A wind turbine independent pitch control method embedded with linear processing predictive control

Abstract

The present invention discloses an independent variable pitch control method for a wind turbine embedded with linear processing predictive control, which belongs to the field of wind power and includes the following steps: establishing a mathematical model of a large wind turbine; establishing a mathematical model of a linear processor; establishing a mathematical model of a model predictive controller; based on the established model predictive controller, using a pitch angle signal as an input of the model predictive controller to perform independent variable pitch control of the wind turbine. The present invention discretizes the variable pitch signal into multiple signal outputs within a communication cycle through a linear processor, reduces signal frequency and amplitude fluctuations, and combines the superiority of model predictive control in multi-input and multi-output systems, thereby reducing fatigue loss of a drive mechanism and improving blade load reduction effects, and is suitable for systems with high-precision control and parameter changes.

Description

Technical Field

[0001] The invention relates to the field of wind power, and in particular to an independent pitch control method for a wind turbine set embedded with linear processing predictive control. Background Art

[0002] The increase in wind turbine power generation has led to the increasingly large-scale structure of wind turbines, which has resulted in significant impacts from wind shear and tower shadow effects. Research on blade pitch control technology can reduce this impact and ensure the stable operation of wind turbines.

[0003] Blade pitch control technology is divided into two categories: integral pitch and independent pitch. Integral pitch technology transmits the pitch angle signal to the blades, and the three blades perform unified pitch action. Generally, PI or PID controllers are used. The advantages of integral pitch technology are large adjustment range, simple structure, and easy operation, but the blade load is relatively large, especially in large wind turbines. Independent pitch technology transmits the pitch angle signal to the blades, and the three blades adjust the pitch angle independently. The advantage of independent pitch technology is that it can compensate for the impact of uneven wind speed and effectively reduce the blade load, but the single pitch drive and high-frequency output will cause fatigue damage. Therefore, how to find a suitable balance between load reduction and structural fatigue loss is a problem that needs to be solved. Traditional independent pitch is achieved by designing different single-input single-output controllers, or by optimizing parameters and using different control methods, but the load control between blades is not completely decoupled. Therefore, it is necessary to use a multi-input and multi-output controller. Model predictive control is such a control method, which has obvious advantages over traditional control methods.

[0004] At present, the application of model predictive control in the field of wind turbine pitch control can be roughly divided into two types: the first is the basic model predictive controller, and the second is the advanced model predictive controller. The control technology based on the basic model predictive controller has been proven to have good effects in controlling the wind speed of the wind turbine, reducing structural load and fatigue loss, and suppressing power fluctuations, but it is still insufficient to consider the nonlinear factors of wind turbines. Advanced model predictive controllers include variable gain model predictive controllers, continuous linearization model predictive controllers, and nonlinear controllers, which better adapt to the nonlinear factors of wind turbines and have better effects in speed control and load reduction. In terms of blade load reduction, in addition to considering multivariable input and output and nonlinear factors, the processing of pitch signal is also very important, because the high-frequency output and fluctuation of pitch signal will directly affect the load fluctuation of blades and the working strength of driving structure. Therefore, ignoring the processing of pitch signal will inevitably affect the control effect of controller, but most of the existing advanced model predictive controllers do not process pitch signal. Summary of the invention

[0005] In order to solve the above technical problems, the present invention provides an independent variable pitch control method for a wind turbine set embedded with linear processing predictive control, which has a simple algorithm and good control performance.

[0006] The technical solution of the present invention to solve the above technical problem is: a wind turbine independent pitch control method embedded with linear processing predictive control, comprising the following steps:

[0007] Step S1: Establishing a mathematical model of a large wind turbine: Based on the blade azimuth, the nonlinear aerodynamic motion equation of the wind turbine is converted into a linear model in a rotating coordinate system through Kellermann transformation, and then the linear model in the rotating coordinate system is converted into a state space equation in a fixed coordinate system;

[0008] Step S2, establishing a mathematical model of a linear processor: the control of the pitch speed corresponds to the pitch speed trapezoidal curve, and the pitch speed trapezoidal curve in different time periods within the communication cycle is integrated to complete the linearization and discretization processing of the pitch signal to obtain multiple discontinuous pitch angle signals;

[0009] Step S3, establishing a mathematical model of a model predictive controller: on the basis of the state space equation obtained in step S1, establishing a complete model predictive controller;

[0010] Step S4, based on the established model predictive controller, the pitch angle signal obtained in step S2 is used as the input of the model predictive controller to perform independent variable pitch control of the wind turbine set.

[0011] In the above-mentioned independent variable pitch control method for wind turbines embedded with linear processing predictive control, the specific steps of step S1 are as follows:

[0012] Step S11: Establish a linear model of a large wind turbine in a rotating coordinate system:

[0013] ;

[0014] Among them, x, y, and u are state quantity, output quantity, and input quantity respectively. is the blade azimuth, , , are the state quantity, output quantity and input quantity at time t, respectively. , They are , The derivative with respect to time t is is the wind wheel speed, It's about time t The wind wheel speed, , , , is the blade azimuth The correlation model matrix of

[0015] Step S12: Convert the linear model in the rotating coordinate system into a state space equation in the fixed coordinate system.

[0016] In the above-mentioned independent pitch control method for wind turbines embedded with linear processing predictive control, the specific steps of step S12 are as follows:

[0017] Step S121: transform the coordinates of the blade, and the transformation formula is as follows:

[0018] ;

[0019] in, is the coordinate transformation matrix;

[0020] From the above formula, we can get The inverse matrix for:

[0021] ;

[0022] Step S122: convert the state variables, and do not convert other variables. The conversion formula of the state variables is as follows:

[0023] ;

[0024] in, , , represents the state quantity after transformation, , , represents the state quantity before transformation, represents a non-state variable, Represents the diagonal transformation matrix of the state variables; the above formula is abbreviated as , is the state variable after transformation;

[0025] The derivative of the transformed blade coordinates with respect to time is:

[0026] ;

[0027] for The derivative of represents the coefficient matrix;

[0028] Expanding the above formula to state variables yields:

[0029] ;

[0030] for The derivative of is the coefficient matrix after differentiation;

[0031] The transformed state variables The derivative with respect to time t for:

[0032] ;

[0033] Step S123: Repeat step S122, replacing the state variables with input variables and output variables. Similarly, the transformed input variables are The output variable after transformation is ,in is the diagonal transformation matrix of the input variables, is the diagonal transformation matrix of the output variables;

[0034] Step S124: Combining the diagonal transformation matrix of the state variables, output variables and input variables, the state space equation in the fixed coordinate system can be obtained as follows:

[0035] ;

[0036] in , , , After transformation , , , , About time t The transformed input variables, About time t The transformed output variable;

[0037]

[0038] for The inverse matrix of for The inverse matrix of .

[0039] In the above-mentioned independent variable pitch control method for wind turbines embedded with linear processing predictive control, in step S2, according to the variable pitch speed curve, the mathematical model of the linear processor is designed as follows:

[0040] ;

[0041] in, is the output pitch angle, is the initial pitch angle, is the pitch speed, , represent the initial pitch speed, the maximum forward acceleration pitch speed and the maximum reverse acceleration pitch speed respectively, and , t represents time, t i Indicates different time points in the communication cycle, subscript , is the pitch acceleration at different time points; for and The product of represents the changing pitch angle.

[0042] In the above-mentioned independent variable pitch control method for wind turbines embedded with linear processing predictive control, the specific steps of step S3 are as follows:

[0043] Step S31: Establish the model predictive control equation of the model predictive controller as follows:

[0044] ;

[0045] In the formula, is the state at time k, is the state at time k+1, is the input at time k, is the output at time k, Yes After the differential augmentation operation, the corresponding coefficient matrix and the predicted output are written in matrix form to obtain the predicted output matrix Y:

[0046] ;

[0047] In the above formula, the coefficient matrix F, The expression is as follows:

[0048] ;

[0049] ;

[0050] in, represents the control time domain, represents the prediction time domain, and ;

[0051] Step S32: setting a cost function;

[0052] The purpose of the model predictive control equation is to make the cost function obtain the optimal value, that is, to make the error between the actual output of the system and the reference value and the system input obtain the optimal solution; the cost function The expression is as follows:

[0053] ;

[0054] In the formula, the first term Indicates the error between the actual output of the system and the reference value. The second term represents the system input, is the output weight matrix, is the input weight matrix, yes The reference trajectory at time, Indicates that at time k, the prediction Output at the moment, Indicates that at time k, the prediction Input of time, subscript , Indicates count;

[0055] Step S33: Set the input constraint conditions, the expression is as follows:

[0056] ;

[0057] In the formula, Indicates the system status. represents the initial state of the system, Indicates the upper and lower limits of the input. Indicates the upper and lower limits of the input change rate.

[0058] The beneficial effects of the present invention are:

[0059] 1. The present invention proposes a model predictive controller combined with a linear processor based on the characteristics of high-frequency output and large amplitude variation of pitch signal during independent pitch control of large wind turbines. The linear processor discretizes the pitch signal into multiple signal outputs within one communication cycle, thereby reducing the frequency change and amplitude fluctuation of the signal, thereby reducing the fatigue loss of the drive mechanism and improving the load reduction effect of the blades. The controller is suitable for systems with high-precision control and parameter changes.

[0060] 2. The present invention adapts well to the nonlinear factors of wind turbines, further improves the effect of the model predictive control method in load shedding, thereby enhancing the application scope and practicality of the technology, and has good adaptability to fluctuations in external loads, and can still have a good load shedding effect under different turbulent wind conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 It is the overall flow chart of the present invention.

[0062] Figure 2 Schematic diagram of a linear processor.

[0063] Figure 3 Schematic diagram of the model predictive controller.

[0064] Figure 4 This is a comparison chart of experimental results of the present invention, the PI controller, and a separate MPC controller under the condition of an average wind speed of 14 m / s.

[0065] Figure 5 This is a comparison chart of experimental results of the present invention, the PI controller, and a separate MPC controller under the condition of an average wind speed of 18 m / s. DETAILED DESCRIPTION

[0066] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0067] like Figure 1 As shown, a wind turbine independent pitch control method embedded with linear processing predictive control comprises the following steps:

[0068] Step S1: Establish a mathematical model of a large wind turbine: Based on the blade azimuth, the nonlinear aerodynamic motion equation of the wind turbine is converted into a linear model in a rotating coordinate system through the Kellermann transformation, and then the linear model in the rotating coordinate system is converted into a state space equation in a fixed coordinate system.

[0069] The specific steps of step S1 are:

[0070] Step S11: Establish a linear model of a large wind turbine in a rotating coordinate system:

[0071] ;

[0072] Among them, x, y, and u are state quantity, output quantity, and input quantity respectively. is the blade azimuth, , , are the state quantity, output quantity and input quantity at time t, respectively. , They are , The derivative with respect to time t is is the wind wheel speed, It's about time t The wind wheel speed, , , , is the blade azimuth The correlation model matrix.

[0073] Step S12: In order to adapt the model predictive controller, the linear model in the rotating coordinate system is converted into a state space equation in the fixed coordinate system.

[0074] The specific steps of step S12 are:

[0075] Step S121: transform the coordinates of the blade, and the transformation formula is as follows:

[0076] ;

[0077] in, is the coordinate transformation matrix;

[0078] From the above formula, we can get The inverse matrix for:

[0079] ;

[0080] Step S122: convert the state variables, and do not convert other variables. The conversion formula of the state variables is as follows:

[0081] ;

[0082] in, , , represents the state quantity after transformation, , , represents the state quantity before transformation, represents a non-state variable, Represents the diagonal transformation matrix of the state variables; the above formula is abbreviated as , is the state variable after transformation;

[0083] The derivative of the transformed blade coordinates with respect to time can be obtained:

[0084] ;

[0085] for The derivative of represents the coefficient matrix;

[0086] Expanding the above formula to state variables yields:

[0087] ;

[0088] for The derivative of is the coefficient matrix after differentiation;

[0089] The transformed state variables Derivatives with respect to time for:

[0090] ;

[0091] Step S123: Repeat step S122, replacing the state variables with input variables and output variables. Similarly, the transformed input variables are The output variable after transformation is ,in is the diagonal transformation matrix of the input variables, is the diagonal transformation matrix of the output variables;

[0092] Step S124: Combining the diagonal transformation matrix of the state variables, output variables and input variables, the state space equation in the fixed coordinate system can be obtained as follows:

[0093] ;

[0094] in , , , After transformation , , , , About time t The transformed input variables, About time t The transformed output variable;

[0095]

[0096] for The inverse matrix of for The inverse matrix of .

[0097] Step S2, establish a mathematical model of the linear processor: the control of the pitch speed corresponds to the pitch speed trapezoidal curve, and the pitch speed curve in different time periods within the communication cycle is integrated to complete the linearization and discretization processing of the pitch signal to obtain multiple discontinuous pitch angle signals.

[0098] like Figure 2 As shown in the figure, according to the pitch speed curve, the mathematical model of the linear processor is designed as follows:

[0099] ;

[0100] in, is the output pitch angle, is the initial pitch angle, is the pitch speed, , represent the initial pitch speed, the maximum forward acceleration pitch speed and the maximum reverse acceleration pitch speed respectively, and , t represents time, t i Indicates different time points in the communication cycle, subscript , is the pitch acceleration at different time points; for and The product of represents the changing pitch angle. Through the above formula, the linearization of the set pitch angle can be completed. For example, in order to ensure the stability of the power generation, the variable pitch speed requirement is , the communication cycle is , then the pitch control command received by the pitch control controller is:

[0101] ;

[0102] In the formula is the pitch angle, is the number of times it is sent. Therefore, the pitch control commands received by the pitch controller are , so as to achieve the purpose of reducing the frequency and amplitude fluctuation of the pitch signal.

[0103] In turbulent wind conditions, wind speed changes frequently, the blade pitch angle will frequently switch between maximum acceleration and deceleration, and the blade load will fluctuate greatly. The processed pitch signal can reduce the severity of the change, making the power output more stable.

[0104] Step S3, establish the mathematical model of the model predictive controller: Figure 3 As shown, a complete model predictive controller is established based on the state space equation obtained in step S1.

[0105] The specific steps of step S3 are:

[0106] Step S31: Establish the model predictive control equation of the model predictive controller as follows:

[0107] ;

[0108] In the formula, is the state at time k, is the state at time k+1, is the input at time k, is the output at time k, Yes After the differential augmentation operation, the corresponding coefficient matrix and the predicted output are written in matrix form to obtain the predicted output matrix Y:

[0109] ;

[0110] In the above formula, the coefficient matrix F, The expression is as follows:

[0111] ;

[0112] ;

[0113] in, represents the control time domain, represents the prediction time domain, and .

[0114] Step S32: setting a cost function;

[0115] The purpose of the model predictive control equation is to make the cost function obtain the optimal value, that is, to make the error between the actual output of the system and the reference value and the system input obtain the optimal solution; the cost function The expression is as follows:

[0116] ;

[0117] In the formula, the first term Indicates the error between the actual output of the system and the reference value. The second term represents the system input, is the output weight matrix, is the input weight matrix, yes The reference trajectory at time, Indicates that at time k, the prediction Output at the moment, Indicates that at time k, the prediction Input of time, subscript , Indicates count.

[0118] Step S33: Set the input constraint conditions, the expression is as follows:

[0119] ;

[0120] In the formula, Indicates the system status. represents the initial state of the system, Indicates that at time k, the prediction Input at the moment, Indicates the upper and lower limits of the input. Indicates the upper and lower limits of the input change rate.

[0121] Step S4, based on the established model predictive controller, the pitch angle signal obtained in step S2 is used as the input of the model predictive controller to perform independent variable pitch control of the wind turbine set.

[0122] Simulation experiment

[0123] All experiments were conducted on the same computer with the following configuration: Intel(R) Core(TM) i7-8700 CPU @ 3.20GHz, RAM 16 GB and NVIDIA GeForce GTX 1050 GPU. In order to verify the effect of the controller, simulation experiments at different wind speeds were carried out in MATLAB / Simulink. Two groups of experiments were designed, namely, a wind rotor diameter of 126m, a hub diameter of 3m, a hub height of 90m, a wind rotor mass of 110t, a cut-in wind speed of 4m / s, an average wind speed of 14m / s, and a wind rotor diameter of 126m, a hub diameter of 3m, a hub height of 90m, a wind rotor mass of 110t, a cut-in wind speed of 4m / s, and an average wind speed of 18m / s. In order to verify the load reduction effect, comparative experimental results of the method of the present invention with a PI controller and a separate MPC controller are given. The experimental results are as follows: Figure 4 , Figure 5 As shown. Figure 4 , Figure 5 It can be seen that the method of the present invention can achieve good load reduction effect under different wind conditions with large frequency of pitch signal amplitude changes, and is superior to PI control and separate MPC control methods. The peak value and overall fluctuation of the experimental results obtained by this method in the figure are lower than those of other methods.

[0124] In summary, the present invention can reduce the amplitude change of the pitch signal, reduce the signal frequency, and obtain a good blade load reduction effect under different wind conditions, thereby improving the problem that the model predictive controller has limited load reduction effect under different wind conditions. The discrete output idea of ​​the linear processor is combined with the model predictive control to introduce independent pitch control, which reduces the fatigue load of the drive structure and the blade load, and improves the service life of the unit. It can be applied to multi-input and multi-output, coupled and nonlinear systems, and broadens the scope of application.

Claims

1. A wind turbine independent pitch control method embedded with linear processing predictive control, characterized in that: The following steps are involved: Step S1: Establishing a mathematical model of a large wind turbine: Based on the blade azimuth, the nonlinear aerodynamic motion equation of the wind turbine is converted into a linear model in a rotating coordinate system through Kellermann transformation, and then the linear model in the rotating coordinate system is converted into a state space equation in a fixed coordinate system; Step S2, establishing a mathematical model of a linear processor: the control of the pitch speed corresponds to the pitch speed trapezoidal curve, and the pitch speed trapezoidal curve in different time periods within the communication cycle is integrated to complete the linearization and discretization processing of the pitch signal to obtain multiple discontinuous pitch angle signals; Step S3, establishing a mathematical model of a model predictive controller: on the basis of the state space equation obtained in step S1, establishing a complete model predictive controller; Step S4, based on the established model predictive controller, the pitch angle signal obtained in step S2 is used as the input of the model predictive controller to perform independent variable pitch control of the wind turbine set.

2. The independent pitch control method for wind turbines embedded with linear processing predictive control according to claim 1 is characterized in that: The specific steps of step S1 are: Step S11: Establish a linear model of a large wind turbine in a rotating coordinate system: ； Among them, x, y, and u are state quantity, output quantity, and input quantity respectively. is the blade azimuth, , , are the state quantity, output quantity and input quantity at time t, respectively. , They are , The derivative with respect to time t is is the wind wheel speed, It's about time t The wind wheel speed, , , , is the blade azimuth The correlation model matrix of Step S12: Convert the linear model in the rotating coordinate system into a state space equation in the fixed coordinate system.

3. The independent pitch control method for wind turbines embedded with linear processing predictive control according to claim 2 is characterized in that: The specific steps of step S12 are: Step S121: transform the coordinates of the blade, and the transformation formula is as follows: ； in, is the coordinate transformation matrix; From the above formula, we can get The inverse matrix for: ； Step S122: convert the state variables, and do not convert other variables. The conversion formula of the state variables is as follows: ； in, , , represents the state quantity after transformation, , , represents the state quantity before transformation, represents a non-state variable, Represents the diagonal transformation matrix of the state variables; the above formula is abbreviated as , is the state variable after transformation; The derivative of the transformed blade coordinates with respect to time is: ； for The derivative of represents the coefficient matrix; Expanding the above formula to state variables yields: ； for The derivative of is the coefficient matrix after differentiation; The transformed state variables The derivative with respect to time t for: ； Step S123: Repeat step S122, replacing the state variables with input variables and output variables. Similarly, the transformed input variables are The output variable after transformation is ,in is the diagonal transformation matrix of the input variables, is the diagonal transformation matrix of the output variables; Step S124: Combining the diagonal transformation matrix of the state variables, output variables and input variables, the state space equation in the fixed coordinate system can be obtained as follows: ； in , , , After transformation , , , , About time t The transformed input variables, About time t The transformed output variable; for The inverse matrix of for The inverse matrix of .

4. The independent pitch control method for wind turbines embedded with linear processing predictive control according to claim 1 is characterized in that: In step S2, according to the pitch speed curve, the mathematical model of the linear processor is designed as follows: ； in, is the output pitch angle, is the initial pitch angle, is the pitch speed, , represent the initial pitch speed, the maximum forward acceleration pitch speed and the maximum reverse acceleration pitch speed respectively, and , t represents time, t i Indicates different time points in the communication cycle, subscript , is the pitch acceleration at different time points; for and The product of represents the changing pitch angle.

5. The independent pitch control method for wind turbines embedded with linear processing predictive control according to claim 3 is characterized in that: The specific steps of step S3 are: Step S31: Establish the model predictive control equation of the model predictive controller as follows: ； In the formula, is the state at time k, is the state at time k+1, is the input at time k, is the output at time k, Yes After the differential augmentation operation, the corresponding coefficient matrix and the predicted output are written in matrix form to obtain the predicted output matrix Y: ； In the above formula, the coefficient matrix F, The expression is as follows: ； ； in, represents the control time domain, represents the prediction time domain, and ; Step S32: setting a cost function; The purpose of the model predictive control equation is to make the cost function obtain the optimal value, that is, to make the error between the actual output of the system and the reference value and the system input obtain the optimal solution; the cost function The expression is as follows: ； In the formula, the first term Indicates the error between the actual output of the system and the reference value. The second term represents the system input, is the output weight matrix, is the input weight matrix, yes The reference trajectory at time, Indicates that at time k, the prediction The output of the moment, Indicates that at time k, the prediction Input of time, subscript , Indicates count; Step S33: Set the input constraint conditions, the expression is as follows: ； In the formula, Indicates the system status. represents the initial state of the system, Indicates the upper and lower limits of the input. Indicates the upper and lower limits of the input change rate.

Images