CN111812983A - A primary frequency modulation load shedding control method for wind turbines based on differential flat active disturbance rejection control - Google Patents

Abstract

The invention belongs to the technical field of wind power generation, and in particular relates to a primary frequency modulation and load shedding control method for wind turbines based on differential flat active disturbance rejection control. The control performance of the pitch angle controller when the wind turbine participates in the primary frequency modulation and load shedding control is improved. Based on the differential flat ADRC control model, an improved particle swarm optimization algorithm is used to analyze the parameters of the differential flat ADRC controller. Automatic optimization; the invention is widely used in the field of primary frequency modulation and load shedding control of wind turbines.

Description

A primary frequency modulation load shedding control method for wind turbines based on differential flat active disturbance rejection control

technical field

The invention belongs to the technical field of wind power generation, and relates to a primary frequency modulation control method for a doubly-fed wind generator set, which is used to improve the control performance of a pitch angle controller when the wind generator set participates in primary frequency modulation and load shedding control, and in particular relates to a differential flat active disturbance rejection control method. The primary frequency modulation load shedding control method for wind turbines.

Background technique

Traditional fossil energy provides human beings with a large amount of energy, but with the continuous development of the global economy, human beings have an increasing demand for energy, traditional fossil energy is increasingly in short supply and causes serious environmental pollution. The development of clean and renewable energy has become a Inevitably, wind power generation has a broad space for development as a member of clean energy. In recent years, more and more wind turbines have been connected to the grid, and the randomness and fluctuation of their power have brought great challenges to the frequency stability of the grid. In order to improve the grid's ability to absorb wind power, wind turbines are required to have the function of participating in the primary frequency regulation of the grid.

In order to enable the wind turbine to have the ability to participate in the primary frequency regulation of the power grid, it is necessary to carry out load shedding control. The main purpose of the load shedding control is to reduce the power generated by the wind turbine to obtain a certain reserve capacity. The methods can be divided into overspeed control and pitch angle control. The overspeed control is mostly used when the wind speed is low, and its power adjustment range is limited. The pitch angle control has a large power adjustment range, which is suitable for the full wind speed section and is an indispensable part of the load shedding control. The realization method of load shedding only through pitch angle control is shown in Figure 1. First, after knowing the generated power P _tar of the wind turbine during load shedding, the wind turbine is calculated from the power speed (P-ω _r ) curve of the wind turbine. When the generated power is P _tar , the corresponding target rotational speed ω _ref , and then the target pitch angle command β _ref is issued by the pitch angle controller, and then the pitch angle actuator responds to the controller command, which keeps the wind turbine rotational speed. In the vicinity of ω _ref , the generated power of the wind turbine is further stabilized in the vicinity of the target power P _tar .

When the load shedding control is carried out by the pitch angle control, the performance of the pitch angle controller has a great influence on the whole load shedding control process. The pitch process of wind turbines is nonlinear and has external disturbances. It is often difficult to achieve the required control effect only through the traditional PI controller. Therefore, how to improve the control performance of the pitch angle controller has become an urgent problem to be solved. At the same time, the controller parameters have a great influence on the control performance of the controller. Usually, the parameters are manually set and are often not optimal. Therefore, how to optimize the controller parameters has become an urgent problem to be solved.

Differential Flat Active Disturbance Rejection Control (DFADRC) defines internal disturbances (model parameter perturbations) and external disturbances as "total disturbances". The total disturbances are observed in real time and canceled by the extended state observer, which has strong robustness and robustness. applicability, and can be used for nonlinear systems. The controller needs to tune three parameters, and there are certain limitations in tuning the parameters only by experience, and the tuned parameters are often not optimal.

SUMMARY OF THE INVENTION

The invention overcomes the deficiencies of the prior art, provides a method for a wind turbine pitch angle controller based on the differential flat ADRC control, and further proposes to automatically optimize the differential flat ADRR through an improved particle swarm optimization algorithm The method of controller parameters improves the control performance of the pitch angle controller when the wind turbine participates in the primary frequency modulation and load shedding control. The invention proposes a method for optimizing the parameters of the differential flat active disturbance rejection control through an improved particle swarm optimization algorithm, and solves the problem of optimizing the controller parameters.

In order to solve the above-mentioned technical problems, the technical scheme adopted in the present invention is: a primary frequency modulation and load shedding control method for wind turbines based on differential flat active disturbance rejection control, comprising:

The pitch angle controller is designed by using the differential flat active disturbance rejection control strategy;

The pitch process of the wind turbine is set as:

Among them: y is the output (ie rotational speed ω _r ), u is the control signal (ie pitch angle β), a ₁ , a ₀ , b are unknown parameters;

当存在外部扰动时，公式(1)可改写为如下形式：It is assumed that the nominal values of some parameters of the controlled object are known

When there is external disturbance, formula (1) can be rewritten as follows:

为系统总扰动，b₀为b的估计值，η为外部扰动；in:

is the total disturbance of the system, b ₀ is the estimated value of b, and η is the external disturbance;

则公式(2)可写为：selected

Then formula (2) can be written as:

in:

Its extended state observer is:

为了减小参数调节的个数并保证扩张状态观测器的稳定性，通过极点配置法将观测器特征方程的根配置在-ω_o处，即：When the value of L=[l ₁ l ₂ l ₃ ] ^T is appropriate, the estimated quantity can be tracked accurately in real time, that is,

In order to reduce the number of parameter adjustments and ensure the stability of the extended state observer, the root of the observer's characteristic equation is placed at -ω _o by the pole placement method, namely:

λ(s)=|sI-(A-LC)|=(s+ω _o ) ³ (5)

Thus, its parameters are:

where: ω _o is the bandwidth of the extended state observer, and ω _o >0;

可以实时准确跟踪y，

若反馈控制律选为：if

y can be accurately tracked in real time,

If the feedback control law is selected as:

Then the control system can be simplified into the following form:

Given the expected tracking value y ^* of the flat output y, the error is e(t)=y ^* (t)-y(t). Since the controlled object is a second-order differential flat system, the linear feedback control law is:

The closed-loop error characteristic equation is:

p(s)=s ² +δ ₁ s+δ ₀ =0 (9)

In order to ensure the stability of the controller, its eigenvalues are arranged at the left half-plane -ω _c of the s-domain, namely:

p(s)=s ² +δ ₁ s+δ ₀ =s ² +2ζ _c ω _c s+ω _c ² (10)

Then δ ₁ =2ζ _c ω _c , δ ₀ =ω _c ² ; where: ω _c is the controller bandwidth, ζ _c is usually 1;

Through the above analysis, the parameters that need to be tuned for the differential flat ADRC control are the controller bandwidth ω _c , the observer bandwidth ω _o and b ₀ .

On the basis of the above-mentioned differential flat ADRC control model, the parameters of the differential flat ADRC controller are optimized based on the improved particle swarm optimization algorithm. The specific steps are as follows:

Step 1: Initialize parameters, including initial position and speed;

Step 2: Calculate the fitness value, record the individual optimal position pbest and the global optimal position gbest;

Step 3: Update the particle velocity and position, as shown in equations (11) and (12) respectively;

为第k次迭代时第i个个体速度，

以及gbest^k分别为第k次迭代时第i个个体最优位置以及全局最优位置，

为第k次迭代时第i个个体位置；where w is the inertia weight, c ₁ and c ₂ are learning factors, as shown in equations (13), (14) and (15), respectively,

is the i-th individual velocity at the k-th iteration,

and gbest ^k are the i-th individual optimal position and the global optimal position in the k-th iteration, respectively,

is the i-th individual position at the k-th iteration;

where w _max is the initial weight (usually 0.9), w _min is the final weight (usually 0.4), n _c is the current number of iterations, and n _max is the maximum number of iterations;

Step 4: Update the individual best pbest and the global best gbest;

Step 5: Map gbest to [0,1], generate a chaotic sequence by formula (16), and demap the sequence to the original solution space;

z _n+1 = μz _n (1-z _n ), n=0,1,2,… (16)

where μ is the control parameter; set z ₀ ∈ [0,1], the Logistic system is completely in a chaotic state; it has randomness and ergodicity;

Then calculate and compare its fitness value, get the best particle, and randomly replace a particle in the original population;

Step 6: If the end condition is met, the optimization ends; otherwise, go to Step 3.

The fitness function in the improved particle swarm optimization algorithm is specifically:

The integral of time multiplied by the absolute value of error (ITAE), as shown in equation (17):

Among them, T _max is the simulation time, and e(t) is the error between the actual speed and the set value of the speed.

Compared with the prior art, the present invention has the beneficial effects as follows: the present invention uses the improved particle swarm optimization algorithm to optimize the parameters of the differential flat ADDR control, solves the problem of optimizing the controller parameters, and realizes automatic Optimize the parameters of the differential flat ADRC controller. The control performance of the pitch angle controller is improved when the wind turbine participates in the primary frequency modulation and load shedding control.

Description of drawings

The present invention will be further described below in conjunction with the accompanying drawings.

Fig. 1 shows the way in which the existing pitch angle control realizes load shedding.

Fig. 2 is the parameter optimization process of the improved particle swarm optimization algorithm of the present invention to optimize the differential flat ADRC controller.

Detailed ways

As shown in FIG. 2, the present invention is further described.

A primary frequency modulation and load shedding control method for wind turbines based on differential flat active disturbance rejection control, comprising:

The pitch angle controller is designed by using the differential flat active disturbance rejection control strategy;

Usually, the model needs to be simplified when analyzing the pitch process of the wind turbine. The pitch process can be regarded as a second-order process, the input is the pitch angle β of the wind turbine, and the output is the rotor speed ω _r of the wind turbine. Since there is a one-to-one correspondence between the rotational speed and power of the wind turbine when only the pitch control is performed, the power generated by the wind turbine can be adjusted through the pitch angle control according to the actual demand.

The pitch process of the wind turbine is set as:

Among them: y is the output (ie rotational speed ω _r ), u is the control signal (ie pitch angle β), a ₁ , a ₀ , b are unknown parameters.

当存在外部扰动时，公式(1)可改写为如下形式：To take advantage of known parts of the model and reduce the tracking pressure of the expanded state observer. The present invention assumes that the nominal values of some parameters of the controlled object are known

When there is external disturbance, formula (1) can be rewritten as follows:

为系统总扰动(其中包括模型参数的不确定性以及外部扰动)，b₀为b的估计值，η为外部扰动。in:

is the total disturbance of the system (including the uncertainty of model parameters and external disturbances), b ₀ is the estimated value of b, and η is the external disturbance.

则公式(2)可写为：selected

Then formula (2) can be written as:

in:

Its extended state observer is:

为了减小参数调节的个数并保证扩张状态观测器的稳定性，通常通过极点配置法将观测器特征方程的根配置在-ω_o处，即：When the value of L=[l ₁ l ₂ l ₃ ] ^T is appropriate, the estimated quantity can be tracked accurately in real time, that is,

In order to reduce the number of parameter adjustments and ensure the stability of the extended state observer, the root of the observer's characteristic equation is usually placed at -ω _o by the pole placement method, that is:

λ(s)=|sI-(A-LC)|=(s+ω _o ) ³ (5)

Thus, its parameters are:

where: ω _o is the bandwidth of the extended state observer, and ω _o >0.

可以实时准确跟踪y，

若反馈控制律选为：if

y can be accurately tracked in real time,

If the feedback control law is selected as:

Then the control system can be simplified into the following form:

Given the expected tracking value y ^* of the flat output y, the error is e(t)=y ^* (t)-y(t). Since the controlled object is a second-order differential flat system, the linear feedback control law is:

The closed-loop error characteristic equation is:

p(s)=s ² +δ ₁ s+δ ₀ =0 (9)

In order to ensure the stability of the controller, its characteristic root can be configured at the left half-plane -ω _c of the s domain, namely:

p(s)=s ² +δ ₁ s+δ ₀ =s ² +2ζ _c ω _c s+ω _c ² (10)

Then δ ₁ =2ζ _c ω _c , δ ₀ =ω _c ² . Where: ω _c is the controller bandwidth, ζ _c is usually 1.

Through the above theoretical analysis, it can be known that the parameters that need to be set for the differential flat ADRC control are the controller bandwidth ω _c , the observer bandwidth ω _o and b ₀ .

Further, parameter optimization of differentially flat ADRC based on improved particle swarm optimization algorithm;

The present invention improves the standard particle swarm optimization algorithm, and the steps are as follows:

Step 1: Initialize parameters (initial position and speed, etc.);

Step 2: Calculate the fitness value, record the individual optimal position pbest and the global optimal position gbest;

Step 3: Update the particle velocity and position, as shown in equations (11) and (12) respectively;

为第k次迭代时第i个个体速度，

以及gbest^k分别为第k次迭代时第i个个体最优位置以及全局最优位置，

为第k次迭代时第i个个体位置。where w is the inertia weight, c ₁ and c ₂ are learning factors, as shown in equations (13), (14) and (15), respectively,

is the i-th individual velocity at the k-th iteration,

and gbest ^k are the i-th individual optimal position and the global optimal position in the k-th iteration, respectively,

is the i-th individual position at the k-th iteration.

where w _max is the initial weight (usually 0.9), w _min is the final weight (usually 0.4), n _c is the current number of iterations, and n _max is the maximum number of iterations

Step 4: Update the individual best pbest and the global best gbest

Step 5: Map gbest to [0, 1], generate a chaotic sequence by formula (16), and demap the sequence to the original solution space.

z _n+1 = μz _n (1-z _n ), n=0,1,2,… (16)

where μ is the control parameter. Let z ₀ ∈ [0,1], the Logistic system is completely chaotic. It is random and ergodic.

Then calculate and compare its fitness value, get the best particle, and randomly replace a particle in the original population;

Step 6: If the end condition is met, the optimization ends; otherwise, go to Step 3.

The fitness function in the improved particle swarm optimization algorithm involved is:

The fitness function in the optimization algorithm is selected as integral time multiplied by absolute value of error (ITAE), as shown in equation (17):

Among them, T _max is the simulation time, and e(t) is the error between the actual speed and the set value of the speed.

The above-mentioned embodiments merely illustrate the principles and effects of the present invention, but are not intended to limit the present invention. Those skilled in the art can make modifications or improvements to the above embodiments without departing from the spirit and scope of the present invention. Therefore, all equivalent modifications or changes made by those with ordinary knowledge in the technical field without departing from the spirit and technical idea disclosed in the present invention should still be covered by the claims of the present invention.

Claims (3)

1. a wind turbine primary frequency modulation load shedding control method based on differential flat active disturbance rejection control, is characterized in that, comprises: The pitch angle controller is designed by using the differential flat active disturbance rejection control strategy; The pitch process of the wind turbine is set as:

Among them: y is the output (ie rotational speed ω _r ), u is the control signal (ie pitch angle β), a ₁ , a ₀ , b are unknown parameters; It is assumed that the nominal values of some parameters of the controlled object are known

When there is external disturbance, formula (1) can be rewritten as follows:

in:

is the total disturbance of the system, b ₀ is the estimated value of b, and η is the external disturbance; selected

Then formula (2) can be written as:

in:

C=[1 0 0] Its extended state observer is:

When the value of L=[l ₁ l ₂ l ₃ ] ^T is appropriate, the estimated quantity can be tracked accurately in real time, that is,

In order to reduce the number of parameter adjustments and ensure the stability of the extended state observer, the root of the observer's characteristic equation is placed at -ω _o by the pole placement method, namely: λ(s)=|sI-(A-LC)|=(s+ω _o ) ³ (5) Thus, its parameters are:

where: ω _o is the bandwidth of the extended state observer, and ω _o >0; if

y can be accurately tracked in real time,

If the feedback control law is selected as:

Then the control system can be simplified into the following form:

Given the expected tracking value y ^* of the flat output y, the error is e(t)=y ^* (t)-y(t). Since the controlled object is a second-order differential flat system, the linear feedback control law is:

The closed-loop error characteristic equation is: p(s)=s ² +δ ₁ s+δ ₀ =0 (9) In order to ensure the stability of the controller, its eigenvalues are arranged at the left half-plane -ω _c of the s-domain, namely: p(s)=s ² +δ ₁ s+δ ₀ =s ² +2ζ _c ω _c s+ω _c ² (10) Then δ ₁ =2ζ _c ω _c , δ ₀ =ω _c ² ; where: ω _c is the controller bandwidth, ζ _c is usually 1; Through the above analysis, the parameters that need to be tuned for the differential flat ADRC control are the controller bandwidth ω _c , the observer bandwidth ω _o and b ₀ . 2. A kind of primary frequency modulation load shedding control method for wind turbine based on differential flat ADRC control according to claim 1, characterized in that, under the condition based on the above-mentioned differential flat ADRC control model, the method based on improved The particle swarm optimization algorithm optimizes the parameters of the differential flat ADRC controller. The specific steps are as follows: Step 1: Initialize parameters, including initial position and speed; Step 2: Calculate the fitness value, record the individual optimal position pbest and the global optimal position gbest; Step 3: Update the particle velocity and position, as shown in equations (11) and (12) respectively;

where w is the inertia weight, c ₁ and c ₂ are learning factors, as shown in equations (13), (14) and (15), respectively, v _i ^k is the i-th individual velocity in the k-th iteration, pbest _i ^k and gbest ^k are the i-th individual optimal position and the global optimal position at the k-th iteration, respectively, and x _i ^k is the i-th individual position at the k-th iteration;

where w _max is the initial weight (usually 0.9), w _min is the final weight (usually 0.4), n _c is the current number of iterations, and n _max is the maximum number of iterations; Step 4: Update the individual best pbest and the global best gbest; Step 5: Map gbest to [0,1], generate a chaotic sequence by formula (16), and demap the sequence to the original solution space; z _n+1 = μz _n (1-z _n ), n=0,1,2,… (16) where μ is the control parameter; let z ₀ ∈ [0,1], the Logistic system is completely in a chaotic state; it has randomness and ergodicity; Then calculate and compare its fitness value, get the best particle, and randomly replace a particle in the original population; Step 6: If the end condition is met, the optimization ends; otherwise, go to Step 3. 3. a kind of wind turbine primary frequency modulation load shedding control method based on differential flat active disturbance rejection control according to claim 2, is characterized in that, the fitness function in described improved particle swarm optimization algorithm is specifically: The integral of time multiplied by the absolute value of error (ITAE), as shown in equation (17):

Among them, T _max is the simulation time, and e(t) is the error between the actual speed and the set value of the speed.

Images