a static three-phase coordinate system, the αβ axis is a static two-phase coordinate system, and the dq axis is a rotating two-phase coordinate system.

According to the definition of the coordinate system in Fig. 5, Clark transform is performed on the grid-side phase voltages e _a , e _b , and e _c :

The phase angle θ _g can be calculated according to e _gα and e _gβ :

According to the definition of the coordinate system in Fig. 5, the given currents _idref and i _qref in the two-phase rotating coordinate system are transformed as follows:

In this embodiment, a model mismatch compensation method for a three-phase grid-connected inverter based on model predictive control is proposed, which includes the following steps:

A model-predictive-controlled three-phase grid-connected inverter model mismatch compensation method, characterized by comprising the following steps:

Step S1: Establish a discrete mathematical model of the three-phase grid-connected inverter based on the L filter, and substitute the sampled signal and the given current value into the discrete mathematical model of the three-phase grid-connected inverter based on the L filter, and calculate the prediction. Initial values of inverter output voltage u _anref0 , u _bnref0 , u _cnref0 ;

Step S2: establish the actual mathematical model of the three-phase grid-connected inverter based on the LCL filter, define the voltage generated by the current flowing into the capacitor as the model disturbance, and define the sum of the voltage generated by the resistance and inductance deviation value and the model disturbance as the system. Unknown disturbance; define the collected grid voltage as known disturbance;

Step S3: define the known inductance as the control gain, take the inverter output voltages u _an , _ubn , u _cn as the input of the model mismatch compensation, and the inverter side currents i _a , _ib , _ic as the model mismatch Compensation output, establish a model mismatch compensation module, and based on the model mismatch compensation module, after tuning β _a1 , β _a2 , β _b1 , β _b2 , β _c1 , β _c2 , the unknown disturbances z _a2 , z _b2 , z of the system are observed. _c2 ;

Step S4: compensating the unknown disturbance of the system to the initial values u _anref0 , u _bnref0 , and u _cnref0 of the predicted inverter output voltage to obtain the three-phase predicted inverter output voltages u _anref , u _bnref , and u _cnref ;

Step S5 : Design a cost function according to u _anref , u _bnref , and u _cnref , send the three-phase predicted inverter output voltage to the model prediction controller for traversal, and select a voltage vector for the inverter to control at the next moment.

In this embodiment, the DC voltage passes through the inverter system, and then passes through the LCL filter to eliminate the high-order harmonics of the current and then merges into the power grid; the phase voltages e _a , e _b , and e _c on the collected grid side are transformed into the two-phase static coordinate system to obtain e _gα , e _gβ , where the calculation formula is as follows:

Calculate the phase angle θ _g of the phase voltage on the grid side, and the calculation formula is as follows:

Transform the given currents i _dref and i _qref in the two-phase rotating coordinate system to the three-phase stationary coordinate system to obtain i _aref , i _bref , i _cref , and the calculation formulas are as follows:

Substitute the given currents i _aref , i _bref , i _cref , the three-phase currents i _a , i _b , _{i c} on the inverter side and the phase voltages _{e a} , eb , and ec on the grid side into the three-phase parallel connection based on the L filter. The discrete mathematical model of the grid inverter obtains the initial values of the predicted inverter output voltage u _anref0 , u _bnref0 , u _cnref0 ,

In this embodiment, step S2 is specifically:

Assuming that there are inductance-resistance deviations ΔL _x and ΔR _x between the phases of the actual system and the prediction model, the actual mathematical model of the three-phase grid-connected inverter based on the LCL filter is as follows:

Let the voltage errors generated by parameter mismatch be the internal unknown disturbances, respectively

f _a ( i _a )=(u _ra +ΔL _a di _a /dt+ΔR _a i _a )

f _b (i _b )=( _urb +ΔL _b di _b /dt+ΔR _b i _b )

f _c ( _ic )=(u _rc +ΔL _{c dic /dt+ΔR c i c ) ,}

The known external disturbances are the three-phase grid side phase voltages w _a (t)=-e _a , w _b (t)=-e _b and w _c (t)=-e _c , respectively, and the control gains are b _a =1 /L _a , _{bb =1/L b and bc} =1/Lc

but:

Let u _an , _ubn , u _cn be system inputs u _as , _ubs , u _cs , _{i a} , ib , ic are system outputs y _a1 , y _b1 , y _c1 ; _ia , _ib , ic are state variables x _a1 , x _b1 , x _c1 construct a first-order system as follows:

Let the state variables x _a2 = f _a ( i _a ), x _b2 = f _b ( i _b ), and x _c2 = f _c ( i _c ), the new system is constructed as:

The model mismatch compensation module can perform the state variables x _a1 , x _b1 , x _c1 and x _a2 , x _b2 , x _c2 according to the system input quantities u _an , _{ub bn} , u _cn and the output quantities i _a , _ib , and _ib . estimate, and then realize the dynamic disturbance feedforward compensation, and establish the following equation for the system:

Taking phase a as an example, it can be seen from the above formula that z _a2 is the value of x _a2 after being filtered by a second-order low-pass filter, and z _a1 is the value of x _a1 that is filtered and then phase compensated. The value of β _a2 directly affects the estimation ability of parameter mismatch compensation. With the increase of β _a2 , the anti-interference of the system is enhanced, that is, the time for the system to recover to a steady state after disturbance occurs is short. However, a further increase of β _a2 will make the system oscillate and be sensitive to noise. At this time, β _a1 should be increased to suppress this problem, but if β _a1 is too large, the system will diverge. To sum up, in the actual parameter tuning process, it is necessary to select the observer pole appropriately according to the specific situation.

The same analysis method can illustrate the b-phase and c-phase current loops, and the analysis will not be repeated here.

Finally, add the observation results to get the compensated grid-side reference voltage at time k:

Preferably, in this embodiment, the cost function is designed only with current tracking as the only target. In the current control of grid-connected inverters, current, voltage or power are usually selected as the main constraint variables, and the cost function g(k) is established by taking the voltage constraint variables as an example:

g(k)=(u _anref (k)-u _a (k)) ² +(u _bnref (k)-u _b (k)) ² +(u _cnref (k)-u _c (k)) ²

Among them, u _a (k), u _b (k), and u _c (k) are the inverter output voltages estimated by each voltage vector, because the three-phase current tracking is the same dimension, the coefficient of each square phase is 1 . After the cost function traverses all the vectors, the voltage vector that minimizes g(k) is obtained, and the voltage vector is used for the control of the three-phase grid-connected inverter at the next moment.

The above descriptions are only preferred embodiments of the present invention, and all equivalent changes and modifications made according to the scope of the patent application of the present invention shall fall within the scope of the present invention.

Claims (6)

1. A model mismatch compensation method for a three-phase grid-connected inverter based on model predictive control is characterized by comprising the following steps:

s1, establishing a three-phase grid-connected inverter discrete mathematical model based on an L filter, substituting a sampling signal and a given current value into the three-phase grid-connected inverter discrete mathematical model based on the L filter, and calculating an initial value u of output voltage of a prediction inverter _anref0 、u _bnref0 、u _cnref0 ；

S2, establishing an actual mathematical model of the three-phase grid-connected inverter based on the LCL filter, defining the voltage generated by the current flowing into the capacitor as model disturbance, and defining the sum of the voltage generated by the resistance inductance deviation value and the model disturbance as system unknown disturbance; defining the collected power grid voltage as known disturbance;

s3, defining the known inductance as a control gain and outputting a voltage u by the inverter _an 、u _bn 、u _cn As input for model mismatch compensation, the inverter side current i _a 、i _b 、i _c As the output of the model mismatch compensation, a model mismatch compensation module is established, and beta is set based on the model mismatch compensation module _a1 、β _a2 、β _b1 、β _b2 、β _c1 、β _c2 The unknown disturbance z of the system is observed after the observation _a2 、z _b2 、z _c2 ；

S4, compensating unknown disturbance of the system to a predicted initial value u of the output voltage of the inverter _anref0 、u _bnref0 、u _cnref0 To obtain the output voltage u of the three-phase prediction inverter _anref 、u _bnref 、u _cnref ；

Step S5: according to u _anref 、u _bnref 、u _cnref And designing a cost function, and sending the output voltage of the three-phase prediction inverter into a model prediction controller for traversal, and then selecting a voltage vector for the control of the inverter at the next moment.

2. The model mismatch compensation method for the model predictive controlled three-phase grid-connected inverter according to claim 1, wherein the step S1 specifically comprises:

the mathematical model under the static abc coordinate system of the grid-connected inverter based on the LCL filter is

Wherein u is _rx ＝-L _x2 di _cx /dt-R _x2 i _cx Is a model disturbance term; l is a radical of an alcohol _x And R _x Respectively, the sum of the inductances on both sides of the capacitor and their parasitic resistances, i.e. L _x ＝L _x1 +L _x2 ，R _x ＝R _x1 +R _x2 ；

Neglecting a model disturbance term, establishing an L filter simplified inverter dynamic model:

because of the sampling period T _s The method is short enough, a forward Euler method can be used, and the discrete mathematical model of the inverter is obtained after discretizing the formula (2):

when the sampling time is less than the preset time, obtaining:

for a given current i in a two-phase rotating coordinate system _dref 、i _qref The following transformations are performed:

substituting equation (5) for equation (3) to obtain an initial value of the predicted inverter output voltage at time k abc in a stationary coordinate system:

3. the model mismatch compensation method for the model predictive controlled three-phase grid-connected inverter according to claim 1, wherein the step S2 specifically comprises:

setting that the inductance resistance deviation Delta L exists between each phase of the actual system and the prediction model _x And Δ R _x Then, the actual mathematical model of the three-phase grid-connected inverter based on the LCL filter specifically includes:

let the voltage error generated by parameter mismatch be an internal unknown disturbance respectively

f _a (i _a )＝(u _ra +ΔL _a di _a /dt+ΔR _a i _a )

f _b (i _b )＝(u _rb +ΔL _b di _b /dt+ΔR _b i _b )

f _c (i _c )＝(u _rc +ΔL _c di _c /dt+ΔR _c i _c )，

The external known disturbances are the three-phase network-side phase voltages w _a (t)＝-e _a 、w _b (t)＝-e _b And w _c (t)＝-e _c The control gains are respectively b _a ＝1/L _a 、b _b ＝1/L _b And b _c ＝1/L _c

Then:

let u _an 、u _bn 、u _cn Input u for the system _as 、u _bs 、u _cs ，i _a 、i _b 、i _c Output y for the system _a1 、y _b1 、y _c1 ；i _a 、i _b 、i _c Is a state variable x _a1 、x _b1 、x _c1 The first order system was constructed as follows:

let the state variable x _a2 ＝f _a (i _a )、x _b2 ＝f _b (i _b )、x _c2 ＝f _c (i _c ) The new system is constructed as follows:

4. the model mismatch compensation method for the model predictive controlled three-phase grid-connected inverter according to claim 3, wherein the step S3 specifically comprises:

will output the inverter voltage u _an 、u _bn 、u _cn Phase current i of three phases on the inversion side _a 、i _b 、i _c And predicting an initial value u of the inverter output voltage _anref0 、u _bnref0 、u _cnref0 Substituting the model mismatch compensation module to calculate as follows:

compensation of a-phase model mismatch:

and b-phase model mismatch compensation:

c-phase model mismatch compensation:

wherein z is _a1 、z _b1 、z _c1 Is an estimate of the three-phase current,

are each z _a1 、z _b1 、z _c1 The derivative of (a) of (b),

are each z _a2 、z _b2 、z _c2 Derivative of e _a1 、e _b1 、e _c1 Inversion side three-phase currentAn error value of the estimated value and the three-phase current sample value.

5. The model mismatch compensation method for the model predictive controlled three-phase grid-connected inverter according to claim 4, wherein the step S4 specifically comprises:

compensating unknown disturbance of the system to the initial value u of the output voltage of the prediction inverter _anref0 、u _bnref0 、u _cnref0 Obtaining the output voltage of the three-phase prediction inverter, and obtaining the compensated network side reference voltage at the moment k:

6. the model mismatch compensation method for the model predictive controlled three-phase grid-connected inverter according to claim 1, wherein the cost function g (k) is specifically:

g(k)＝(u _anref (k)-u _a (k)) ² +(u _bnref (k)-u _b (k)) ² +(u _cnref (k)-u _c (k)) ²

wherein u is _a (k)、u _b (k)、u _c (k) Is the predicted inverter output voltage of each voltage vector.

Images