TL;DR
This paper introduces a novel gain-scheduled H∞ control strategy for wind turbines to achieve active power control (APC). By coordinating blade pitch and generator torque, the controller can precisely track power commands and minimize structural loads across the entire operational envelope, addressing the limitations of traditional turbine control.
What Problem Does This Paper Address?
Traditional wind turbines are designed to maximize power capture at low wind speeds and maintain rated power at high speeds. However, they lack the ability to actively control their power output to respond to grid demands, such as frequency regulation or ancillary services. This limitation becomes critical as wind power penetration increases, potentially destabilizing the grid. While some methods exist to reserve power, they often operate along specific curves within the power envelope. This paper aims to provide a unified control strategy that allows wind turbines to operate anywhere within their power envelope, actively tracking power commands while simultaneously considering structural load mitigation.
Methods
The core of the proposed solution is a gain-scheduled H∞ controller. Here's a breakdown of the methodology:
- Multiple Input, Multiple Output (MIMO) Control: The controller coordinates two key actuators: blade pitch angle and generator torque. This is a departure from traditional single-input, single-output (SISO) control strategies.
- H∞ Control Synthesis: For specific operating points (trim conditions) within the turbine's power envelope, H∞ control is used. This robust control technique is chosen to handle uncertainties and ensure good performance in the presence of disturbances.
- Gain Scheduling: To address the nonlinear dynamics of the wind turbine across its entire operating range, the controller is "gain-scheduled." This involves:
- Linearization: The nonlinear turbine model is linearized at multiple "trim points" across the wind speed and power output envelope.
- Controller Design: An H∞ controller is designed for each of these trim points.
- Interpolation: For any operating condition between these trim points, the controller gains are calculated by linearly interpolating the gains from the nearest designed trim points. This simplifies the design process compared to more complex linear parameter-varying (LPV) methods.
- Decoupled vs. Coupled Control: The paper initially considers a decoupled control architecture (separate controllers for pitch and torque) but opts for a coupled MIMO H∞ approach due to the inherent coupling between power and rotor speed dynamics, which complicates tuning in a decoupled system.
Datasets
Not discussed in the paper. The evaluation relies on simulations using the NREL FAST simulation package, which models a WindPACT 1.5 MW wind turbine.
Key Insights and Ideas
- Unified Control for the Entire Envelope: The gain-scheduling approach allows for a single controller structure to operate effectively across the entire power/wind speed envelope, simplifying implementation and enabling consistent performance.
- Balancing Power Tracking and Load Mitigation: The H∞ framework, with carefully chosen performance weights, allows for the simultaneous optimization of tracking a desired power reference and minimizing structural loads.
- Trade-offs in Control Strategy: The paper highlights that achieving APC can involve trade-offs. For instance, shedding power by increasing blade pitch (as proposed for low wind speeds) can increase pitch actuator usage.
- Importance of Gain Scheduling for Nonlinearity: Simulations demonstrate that a single H∞ controller designed at one operating point performs well near that point but shows discrepancies when deviating significantly, underscoring the necessity of gain scheduling for handling turbine nonlinearities.
Key Results
- Improved Power Tracking: Simulations show that the gain-scheduled controller achieves good power tracking performance, with a 90% settling time of less than 10 seconds for step commands.
- Reduced Structural Loads: Compared to a standard control law, the gain-scheduled APC shows significant reductions (over 20%) in damage equivalent loads (DEL) for blade root flapwise bending and tower fore-aft bending under rated wind conditions.
- Rotor Speed Regulation: The controller demonstrates improved rotor speed regulation, with a substantial decrease in the RMS rotor speed tracking error.
- Above-Rated Performance: In above-rated wind conditions, the gain-scheduled controller also shows good performance, with better rotor speed regulation than a traditional PI blade pitch controller, though transients after large power command steps can still induce higher shaft loads.
- Actuator Usage: The APC strategy, particularly in below-rated conditions, utilizes blade pitch actuation more than the standard control, leading to higher RMS and maximum pitch rates.
Limitations and Future Work
- Performance Guarantees: The paper acknowledges that the linear interpolation used in gain scheduling does not offer formal performance guarantees, unlike more rigorous LPV methods.
- Actuator Saturation: The current design does not explicitly account for actuator saturation, which could limit performance in extreme conditions.
- Turbulent Wind Conditions: While simulations were performed in turbulent wind, the paper suggests that further consideration is needed for transient performance in these conditions.
- Future Work Suggestions: The authors propose several avenues for future research:
- Investigating the maximum power tracking bandwidth.
- Developing additional outer-loop controls for different ancillary services.
- Establishing upper bounds on command power in turbulent wind conditions.
- Incorporating actuator saturation into the design.
- Exploring more rigorous LPV designs for potentially improved performance guarantees.
Summary updated