Model-free self-adaptive independent variable pitch control method and system for wind turbine generator based on multivariable decoupling

Abstract

The invention discloses a model-free self-adaptive independent variable pitch control method and system for a wind turbine generator based on multivariable decoupling, and aims to solve the problem of blade fatigue damage caused by blade root asymmetric load under a high-wind-speed turbulence working condition. A multi-input multi-output control architecture is designed, control parameters are dynamically adjusted by using a model-free adaptive control (MFAC) algorithm, a cooperative adjustment instruction of an independent variable pitch angle is generated, optimal control of a three-blade pitch angle is realized, and power output stability is maintained while a blade root load is suppressed. Compared with traditional independent pitch control (IPC), the method has the advantages that model dependence is eliminated through pseudo-gradient matrix multivariable decoupling, the anti-interference capacity under the turbulent wind condition is remarkably enhanced, and technical guarantee is provided for safe operation of a large wind turbine generator.

Description

Technical Field

[0001] The present invention belongs to the technical field of intelligent control of wind turbines, and specifically relates to a model-free adaptive independent pitch control method and system based on multivariable decoupling. Aiming at the problems of blade root periodic load fluctuation and power instability of large wind turbines under high wind speed turbulence conditions, a multi-input multi-output (MIMO) model-free adaptive control (MFAC) algorithm is used to achieve coordinated optimization of load balancing and power stability. Background Art

[0002] With the popularity of large wind turbines above 5MW, the continuous increase in rotor diameter has led to an intensification of the wind speed gradient effect on the rotor surface. The asymmetric loads generated by the three blades under the tower shadow effect, wind shear and turbulence excitation show significant time-varying characteristics, which seriously threatens the structural safety. Traditional independent pitch control (IPC) relies on an aerodynamic-structural coupling model, the blade root load of the unit is high, and the power output stability needs to be further improved. Summary of the invention

[0003] Purpose of the invention: In view of the shortcomings of the prior art, the present invention proposes a model-free adaptive independent pitch control method and system for wind turbines based on multivariable decoupling, aiming to solve the problems of model dependence and multivariable coupling, reduce the blade root load of the unit, improve the power output stability, and provide technical support for the safe and efficient operation of large wind turbines under complex working conditions.

[0004] Technical solution: A model-free adaptive independent pitch control method for a wind turbine based on multivariable decoupling is provided, comprising the following steps:

[0005] S1: Data collection and feedback

[0006] Real-time wind speed collection , Wind turbine azimuth , generator speed The blade root flapping moment of the three blades of the wind turbine , , , which is fed back to the controller through the sensor module of the perception layer;

[0007] S2: Unified pitch reference angle generation

[0008] Generate a uniform pitch reference angle based on the generator speed , whose expression is:

[0009]

[0010] Where: is the pitch angle at rated wind speed; is the rated speed of the generator; is the empirical coefficient; is the number of discrete time steps;

[0011] S3: Kalman Transform

[0012] Convert blade root moment to pitch moment in hub coordinate system and yaw moment :

[0013]

[0014] Where: For the The azimuth angle of the blades ( =1, 2, 3).

[0015] S4: Model-free adaptive control (MFAC) algorithm

[0016] A model-free adaptive control model is established based on the compact dynamic linearization (CFDL) method, and the pseudo gradient matrix is ​​dynamically updated. The pitch angle adjustment in the pitch and yaw directions is calculated by the control law and ; Specifically include:

[0017] S41: Dynamic Linearization Modeling

[0018] Define the system output as y ( k ) = [ M t ( k ), M y ( k )] T , the control input is u ( k ) = [ Δ β t ( k ), Δ β y ( k )] T , establish a linear time-varying model:

[0019]

[0020] Where: , ; [ ] T Represents the transpose of a matrix; is the pseudo gradient matrix;

[0021] S42: Control Law Design

[0022] The objective function is:

[0023]

[0024] Where: is the expected torque; is the weight coefficient; Representation Matrix The norm of ;

[0025] S43: Solve the control input update law

[0026]

[0027] Where: is the step size factor;

[0028] S44: Pseudo-gradient matrix online update

[0029] Update the pseudo gradient matrix using the projection algorithm , constraining the update amplitude of non-diagonal elements to be ≤ 0.1.

[0030] S5: Inverse Kalman Transform

[0031] Generate independent pitch adjustment for each blade:

[0032]

[0033] S6: Independent pitch execution

[0034] Correct the final pitch angle of each blade:

[0035]

[0036] Beneficial effects: The present invention achieves significant effects in terms of load suppression and power stability of wind turbines through a model-free adaptive independent pitch control method and system for wind turbines based on multivariable decoupling. The specific effects are as follows:

[0037] (1) Load suppression: standard deviation of blade root bending moment is reduced by ≥30%;

[0038] (2) Power stability: fluctuation rate ≤ 2%. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] Figure 1 It is the overall architecture diagram of the control system;

[0040] Figure 2 It is the flow chart of MFAC algorithm;

[0041] Figure 3 It is a comparison chart of load suppression effect;

[0042] Figure 4 This is an output power comparison chart. DETAILED DESCRIPTION

[0043] In order to explain the technical solution of the present invention in detail, the following is further described in conjunction with the drawings and specific embodiments of the specification. It should be understood that the described embodiments are only used to explain the present invention and are part of the embodiments of the present invention, rather than all of the embodiments.

[0044] See also Figure 1The present invention implements control for wind turbines, and the control system includes a perception layer, a control layer and an execution layer; the perception layer includes a fiber optic strain sensor (sampling rate 1kHz), a lidar anemometer (sampling rate 20Hz), an azimuth encoder (accuracy ±0.1°) and a speed sensor (accuracy ±0.1rpm); the control layer includes a ROSCO main controller, an MFAC coprocessor (calculation cycle 10ms) and a pitch angle correction module; the execution layer includes a hydraulic pitch actuator (position control accuracy ±0.1°, response time ≤50ms). The implementation of the control method includes 6 steps: data acquisition and feedback, unified pitch reference angle generation, Kalman transform, model-free adaptive control (MFAC) algorithm, Kalman inverse transform and independent pitch execution. Real-time collection of wind speed , wind turbine azimuth , generator speed The blade root flapping moment of the three blades of the wind turbine , , , which is fed back to the controller through the sensor module of the perception layer; a unified pitch reference angle is generated according to the generator speed ; Through Kalman transformation, the blade root moment is converted into the pitch moment in the hub coordinate system and yaw moment ; A model-free adaptive control model is established based on the compact dynamic linearization (CFDL) method, and the pseudo gradient matrix is ​​dynamically updated. The pitch angle adjustment in the pitch and yaw directions is calculated by the control law and ; Generate independent pitch adjustment values ​​for each blade through inverse Kalman transform; Correct the final pitch angle of each blade, and drive the blades to perform pitch operation through the hydraulic pitch system; Optimize the control of the three-blade pitch angle to maintain power output stability while suppressing blade root load.

[0045] See also Figure 2 ,The model-free adaptive control (MFAC) algorithm in the present invention specifically includes 5 steps: initialization, dynamic linearization, control rate calculation, pseudo-gradient update and update iteration.

[0046] Example 1

[0047] In order to verify the effect of the control method and system on load suppression and power stability, according to Figure 1 The control system architecture and Figure 2 The model-free adaptive control (MFAC) algorithm flow is described, and a simulation model of a wind turbine model-free adaptive independent pitch control system is built on the OPENFAST and Matlab / simulink software platforms for simulation analysis.

[0048] Among them, wind turbine parameters: NREL 5MW onshore wind turbine, rated power 5MW, rotor radius 63m, hub height 90m, rated wind speed 11.4m / s. Blade parameters: blade length 61.5m, pre-bent design, blade root flapping stiffness. Sensor configuration: The sampling rate of the optical fiber strain sensor is 1kHz, which is used to measure the blade root flapping torque; the sampling rate of the lidar anemometer is 20Hz, which is used to measure the wind speed 50m in front of the hub; the accuracy of the azimuth encoder is ±0.1°, which is used to measure the blade azimuth; the accuracy of the speed sensor is ±0.1rpm, which is used to measure the generator speed. Controller hardware configuration: The main controller is based on the ROSCO (Reference Open Source Controller) architecture, integrating the benchmark pitch control module to achieve a smooth transition from area 2 (maximum power tracking) to area 3 (constant power output); the MFAC coprocessor uses the TI TMS320C6748 DSP model (main frequency 200MHz), which runs the MFAC algorithm in real time and calculates the cycle =10 ms, memory is 512kB RAM, 2MB Flash. Model-free adaptive control (MFAC) algorithm parameter settings: Error weight =0.3, step factor =0.2, sampling period T=10ms. Pseudo gradient matrix The initialization value is Φ ( 0 ) = [ − 60 10 1 − 1 ] , learning rate =0.1, prevent odd and abnormal numbers =10 -5 , empirical coefficient =0.15, pitch angle at rated wind speed =2∘, rated speed of the generator =12.1rpm. Actuator configuration: The position control accuracy of the hydraulic pitch system is ±0.1°, the response time is ≤50ms, and the oil pressure is set to 20MPa. Wind condition setting: IEC Class B wind condition with turbulence intensity of 15%.

[0049] Load suppression effects such as Figure 3 The standard deviations of the blade root flapping moments of the three blades are 903.74 kNm (1342.71 kNm for traditional IPC), 912.83 kNm (1351.91 kNm for traditional IPC), and 913.06 kNm (1339.44 kNm for traditional IPC); the maximum blade root flapping moments of the three blades are 6591 kNm (7465 kNm for traditional IPC), 7072 kNm (7459 kNm for traditional IPC), and 6941 kNm (7396 kNm for traditional IPC).

[0050] Output power Figure 4 The maximum output power is 5075kW (the traditional IPC is 5138kW), the standard deviation of the output power is 45.59kW (the traditional IPC is 48.44kW), and the output power fluctuation range is ±1.28% (the traditional IPC is ±2.76%).

[0051] Result analysis:

[0052] (1) Load suppression effect: The standard deviation of the blade root bending moment is reduced by 32.69%, 32.48%, and 31.83%, that is, the standard deviation of the blade root bending moment is reduced by ≥30%.

[0053] (2) Power stabilization effect: The output power fluctuation rate is ±1.28%≤2%, which meets the requirement of grid fluctuation rate ≤5%. Compared with traditional IPC, MFAC further reduces it by 53.6%; the output power standard deviation is reduced by 5.9%.

[0054] The above results show that the model-free adaptive independent pitch control method and system for wind turbines based on multivariable decoupling described in the present invention can significantly reduce the blade root load of the unit, improve the power output stability, and provide technical support for the safe and efficient operation of large wind turbines under complex working conditions.

Claims

1. A model-free adaptive independent pitch control method and system for wind turbines based on multivariable decoupling, characterized in that: The following steps are involved: S1: Data collection and feedback Real-time wind speed collection , Wind turbine azimuth , generator speed The blade root flapping moment of the three blades of the wind turbine , , , which is fed back to the controller through the sensor module of the perception layer; S2: Unified pitch reference angle generation Generate a uniform pitch reference angle based on the generator speed , whose expression is: Where: is the pitch angle at rated wind speed; is the rated speed of the generator; is the empirical coefficient; is the number of discrete time steps; S3: Kalman Transform Convert blade root moment to pitch moment in hub coordinate system and yaw moment : Where: For the The azimuth angle of the blades ( =1, 2, 3). S4: Model-free adaptive control (MFAC) algorithm A model-free adaptive control model is established based on the compact dynamic linearization (CFDL) method, and the pseudo gradient matrix is ​​dynamically updated. The pitch angle adjustment in the pitch and yaw directions is calculated by the control law and ; S5: Inverse Kalman Transform Generate independent pitch adjustment for each blade: S6: Independent pitch execution Correct the final pitch angle of each blade:

2. The method and system according to claim 1, characterized in that: The model-free adaptive control (MFAC) algorithm in step S4 includes: S41: Dynamic Linearization Modeling Define the system output as , the control input is , establish a linear time-varying model: Where: , ; Represents the transpose of a matrix; is the pseudo gradient matrix; S42: Control Law Design The objective function is: Where: is the expected torque; is the weight coefficient; Representation Matrix The norm of ; S43: Solve the control input update law Where: is the step size factor; S44: Pseudo-gradient matrix online update Update the pseudo gradient matrix using the projection algorithm , constraining the update amplitude of non-diagonal elements to be ≤ 0.

1.

3. The method and system according to claim 1, characterized in that: The control system includes a perception layer, a control layer and an execution layer; the perception layer includes a fiber optic strain sensor (sampling rate 1kHz), a lidar anemometer (sampling rate 20Hz), an azimuth encoder (accuracy ±0.1°) and a speed sensor (accuracy ±0.1rpm); the control layer includes a ROSCO main controller, an MFAC coprocessor (calculation cycle 10ms) and a pitch angle correction module; the execution layer includes a hydraulic variable pitch actuator (position control accuracy ±0.1°, response time ≤50ms).

4. The method and system according to claim 1, characterized in that: According to the method, under IEC Class B wind conditions with a turbulence intensity of 15%, the standard deviation of the blade root bending moment is reduced by ≥30% and the power fluctuation rate is ≤2%.

Images