Wind Turbine MPC Controller Using Single Linear Model
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current control methods for wind turbines, such as Linear Quadratic Regulators (LQR) and Model Predictive Control (MPC), face challenges like high computational overhead and instability due to slow dynamics and non-linear aerodynamic behavior, making them impractical for real-world application, especially in zone III where wind gusts occur.
Innovation Solution
A single internal linear model-based MPC method is proposed, which uses a 3-mass model to describe the dynamic behavior of the wind turbine's rotor, drive train, and generator, with aerodynamic torque as a disturbance and generator torque as a manipulated variable, allowing for reduced computational load and improved structural load management.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Reliability
If LQR controller is used to reduce torsional vibrations, then vibration reduction is achieved, but computational overhead becomes excessively high due to slow turbine dynamics requiring very long prediction horizons
Solution Approach 1:
The patent changes the fundamental parameter of prediction horizon length. Instead of using very long prediction horizons required by LQR for slow turbine dynamics, the MPC controller uses a finite, optimized prediction horizon that is sufficiently long to capture relevant dynamics but short enough to reduce computational overhead to acceptable levels for real-time control.
2Productivity
If LQR controller is used without considering variable restrictions, then optimal control path is computed, but system becomes unstable when manipulated variable saturation occurs
Solution Approach 1:
The MPC controller incorporates feedback mechanisms that continuously monitor the system state and adjusted the control inputs based on actual measurements. This feedback loop allows the controller to detect when manipulated variables approach saturation limits and adjust the control strategy accordingly, preventing instability while maintaining optimization efficiency.
Solution Approach 2:
The patent implements a dynamic control strategy where the prediction horizon and control inputs are continuously updated based on current system conditions. This dynamic approach allows the controller to adapt to changing operating conditions and variable restrictions in real-time, maintaining stability while achieving optimal control performance.
3Measurement precision
If multiple linear models are used for different operating points, then control accuracy is improved, but device complexity and computational load increase significantly
Solution Approach 1:
The patent implements a universal MPC controller that can handle multiple operating conditions using a single unified framework. Instead of maintaining separate linear models for different operating points, the controller uses a single predictive model that adapts to various operating conditions through optimization, reducing complexity while maintaining control accuracy across the full operating range.
Solution Approach 2:
The patent changes the approach from using multiple fixed models to using a single adaptive model where parameters such as prediction horizon and control weights are adjusted based on operating conditions. This parameter adaptation allows the controller to maintain accuracy across different operating points without the complexity of multiple internal models.
4Reliability
If non-linear aerodynamic behavior is fully modeled, then control performance in zone III is improved, but computational overhead increases making real-time implementation impractical
Solution Approach 1:
The patent changes the modeling approach by using a simplified linear or reduced-order model combined with MPC's optimization capabilities. Instead of implementing full non-linear aerodynamic models that are computationally intensive, the controller uses a simpler model with adjusted parameters and relies on the MPC optimization framework to achieve the necessary control performance in zone III while keeping computational overhead manageable for real-time implementation.
Data Source
AI summary
Model-based predictive control method (MPC) for the reduction of structural load in wind turbines comprising: exclusively proposing a single internal linear model for the MPC for the entire operating range of the turbine; obtaining the adjustable parameters of the linear internal model from the experimental data previously measured in the turbine; choosing the discrete time values for the control and prediction horizons; adjusting the MPC controller and performing a practical implementation test.