阅读说明：本技术 基于时变扰动补偿的三电平发电系统模型预测控制方法 (Time-varying disturbance compensation based three-level power generation system model prediction control method ) 是由 王军晓 刘义宾 杨海 胡开林 徐建明 俞立 于 2021-08-20 设计创作，主要内容包括：本发明公开了基于时变扰动补偿的三电平发电系统模型预测控制方法；基于机侧、网侧变换器采用三电平结构,在新的拓扑下建立数学模型；模型离散化处理；实时采样被控对象信息并进行坐标变换；设计成本函数作为内环控制器；引入扰动状态变量构建状态空间模型；设计观测器来估计外环的状态；结合观测器估计信息设计外环控制器。本发明外环改进的扩张状态观测器能够对时变扰动很好地抑制,另一方面内环的模型预测控制由于无需调制环节使得系统的动态响应速度加快。(The invention discloses a three-level power generation system model prediction control method based on time-varying disturbance compensation; establishing a mathematical model under a new topology by adopting a three-level structure based on a machine side converter and a network side converter; carrying out model discretization treatment; sampling the information of the controlled object in real time and carrying out coordinate transformation; designing a cost function as an inner loop controller; introducing a disturbance state variable to construct a state space model; designing an observer to estimate the state of the outer ring; and designing an outer loop controller by combining observer estimation information. The extended state observer improved by the outer ring can well inhibit time-varying disturbance, and on the other hand, the dynamic response speed of the system is accelerated due to the fact that the model prediction control of the inner ring does not need a modulation link.)

1. The time-varying disturbance compensation based three-level power generation system model prediction control method is characterized by comprising the following steps of:

step 1: determining a given speed value omega of a machine-side speed ring_ref；

Step 2: establishing a machine side mathematical model;

and step 3: sampling the machine side current and speed, and converting the current information under the three-phase static coordinate into a d-q coordinate system;

and 4, step 4: establishing a discrete permanent magnet synchronous motor current prediction model, wherein the process is as follows:

4.1: determination of the three-level inverter output voltage vector:

let the three-phase sinusoidal voltage expression be:

defining the inverter output voltage as:

then

And U is also provided_aN+U_bN+U_cNWhen the value is equal to 0, then

The relation between the three-bridge arm switching state of the three-level inverter and the output voltage of the inverter can be obtained:

wherein the content of the first and second substances,

the corresponding space voltage vector is defined as:

wherein the content of the first and second substances,

because each bridge arm corresponds to three switch states, 27 groups of switch states can be obtained, and 27 voltage vectors can be obtained by substituting the 27 groups of switch states into a defined space voltage vector formula;

4.2, determining a permanent magnet synchronous motor current prediction model:

discretizing a current state equation by adopting a forward Euler formula to obtain a discrete permanent magnet synchronous motor current prediction model in the following form:

wherein i_d(k),i_q(k) Representing the stator current component under the two-phase synchronous rotation d-q coordinate system at the current moment; i.e. i_d(k+1)，i_q(k +1) is the stator current d, q-axis component at the next moment; u. of_d,u_qThe d and q axis voltage components correspond to 27 switch states; l is_sThe inductance of a stator under a d-q coordinate system in a surface-mounted permanent magnet synchronous motor is obtained; t is_sIs a sampling period;

step 5, constructing a cost function;

since the current loop on the machine side adopts predictive current control, the cost function J₁Designed in the following form:

wherein the content of the first and second substances,a reference value representing a stator current d, q-axis component; i.e. i_d(k+1),i_q(k +1) are (k +1) T respectively_sAt the moment d, a predicted value of the stator current of the q axis is obtained;

step 6, selecting an optimal voltage vector;

firstly, determining an output voltage vector of a three-level inverter by the switching states of three bridge arms of the three-level inverter; then under the action of a prediction model, a prediction value at the current moment can be obtained; finally, selecting an optimal voltage vector u according to a designed cost function_{opt_1}：

u_{opt_1}＝arg min J₁

Step 7, introducing a state variable d_ωlDetermining a new state space model;

considering the uncertainty of the system parameters and the influence of external disturbance, the mechanical motion equation in step 2 can be organized as follows:

wherein the content of the first and second substances,representing machine side rotating speed ring lumped disturbance; b_ω0Is about b_ωWherein, in A reference value representing a q-axis component of the stator current;

let x₁＝ω，x₂＝d_ωlThen the new state space model is:

wherein h is₁Denotes d_ωlDifferentiation of (1); b_ω0Is about b_ωWherein, in A reference value representing a q-axis component of the stator current;

and 8, expanding the design of the state observer, wherein the process is as follows:

designing an extended state observer according to the new state space model in the step 7, wherein the conventional extended state observer is in the form of:

wherein the content of the first and second substances,an estimate representing ω;representing lumped disturbances d_ωlAn estimated value of (d); beta is a₁,β₂Representing the gain of the extended state observer;

defining error variablesThe form of the error state space model is as follows:

when in useWhen the time is that the machine side outer ring lumped disturbance is a constant value, and the coefficient matrix of the error state space model is a Helvelz matrix, the estimated error asymptotically converges to 0, namely the estimated value asymptotically is tracked without error in an actual state;

if the machine side outer ring lumped disturbance is time-varying disturbance, the extended state observer cannot realize asymptotic error-free tracking, so that improvement needs to be carried out on the basis of the observer to achieve the purpose of realizing the time-varying disturbance;

the modified extended state observer is of the form:

wherein the content of the first and second substances,an estimate representing ω;representing lumped disturbances d_ωlAn estimated value of (d); beta is a₁₁,β₁₂,β₁₃Representing the gain of the modified extended state observer;

defining new error variablesThen

From the new error equation above, we can get:

continued derivation of the equation at both ends can yield:

selecting a state variable:arranging into a state space form:

when in useNamely machine side outer ring lumped disturbance satisfies a₁+a₂When t type time-varying disturbance occurs, and the coefficient matrix of the new error state space model is a Helvelz matrix, the estimation error is asymptotically converged to 0;

and 9, designing a machine side outer ring control law, wherein the process is as follows:

selecting an appropriate observer gain β₁₁,β₁₂,β₁₃The estimated value of the actual rotation speed can be obtained by the extended state observer modified in step 8And estimate of outer loop lumped disturbancesThe estimated value obtained by the extended state observer can be used for the design of the controller, and the specific form is as follows:

wherein the content of the first and second substances,an estimate representing ω; omega_refA reference value representing an outer ring of rotational speeds; u. of_ω0Representing the machine side controller output; k is a radical of_ωpRepresenting the controller gain;

step 10, establishing a direct current link mathematical model;

step 11, establishing a network side mathematical model;

step 12, sampling and coordinate transformation of current and voltage on the network side;

step 13, establishing a discrete inner loop power prediction model;

step 14, constructing a cost function;

step 15, selecting an optimal voltage vector;

step 16, introducing a state variable d_ulDetermining a new state space model;

step 17, designing an extended state observer;

and step 18, designing a network side outer ring control law.

2. The time-varying disturbance compensation based three-level power generation system model predictive control method according to claim 1, wherein in the step 2, the specific process is as follows:

the mathematical model of the permanent magnet synchronous motor in the d-q coordinate system can be expressed as follows:

the voltage equation is:

in the formula: u. of_d,u_qRepresenting the d-q axis component of the stator voltage; i.e. i_d,i_qRepresenting the d-q axis component of the stator current; l is_sThe stator inductance under a d-q coordinate system in the surface-mounted permanent magnet synchronous motor meets the requirement of L_s＝L_d＝L_q；R_sRepresenting the stator resistance; omega_reRepresents an electrical angular velocity; psi_fRepresents the permanent magnet flux;

the electromagnetic torque equation is:

wherein p is_nRepresenting the number of pole pairs; t is_eRepresents an electromagnetic torque;

the mechanical equation of motion is:

wherein ω represents a mechanical angular velocity; j represents moment of inertia; b represents a friction coefficient; t is_mRepresenting the drive torque.

3. The time-varying disturbance compensation based three-level power generation system model predictive control method as claimed in claim 1, wherein in said step 10, the current at the dc-side capacitance node P, O, N is represented as:

i_c1＝i_pm-i_pg

i_c1+i_om＝i_c2+i_og

i_c2+i_nm＝i_ng

wherein, C₁,C₂Represents a dc filter capacitance; u. of_c1,u_c2Representing the voltage on the dc bus capacitance; i.e. i_c1,i_c2Representing the current flowing through the dc filter capacitor; i.e. i_pm,i_om,i_nmRepresenting the current, i, flowing through the machine side at node P, O, N_pg,i_og,i_ngIndicating current flow to the net side node P, O, N.

4. The time-varying disturbance compensation based three-level power generation system model predictive control method of claim 1, further characterized by: in the step 11, the net side mathematical model in the d-q coordinate system is as follows:

wherein u is_d,u_qOutputting components of voltage under d, q coordinate system for the three-level inverter; e.g. of the type_d,e_qThe component of the grid side voltage under a d, q coordinate system is shown; i.e. i_d,i_qThe component of the grid side current in a d, q coordinate system is shown; l represents a network side filter inductor; r represents the equivalent resistance of the output end; omega_geRepresenting the grid angular velocity.

5. The time-varying disturbance compensation based three-level power generation system model predictive control method of claim 1, further characterized by: in step 13, the grid-side inverter adopts a voltage-oriented control method, so that a grid-side inverter current equation based on grid voltage vector orientation can be expressed as:

wherein u is_d,u_qOutputting components of voltage under d, q coordinate system for the three-level inverter; e.g. of the type_dIs the d-axis component of the net side voltage; i.e. i_d,i_qThe component of the grid side current in a d, q coordinate system; l represents a network side filter inductor; r represents the equivalent resistance of the output end; omega_geRepresenting the grid angular velocity.

6. The time-varying disturbance compensation based three-level power generation system model predictive control method of claim 1, further characterized by: in the step 14, a cost function, cost function J, is constructed₂The form is as follows:

J₂＝|P^*-P(k+1)|+|Q^*-Q(k+1)|

wherein, P^*,Q^*Representing active power and reactive power reference values; p (k +1) and Q (k +1) are (k +1) T_sAnd predicting values of active power and reactive power at the moment.

7. The time-varying disturbance compensation based three-level power generation system model predictive control method of claim 6, wherein: in step 15, the cost function J is selected from the 27 voltage vectors output from the grid-side inverter₂Minimum voltage vector u_{opt_2}：

u_{opt_2}＝arg min J₂

8. The time-varying based of claim 1The disturbance compensation three-level power generation system model prediction control method is characterized by comprising the following steps of: in the step 16, a state variable d is introduced_ulConstructing a new state space model;

output power P of machine side rectifier without considering converter loss_mCan be expressed as:

P_m＝u_dci_m

wherein u is_dcRepresenting the dc bus voltage, which may be denoted u_dc＝u_c1+u_c2；i_mRepresenting the current output by the machine side converter to the dc bus;

the current flowing through the dc-side capacitor is:

where C represents a dc-side capacitance, and may be represented as C ═ C₁＝C₂；i_gRepresents the current input to the grid-side inverter;

the active power P input from the dc side to the grid side inverter is:

P＝u_dci_g

from the above equation, one can obtain:

is equivalent to

Wherein the content of the first and second substances,representing net side voltage loop lumped disturbances; b_u0Is about b_uWherein, inP^*Representing an active power reference value;

let z₁＝u_dc；z₂＝d_ulThen the new state space model is:

wherein h is₂Denotes d_ulDifferentiation of (1); b_u0Is about b_uWherein, inP^*Representing the active power reference value.

9. The time-varying disturbance compensation based three-level power generation system model predictive control method of claim 1, further characterized by: in step 17, the representation of the modified extended state observer of the outer loop is as follows:

wherein the content of the first and second substances,represents u_dcAn estimated value of (d);representing lumped disturbances d_ulAn estimated value of (d); l₁,l₂,l₃Representing the gain of the modified extended state observer;

when in useThe lumped disturbance of the outer loop of the machine side is constantWhen the error state is in a value, the coefficient matrix of the error state space model is a Helvelz matrix, and the estimated error asymptotically converges to 0, namely the estimated value asymptotically is tracked in an actual state without error;

when in useNamely, the network side outer ring lumped disturbance satisfies a₁+a₂And when t-type time-varying disturbance occurs, the designed extended state observer can realize error-free asymptotic convergence.

10. The time-varying disturbance compensation based three-level power generation system model predictive control method of claim 1, further characterized by: in step 18, the outer loop control law is designed as follows:

selecting an appropriate observer gain l₁,l₂,l₃The estimated value of the dc bus voltage can be obtained by the extended state observer designed in step 17And estimate of outer loop lumped disturbancesThe estimated value obtained by the extended state observer can be used for the design of the controller, and the specific form is as follows:

wherein the content of the first and second substances,a reference value representing the outer loop of the voltage; u. of_u0Representing the network side controller output; k is a radical of_upRepresenting the controller gain.

Technical Field

The invention relates to the technical field of wind power generation, in particular to a time-varying disturbance compensation-based three-level power generation system model prediction control method.

Background

With the rapid development of the industry, the supply of conventional energy sources is increasingly tense, and the problem of environmental pollution caused by the conventional energy sources is also increasingly tense. The wind energy is a green and environment-friendly renewable energy source, and can effectively relieve the problem of energy supply. At present, the proportion of wind power generation in a power grid is continuously enlarged, so that the research on a wind power generation system has greater practical significance.

The permanent magnet direct-drive wind power generation system is concerned by the advantages of high energy conversion efficiency, high reliability, flexible grid connection and the like. However, the natural wind has the characteristics of randomness, instability and the like, and the permanent magnet synchronous generator has the characteristics of nonlinearity, strong coupling and the like, so that the whole wind power generation system becomes a complex nonlinear system. With the improvement of control requirements, the traditional PID control is difficult to meet the requirements, and scholars at home and abroad put forward a large number of advanced control strategies, such as sliding mode control, active disturbance rejection control, model prediction control and the like.

Compared with PID control, the controller of the finite set model predictive control is designed more flexibly, a cost function can be constructed according to an actual control target, and on the other hand, the finite set model predictive control can directly act on the optimal switching state of the converter according to the constructed cost function output without a modulation link, so that the dynamic response of the system is greatly accelerated; the active disturbance rejection control adopts a two-degree-of-freedom structure, so that the tracking performance and the disturbance rejection performance can be well balanced. External disturbance and uncertainty factors of the system are estimated in real time by using the extended state observer, compensation is performed at the controller end, and the disturbance suppression capability of the system is enhanced while the tracking control performance is met.

In order to improve the dynamic response and the anti-interference performance of the permanent magnet direct-drive wind power generation system, the outer ring adopts active anti-interference control, and the inner ring adopts finite set model prediction control. In the outer-loop active disturbance rejection control, the conventional extended state observer can achieve a good suppression effect on constant-value disturbance and disturbance with slow transformation, and cannot perform estimation well on time-varying disturbance. On the other hand, in order to reduce output harmonic waves, the conventional two-level topology structure can only increase the switching loss by increasing the switching frequency, but an excessively high switching frequency increases the switching loss, and the multi-level topology can enable the switching device to be switched once in each period to achieve the same effect as that of the conventional inverter for several times, and can eliminate more harmonic components under the same switching frequency. Therefore, in order to reduce harmonic components, the converter on the machine side network of the permanent magnet direct-drive wind power generation system adopts a diode-clamped three-level topological structure, and the problems of large switching loss and high harmonic are improved to a certain extent.

Disclosure of Invention

In order to overcome the defects of the prior art, the invention provides a three-level power generation system model prediction control method based on time-varying disturbance compensation. Under a three-level topological structure, the outer ring adopts an improved extended state observer to estimate time-varying disturbance and compensate at a controller end; the inner ring adopts finite set model predictive control, thereby solving the technical problem.

In order to solve the technical problems, the invention provides the following technical scheme:

the time-varying disturbance compensation based three-level power generation system model prediction control method is characterized by comprising the following steps of:

step 1: determining a given speed value omega of a machine-side speed ring_ref；

Step 2: establishing a machine side mathematical model;

and step 3: sampling the machine side current and speed, and converting the current information under the three-phase static coordinate into a d-q coordinate system;

and 4, step 4: establishing a discrete permanent magnet synchronous motor current prediction model, wherein the process is as follows:

4.1: determination of the three-level inverter output voltage vector:

let the three-phase sinusoidal voltage expression be:

defining the inverter output voltage as:

then

And U is also provided_aN+U_bN+U_cNWhen the value is equal to 0, then

The relation between the three-bridge arm switching state of the three-level inverter and the output voltage of the inverter can be obtained:

wherein the content of the first and second substances,

the corresponding space voltage vector is defined as:

wherein the content of the first and second substances,

because each bridge arm corresponds to three switch states, 27 groups of switch states can be obtained, and 27 voltage vectors can be obtained by substituting the 27 groups of switch states into a defined space voltage vector formula;

4.2, determining a permanent magnet synchronous motor current prediction model:

discretizing a current state equation by adopting a forward Euler formula to obtain a discrete permanent magnet synchronous motor current prediction model in the following form:

wherein i_d(k),i_q(k) Representing the stator current component under the two-phase synchronous rotation d-q coordinate system at the current moment; i.e. i_d(k+1)，i_q(k +1) is the stator current d, q-axis component at the next moment; u. of_d,u_qThe d and q axis voltage components correspond to 27 switch states; l is_sThe inductance of a stator under a d-q coordinate system in a surface-mounted permanent magnet synchronous motor is obtained; t is_sIs a sampling period;

step 5, constructing a cost function;

since the current loop on the machine side adopts predictive current control, the cost function J₁Designed in the following form:

wherein the content of the first and second substances,a reference value representing a stator current d, q-axis component; i.e. i_d(k+1),i_q(k +1) are (k +1) T respectively_sAt the moment d, a predicted value of the stator current of the q axis is obtained;

step 6, selecting an optimal voltage vector;

firstly, determining an output voltage vector of a three-level inverter by the switching states of three bridge arms of the three-level inverter; then under the action of a prediction model, a prediction value at the current moment can be obtained; finally, selecting an optimal voltage vector u according to a designed cost function_{opt_1}：

u_{opt_1}＝arg min J₁

Step 7, introducing a state variable d_ωlDetermining a new state space model;

considering the uncertainty of the system parameters and the influence of external disturbance, the mechanical motion equation in step 2 can be organized as follows:

wherein the content of the first and second substances,representing machine side rotating speed ring lumped disturbance; b_ω0Is about b_ωWherein, inA reference value representing a q-axis component of the stator current;

let x₁＝ω，x₂＝d_ωlThen the new state space model is:

wherein h is₁Denotes d_ωlDifferentiation of (1); b_ω0Is about b_ωWherein, inA reference value representing a q-axis component of the stator current;

and 8, expanding the design of the state observer, wherein the process is as follows:

designing an extended state observer according to the new state space model in the step 7, wherein the conventional extended state observer is in the form of:

wherein the content of the first and second substances,an estimate representing ω;representing lumped disturbances d_ωlAn estimated value of (d); beta is a₁,β₂Representing the gain of the extended state observer;

defining error variablesThe form of the error state space model is as follows:

when in useWhen the time is that the machine side outer ring lumped disturbance is a constant value, and the coefficient matrix of the error state space model is a Helvelz matrix, the estimated error asymptotically converges to 0, namely the estimated value asymptotically is tracked without error in an actual state;

if the machine side outer ring lumped disturbance is time-varying disturbance, the extended state observer cannot realize asymptotic error-free tracking, so that improvement needs to be carried out on the basis of the observer to achieve the purpose of realizing the time-varying disturbance;

the modified extended state observer is of the form:

wherein the content of the first and second substances,an estimate representing ω;representing lumped disturbances d_ωlAn estimated value of (d); beta is a₁₁,β₁₂,β₁₃Representing the gain of the modified extended state observer;

defining new error variablesThen

From the new error equation above, we can get:

continued derivation of the equation at both ends can yield:

selecting a state variable:arranging into a state space form:

when in useNamely machine side outer ring lumped disturbance satisfies a₁+a₂When t type time-varying disturbance occurs, and the coefficient matrix of the new error state space model is a Helvelz matrix, the estimation error is asymptotically converged to 0;

and 9, designing a machine side outer ring control law, wherein the process is as follows:

selecting an appropriate observer gain β₁₁,β₁₂,β₁₃The estimated value of the actual rotation speed can be obtained by the extended state observer modified in step 8And estimate of outer loop lumped disturbancesThe estimated value obtained by the extended state observer can be used for the design of the controller, and the specific form is as follows:

wherein the content of the first and second substances,an estimate representing ω; omega_refA reference value representing an outer ring of rotational speeds; u. of_ω0Representing the machine side controller output; k is a radical of_ωpRepresenting the controller gain;

step 10, establishing a direct current link mathematical model;

step 11, establishing a network side mathematical model;

step 12, sampling and coordinate transformation of current and voltage on the network side;

step 13, establishing a discrete inner loop power prediction model;

step 14, constructing a cost function;

step 15, selecting an optimal voltage vector;

step 16, introducing a state variable d_ulDetermining a new state space model;

step 17, designing an extended state observer;

and step 18, designing a network side outer ring control law.

The time-varying disturbance compensation-based three-level power generation system model prediction control method is characterized in that in the step 2, the specific process is as follows:

the mathematical model of the permanent magnet synchronous motor in the d-q coordinate system can be expressed as follows:

the voltage equation is:

in the formula: u. of_d,u_qRepresenting the d-q axis component of the stator voltage; i.e. i_d,i_qRepresenting the d-q axis component of the stator current; l is_sThe stator inductance under a d-q coordinate system in the surface-mounted permanent magnet synchronous motor meets the requirement of L_s＝L_d＝L_q；R_sRepresenting the stator resistance; omega_reRepresents an electrical angular velocity; psi_fRepresents the permanent magnet flux;

the electromagnetic torque equation is:

wherein p is_nRepresenting the number of pole pairs; t is_eRepresents an electromagnetic torque;

the mechanical equation of motion is:

wherein ω represents a mechanical angular velocity; j represents moment of inertia; b represents a friction coefficient; t is_mRepresenting the drive torque.

The time-varying disturbance compensation based three-level power generation system model prediction control method is characterized in that in the step 10, the current at the direct-current side capacitor node P, O, N is represented as:

i_c1＝i_pm-i_pg

i_c1+i_om＝i_c2+i_og

i_c2+i_nm＝i_ng

wherein, C₁,C₂Represents a dc filter capacitance; u. of_c1,u_c2Representing the voltage on the dc bus capacitance; i.e. i_c1,i_c2Representing the current flowing through the dc filter capacitor; i.e. i_pm,i_om,i_nmRepresenting the current, i, flowing through the machine side at node P, O, N_pg,i_og,i_ngIndicating current flow to the net side node P, O, N.

The time-varying disturbance compensation-based three-level power generation system model prediction control method is characterized by comprising the following steps of: in the step 11, the net side mathematical model in the d-q coordinate system is as follows:

wherein u is_d,u_qOutputting components of voltage under d, q coordinate system for the three-level inverter; e.g. of the type_d,e_qThe component of the grid side voltage under a d, q coordinate system is shown; i.e. i_d,i_qThe component of the grid side current in a d, q coordinate system is shown; l represents a network side filter inductor; r represents the equivalent resistance of the output end; omega_geRepresenting the grid angular velocity.

The time-varying disturbance compensation-based three-level power generation system model prediction control method is characterized by comprising the following steps of: in step 13, the grid-side inverter adopts a voltage-oriented control method, so that a grid-side inverter current equation based on grid voltage vector orientation can be expressed as:

wherein u is_d,u_qOutputting components of voltage under d, q coordinate system for the three-level inverter; e.g. of the type_dIs the d-axis component of the net side voltage; i.e. i_d,i_qThe component of the grid side current in a d, q coordinate system; l represents a network side filter inductor; r represents the equivalent resistance of the output end; omega_geRepresenting the grid angular velocity.

The three-level power generation system model prediction control method based on time-varying disturbance compensation is characterized by comprising the following steps: in the step 14, a cost function, cost function J, is constructed₂The form is as follows:

J₂＝|P^*-P(k+1)|+|Q^*-Q(k+1)|

wherein, P^*,Q^*Representing active power and reactive power reference values; p (k +1) and Q (k +1) are (k +1) T_sAnd predicting values of active power and reactive power at the moment.

The time-varying disturbance compensation-based three-level power generation system model prediction control method is characterized by comprising the following steps of: in step 15, the cost function J is selected from the 27 voltage vectors output from the grid-side inverter₂Minimum voltage vector u_{opt_2}：

u_{opt_2}＝arg min J₂

The time-varying disturbance compensation-based three-level power generation system model prediction control method is characterized by comprising the following steps of: in the step 16, a state variable d is introduced_ulConstructing a new state space model;

output power P of machine side rectifier without considering converter loss_mCan be expressed as:

P_m＝u_dci_m

wherein u is_dcRepresenting the dc bus voltage, which may be denoted u_dc＝u_c1+u_c2；i_mRepresenting the current output by the machine side converter to the dc bus;

the current flowing through the dc-side capacitor is:

where C represents a dc-side capacitance, and may be represented as C ═ C₁＝C₂；i_gRepresents the current input to the grid-side inverter;

the active power P input from the dc side to the grid side inverter is:

P＝u_dci_g

from the above equation, one can obtain:

is equivalent to

Wherein the content of the first and second substances,representing net side voltage loop lumped disturbances; b_u0Is about b_uWherein, inP^*Representing an active power reference value;

let z₁＝u_dc；z₂＝d_ulThen the new state space model is:

wherein h is₂Denotes d_ulDifferentiation of (1); b_u0Is about b_uWherein, inP^*Representing the active power reference value.

The time-varying disturbance compensation-based three-level power generation system model prediction control method is characterized by comprising the following steps of: in step 17, the representation of the modified extended state observer of the outer loop is as follows:

wherein the content of the first and second substances,represents u_dcAn estimated value of (d);representing lumped disturbances d_ulAn estimated value of (d); l₁,l₂,l₃Representing the gain of the modified extended state observer;

when in useWhen the time is that the machine side outer ring lumped disturbance is a constant value, and the coefficient matrix of the error state space model is a Helvelz matrix, the estimated error asymptotically converges to 0, namely the estimated value asymptotically is tracked without error in an actual state;

when in useNamely, the network side outer ring lumped disturbance satisfies a₁+a₂And when t-type time-varying disturbance occurs, the designed extended state observer can realize error-free asymptotic convergence.

The time-varying disturbance compensation-based three-level power generation system model prediction control method is characterized by comprising the following steps of: in step 18, the outer loop control law is designed as follows:

selecting an appropriate observer gain l₁,l₂,l₃The estimated value of the dc bus voltage can be obtained by the extended state observer designed in step 17And estimate of outer loop lumped disturbancesThe estimated value obtained by the extended state observer can be used for the design of the controller, and the specific form is as follows:

wherein the content of the first and second substances,a reference value representing the outer loop of the voltage; u. of_u0Representing the network side controller output; k is a radical of_upRepresenting the controller gain.

The invention has the beneficial effects that: an integral link is added to a disturbance estimation item of an outer ring extended state observer of the wind power generation system, time-varying disturbance can be effectively inhibited, a real-time estimated disturbance value can be compensated at a controller end, and the anti-interference performance of the system is improved. On the other hand, the cost function is constructed by utilizing the control target as the controller, the optimal switching state can be directly acted on the converter, the modulation link in vector control is omitted, and the dynamic response speed of the system is greatly accelerated.

Drawings

FIG. 1 is an overall structure diagram of a three-level permanent magnet direct-drive wind power generation system;

FIG. 2 is a three-level inverter space vector diagram;

FIG. 3 is a block diagram of machine side first order active disturbance rejection control based finite set model predictive current control;

FIG. 4 is a block diagram of network side prediction power control based on a finite set model of first-order active disturbance rejection control;

FIG. 5 is a simulation graph of the rotating speed waveform when the wind speed increases at 0.5 s;

FIG. 6 is a simulation of electromagnetic torque waveforms at 0.5s when wind speed increases;

FIG. 7 is a simulation diagram of a stator current component tracking waveform of a q-axis of a current loop at the time of wind speed rise at 0.5 s;

FIG. 8 is a graph of voltage loop voltage waveform simulation at 0.5s with wind speed increasing;

FIG. 9 is a simulation diagram of the net side active power tracking waveform when the wind speed increases at 0.5 s;

FIG. 10 is a simulation graph of the waveform of the grid side phase A output voltage and current when the wind speed increases at 0.5 s;

FIG. 11 is a graph of voltage loop voltage waveform simulation for grid voltage change at 0.7 s;

FIG. 12 is a simulation diagram of the waveform of the output voltage and current of the A phase on the grid side when the grid voltage changes at 0.7 s;

FIG. 13 is a simulation diagram of a rotational speed waveform at the time of a torque change at 0.8 s;

fig. 14 is a graph showing a simulation of the waveform of the grid-side a-phase output voltage and current at the time of a torque change at 0.8 s.

Detailed Description

In order to make the technical solution of the present invention clearer, the following detailed description is made with reference to the accompanying drawings. The embodiments described herein are merely illustrative and are not intended to be limiting.

Referring to fig. 1 to 14, a time-varying disturbance compensation based three-level power generation system model predictive control method includes the following steps:

step 1, determining a given speed value omega of a machine side speed ring_ref；

Step 2, establishing a machine side mathematical model, wherein the process is as follows:

the mathematical model of the permanent magnet synchronous motor in the d-q coordinate system can be expressed as follows:

the voltage equation is:

in the formula: u. of_d,u_qRepresenting the d-q axis component of the stator voltage; i.e. i_d,i_qRepresenting the d-q axis component of the stator current; l is_sThe stator inductance under a d-q coordinate system in the surface-mounted permanent magnet synchronous motor meets the requirement of L_s＝L_d＝L_q；R_sRepresenting the stator resistance; omega_reRepresents an electrical angular velocity; psi_fRepresenting the permanent magnet flux.

The electromagnetic torque equation is:

wherein p is_nRepresenting the number of pole pairs; t is_eRepresenting an electromagnetic torque.

The mechanical equation of motion is:

wherein ω represents a mechanical angular velocity; j represents moment of inertia; (ii) a B represents a friction coefficient; t is_mRepresenting the drive torque.

Step 3, sampling and coordinate transformation of machine side current and speed;

in order to realize effective control of the permanent magnet synchronous motor on the machine side, double closed loop control is adopted, current information of the permanent magnet synchronous motor needs to be known for control of a current loop, and the current information acquired in real time is in a three-phase static coordinate system, so that the current information in the three-phase static coordinate system needs to be converted into a d-q coordinate system for convenient control.

Clark transformation:

park transformation:

wherein, theta_reTo rotate the electrical angle, satisfy

Step 4, establishing a discrete permanent magnet synchronous motor current prediction model, wherein the process is as follows:

and 4.1, determining an output voltage vector of the three-level inverter.

Let the three-phase sinusoidal voltage expression be:

defining the inverter output voltage as:

then

And U is also provided_aN+U_bN+U_cNWhen the value is equal to 0, then

The relation between the three-bridge arm switching state of the three-level inverter and the output voltage of the inverter can be obtained:

wherein the content of the first and second substances,

the corresponding space voltage vector is defined as:

wherein the content of the first and second substances,

because each bridge arm corresponds to three switch states, 27 groups of switch states can be obtained, and 27 voltage vectors can be obtained by substituting the switch states into a defined space voltage vector formula.

And 4.2, determining a permanent magnet synchronous motor current prediction model.

Discretizing a current state equation by adopting a forward Euler formula to obtain a discrete permanent magnet synchronous motor current prediction model in the following form

Wherein i_d(k),i_q(k) Representing the stator current component under the two-phase synchronous rotation d-q coordinate system at the current moment; i.e. i_d(k+1)，i_q(k +1) is the stator current d, q-axis component at the next moment; u. of_d,u_qThe d and q axis voltage components correspond to 27 switch states; l is_sThe inductance of a stator under a d-q coordinate system in a surface-mounted permanent magnet synchronous motor is obtained; r_sRepresenting the stator resistance; psi_fRepresents a permanent magnet flux linkage; omega_re(k) Represents the electrical angular velocity at the present moment; t is_sIs the sampling period.

Step 5, constructing a cost function;

since the current loop on the machine side adopts predictive current control, the cost functionJ₁Designed in the following form:

wherein the content of the first and second substances,a reference value representing a stator current d, q-axis component; i.e. i_d(k+1),i_q(k +1) are (k +1) T respectively_sAnd d, a predicted value of the q-axis stator current.

Step 6, selecting an optimal voltage vector;

firstly, determining an output voltage vector of a three-level inverter by the switching states of three bridge arms of the three-level inverter; then under the action of a prediction model, a prediction value at the current moment can be obtained; finally, selecting an optimal voltage vector u according to a designed cost function_{opt_1}。

u_{opt_1}＝arg min J₁

Step 7, introducing a state variable d_ωlDetermining a new state space model;

considering the uncertainty of the system parameters and the influence of external disturbance, the mechanical motion equation in step 2 can be organized as follows:

wherein the content of the first and second substances,representing machine side rotating speed ring lumped disturbance; b_ω0Is about b_ωWherein, inA reference value representing the q-axis component of the stator current.

Let x₁＝ω，x₂＝d_ωlThen the new state space model is:

wherein h is₁Denotes d_ωlDifferentiation of (1); b_ω0Is about b_ωWherein, inA reference value representing the q-axis component of the stator current.

And 8, expanding the design of the state observer, wherein the process is as follows:

designing an extended state observer according to the new state space model in the step 7, wherein the conventional extended state observer is in the form of:

wherein the content of the first and second substances,an estimate representing ω;representing lumped disturbances d_ωlAn estimated value of (d); beta is a₁,β₂Representing the gain of the extended state observer.

Defining error variablesThe form of the error state space model is as follows:

when in useWhen the lumped disturbance of the outer ring of the time-machine side is a constant value, and the coefficient matrix of the error state space model is HellvinAnd (5) the matrix is used, and the estimated error asymptotically converges to 0, namely the estimated value asymptotically is tracked without error to an actual state.

If the machine-side outer-ring lumped disturbance is time-varying disturbance, the extended state observer cannot realize asymptotic error-free tracking, and therefore improvement needs to be performed on the basis of the observer to achieve the purpose of realizing the time-varying disturbance.

The modified extended state observer is of the form:

wherein the content of the first and second substances,an estimate representing ω;representing lumped disturbances d_ωlAn estimated value of (d); beta is a₁₁,β₁₂,β₁₃Representing the gain of the modified extended state observer.

Defining new error variablesThen

From the new error equation above, we can get:

continued derivation of the equation at both ends can yield:

selecting a state variable:arranging into a state space form:

when in useNamely machine side outer ring lumped disturbance satisfies a₁+a₂And when t type time-varying disturbance occurs, and the coefficient matrix of the new error state space model is a Helvelz matrix, the estimation error is gradually converged to 0.

And 9, designing a machine side outer ring control law, wherein the process is as follows:

selecting an appropriate observer gain β₁₁,β₁₂,β₁₃The estimated value of the actual rotation speed can be obtained by the modified extended state observer in step 8And estimate of outer loop lumped disturbancesThe estimated value obtained by the extended state observer can be used for the design of the controller, and the specific form is as follows:

wherein the content of the first and second substances,an estimate representing ω; omega_refA reference value representing an outer ring of rotational speeds; u. of_ω0Representing the machine side controller output; k is a radical of_ωpRepresenting the controller gain.

Step 10, establishing a direct current link mathematical model

The current at dc-side capacitor node P, O, N is represented as:

i_c1＝i_pm-i_pg

i_c1+i_om＝i_c2+i_og

i_c2+i_nm＝i_ng

wherein, C₁,C₂Represents a dc filter capacitance; u. of_c1,u_c2Representing the voltage on the dc bus capacitance; i.e. i_c1,i_c2Representing the current flowing through the dc filter capacitor; i.e. i_pm,i_om,i_nmRepresenting the current, i, flowing through the machine side at node P, O, N_pg,i_og,i_ngIndicating current flow to the net side node P, O, N,

step 11, establishing a network side mathematical model, wherein the process is as follows:

the net side mathematical model in the d-q coordinate system is:

wherein u is_d,u_qOutputting components of voltage under d, q coordinate system for the three-level inverter; e.g. of the type_d,e_qThe component of the grid side voltage under a d, q coordinate system is shown; i.e. i_d,i_qThe component of the grid side current in a d, q coordinate system is shown; l represents a network side filter inductor; r represents the equivalent resistance of the output end; omega_geRepresenting the grid angular velocity.

Step 12, sampling and coordinate transformation of current and voltage on the network side;

in order to effectively control the grid-side grid-connected inverter and simplify the design of a control system, information acquired in real time under a three-phase static coordinate system needs to be converted into a d-q coordinate system.

Clark transformation:

park transformation:

wherein, theta_geIs the spatial angle of the power grid.

Step 13, establishing a discrete inner loop power prediction model;

firstly, the relation between the switching state of the three-level inverter and the output voltage vector can be obtained from step 4, and then the three-level inverter can be substituted into a prediction model for prediction according to the collected current, voltage information and inverter parameter information.

Grid voltage orientation control is often adopted for controlling a grid-side inverter of a wind power generation system, so that a grid-side inverter current equation based on grid voltage vector orientation can be expressed as follows:

wherein u is_d,u_qOutputting components of voltage under d, q coordinate system for the three-level inverter; e.g. of the type_dIs the d-axis component of the net side voltage; i.e. i_d,i_qThe component of the grid side current in a d, q coordinate system; l represents a network side filter inductor; r represents the equivalent resistance of the output end; omega_geRepresenting the grid angular velocity.

According to the instantaneous power theory and the directional control of the grid voltage, the active power P and the reactive power Q of the grid-side inverter can be expressed as:

wherein: e.g. of the type_d,e_q,i_d,i_qThe components of the grid voltage and current on the d, q axes, respectively.

Since the inner ring on the network side adopts model prediction power control, the power calculation formula needs to be discretized.

At kT_sThe time of day can be:

wherein e is_d(k),e_q(k),i_d(k),i_q(k) The components of the grid voltage and the current on d and q axes at the current moment are respectively; p (k), q (k) represent the active and reactive power at the current moment.

At (k +1) T_sThe time of day can be:

wherein e is_d(k+1),e_q(k+1),i_d(k+1),i_q(k +1) components of predicted values of the grid voltage and the current at the next moment on d and q axes respectively; p (k +1), Q (k +1) respectively represent the active power and reactive power predicted values at the next time.

When sampling time T_sWhen it is sufficiently small, it can be considered that e_d(k+1)＝e_d(k),e_q(k+1)＝e_q(k +1), then

Therefore, it is

Using the forward Euler methodDiscretizing the current state equation of the grid-side inverter can obtain:

the above equation can be compiled from a power prediction model:

wherein u is_d(k),u_q(k) And represents the inverter output voltage component in the d-q coordinate system corresponding to 27 switching states of the three-level inverter.

Step 14, constructing a cost function;

the control targets of the network side are power tracking control and DC side voltage balance, so the cost function J₂Can be designed as follows:

J₂＝|P^*-P(k+1)|+|Q^*-Q(k+1)|

wherein, P^*,Q^*Representing active power and reactive power reference values; p (k +1) and Q (k +1) are (k +1) T_sPredicting values of active power and reactive power at the moment;

step 15, selecting an optimal voltage vector;

the cost function J is selected from the 27 voltage vectors output by the grid-side inverter₂Minimum voltage vector u_{opt_2}。

u_{opt_2}＝arg min J₂

Step 16, introducing a state variable d_ulDetermining a new state space model;

output power P of machine side rectifier without considering converter loss_mCan be expressed as:

P_m＝u_dci_m

wherein u is_dcRepresenting the dc bus voltage, which may be denoted u_dc＝u_c1+u_c2；i_mRepresenting the current output by the machine side inverter to the dc bus.

The current flowing through the dc-side capacitor is:

where C represents a dc-side capacitance, and may be represented as C ═ C₁＝C₂；i_gRepresenting the current input to the grid-side inverter.

The active power P input from the dc side to the grid side inverter is:

P＝u_dci_g

from the above equation, one can obtain:

is equivalent to

Wherein the content of the first and second substances,representing net side voltage loop lumped disturbances; b_u0Is about b_uWherein, inP^*Representing the active power reference value. Let z₁＝u_dc；z₂＝d_ulThen the new state space model is:

wherein h is₂Denotes d_ulDifferentiation of (1); b_u0Is about b_uWherein, inP^*Representing the active power reference value.

Step 17, designing an extended state observer;

designing an extended state observer according to the new state space model in the step 16, wherein the network side extended state observer adopts an extended state observer which is added with an integral link as the machine side, and the specific form is as follows:

the modified extended state observer form is as follows:

wherein the content of the first and second substances,represents u_dcAn estimated value of (d);representing lumped disturbances d_ulAn estimated value of (d); l₁,l₂,l₃Representing the gain of the modified extended state observer.

When in useAnd when the machine side outer ring lumped disturbance is a constant value, the coefficient matrix of the error state space model is a Helvelz matrix, the estimated error asymptotically converges to 0, and the estimated value asymptotically is tracked in an actual state without error.

When in useNamely, the network side outer ring lumped disturbance satisfies a₁+a₂And when t-type time-varying disturbance occurs, the designed extended state observer can realize error-free asymptotic convergence.

Step 18, designing a network side outer ring control law, wherein the process is as follows:

selecting an appropriate observer gain l₁,l₂,l₃The estimated value of the dc bus voltage can be obtained by the extended state observer designed in step 17And estimate of outer loop lumped disturbancesThe estimated value obtained by the extended state observer can be used for the design of the controller, and the specific form is as follows:

wherein the content of the first and second substances,a reference value representing the outer loop of the voltage; u. of_u0Representing the network side controller output; k is a radical of_upRepresenting the controller gain.

Finally, the algorithm is realized in Matlab-simulink software, and the simulation results are shown in FIGS. 5-14.

The wind speed is changed on the premise of ensuring the maximum power tracking, the reference rotating speed of the permanent magnet synchronous motor changes along with the change of the wind speed, as shown in fig. 5, the wind speed is increased at the moment of 0.5s, and the permanent magnet synchronous motor can quickly track the reference rotating speed to reach a new stable state after the wind speed changes; fig. 6 and 7 reflect the tracking conditions of the electromagnetic torque and the q-axis current of the permanent magnet synchronous motor after the wind speed is increased at the time of 0.5s, and it can be found that the reference values of the electromagnetic torque and the q-axis current can be quickly tracked to reach a new steady state; after the wind speed is changed in fig. 8, the voltage of the direct current bus is quickly recovered to a stable state, and the consistency of the voltage of the direct current bus is kept; in fig. 9, the active power at the grid side can well track the active power reference value obtained by the outer loop; in fig. 10, the voltage and current of the power grid still keep the same phase under the condition that the wind speed changes, and full power factor grid connection is realized; fig. 11 and 12 reflect that when the grid voltage changes at 0.7s, the dc bus voltage can be kept stable and full power factor grid connection can still be realized; it can be seen from fig. 13 and 14 that, in steady-state operation, when the torque changes, the rotation speed of the permanent magnet synchronous motor can be quickly recovered to the set value, and at this time, full power factor grid connection can still be realized. The simulation result shows that full power factor grid connection can be realized when wind speed changes, grid voltage changes and torque changes, and when the changes occur, the system can quickly recover a steady state, the changes are well restrained, and the dynamic performance and the anti-interference performance of the system are improved to a certain extent.