https://orcid.org/0000-0003-4911-0481
![Sample our Engineering & Technology journals, sign in here to start your access, latest two full volumes FREE to you for 14 days]({"type":"image" , "src" : "/sda/5933/TOC-Subject-Sample-2021-Engineering-Tech.jpg"})
ABSTRACT
Digital hydraulics is a potential technology for the Hydraulic Pitch System (HPS) in Wind Turbine (WT). Digital Hydraulics Pitch System (DHPS) uses Digital Flow Control Units (DFCU) to develop the precise pitching action. In this paper, a novel Intelligent Digital Pitch Controller (IDPC) is proposed. The proposed controller is designed and implemented on a developed lab-scale DHPS Hardware-in-the-Loop (HIL) simulator. The various parameters of DHPS-hardware were designed using the bottom-up design methodology. The IDPC comprises Machine Learning (ML)-based WECS and DHPS controllers in the outer and inner loop respectively. HIL simulations were conducted with the implemented IDPC. The ML-based WECS controller predicts the reference pitch angle close to its desired value. The ML-based DHPS controller predicts the states of DFCU to develop real-time pitching action in DHPS-hardware. Several case studies were conducted to validate the effectiveness of the proposed IDPC. A study shows that IDPC controlled DHPS exhibits better performance than an ML-Proportional Integral (PI) controlled HPS with proportional flow control valve. Subsequently, the performance of the IDPC is compared with PI-ML cascade controller. This study shows that the Maximum Absolute Error (MAE) between the generator speed and its rated speed is 0.87% and 19.29% for the proposed controller and PI-ML cascade controller, respectively. Similarly, MAE (error between generator torque and its rated torque) of torque is 0.85% and 5.46% for the proposed controller and PI-ML cascade controller, respectively. Thus, the implementation of the IDPC develops optimal power with minimal speed/torque fluctuations.
Acknowledgments
Authors wish to show our gratitude to Prof. Raghunathan Rengasamy, Institute chair professor, Robert Bosch Centre for Data Science and Artificial Intelligence, Indian Institute of Technology (IIT)-Madras for his great support and help. The authors would like to thank Aeolus, for the supporting toolbox in Simulink and National Renewable Energy Laboratory (NREL).
Abbreviations
| WT | = | Wind Turbine |
| GWEC | = | Global Wind Energy Council |
| GW | = | Gigawatt |
| MW | = | Megawatt |
| HIL | = | Hardware-in-the-Loop |
| HPS | = | Hydraulic Pitch System |
| MPPT | = | Maximum Power Point Tracking |
| PFCV | = | Proportional Flow Control Valve |
| DHPS | = | Digital Hydraulic Pitch System |
| WECS | = | Wind Energy Conversion System |
| MPa | = | Mega Pascal |
| EB | = | Electromagnetic Brake |
| FS | = | Full Scale |
| DH | = | Digital Hydraulic |
| DFCU | = | Digital Flow Control Unit |
| PNM | = | Pulse Number Modulation |
| PCM | = | Pulse Code Modulation |
| DFP | = | Digital Fluid Power |
| PID | = | Proportional Integral Derivative |
| P | = | Proportional |
| PI | = | Proportional Integral |
| FLC | = | Fuzzy Logic Controller |
| ML | = | Machine Learning |
| IDPC | = | Intelligent Digital Pitch Controller |
| FFBP-NN | = | Feed Forward Back Propagation-Neural Network |
| FDUHP | = | Fixed Displacement Unidirectional Hydraulic Pump |
| HM | = | Hydraulic Motor |
| ADE | = | Angular Displacement Encoder |
| MAE | = | Maximum Absolute Error |
| MSE | = | Mean Square Error |
| NREL | = | National Renewable Energy Laboratory |
| LM | = | Levenberg–Marquardt |
| SD | = | Steepest Descent |
| GN | = | Gauss-Newton |
| ODE23t | = | Ordinary Differential Equation (Equation23 |
Notations
| Paero | = | Aerodynamic power W |
| ρair | = | Density of air kg/m3 |
| ν | = | Wind speed m/s |
| Arot | = | Swept area of the rotor m2 |
| Cp | = | Coefficient of power – |
| β | = | Pitch angle deg |
| λ | = | Tip speed ratio – |
| R | = | Radius of the rotor m |
| ωrot | = | Rotor angular speed rad/s |
| Taero | = | Aerodynamic torque Nm |
| CT | = | Coefficient of torque – |
| Jrot | = | Inertia of the rotor kg-m2 |
| Krot | = | External damping coefficient of the rotor – |
| Tbt | = | Low-speed shaft torque Nm |
| ωls | = | Speed of the low-speed shaft rad/s |
| ϕrot | = | Angular deviations of rotor-side deg |
| ϕls | = | Angular deviations of gearbox-side deg |
| Kls | = | Low-speed shaft damping Ns/m |
| Fls | = | Low-speed shaft stiffness N/m |
| Ths | = | Torque at high-speed shaft Nm |
| Jgen | = | Generator inertia kg-m2 |
| Tgen | = | Generator torque Nm |
| ωgen | = | Generator speed rad/s |
| Kgen | = | Generator external damping Ns/m |
| ig | = | Gearbox gear ratio – |
| v*cut-in | = | Cut-in wind speed m/s |
| v*rated | = | Rated wind speed m/s |
| v*cut-out | = | cut-out wind speed m/s |
| Cp-max | = | Maximum power coefficient – |
| λop | = | Optimal tip speed ratio – |
| βop | = | Optimal pitch angle deg |
| tgen | = | Generator time constant s |
| Tref | = | Reference generator torque Nm |
| Pgen | = | Generated power W |
| Pgen-rated | = | Rated generator power W |
| ωgen-rated | = | Rated speed of the generator rad/s |
| Ipl | = | Total blade inertia moment kg-m2 |
| Iy1 and Ix1 | = | Moment of inertia of the airfoil section about the axis o-y1 and o-x1 kg-m2 |
| ρb | = | Blade material density kg/m3 |
| ∆r | = | Blade incremental radius m |
| t | = | Thickness of the airfoil m |
| l | = | Chord length of the airfoil m |
| Lp | = | Actual dynamic pitch load Nm |
| Lp-max | = | Maximum dynamic pitch load Nm |
| η | = | Pitch bearing efficiency – |
| ipg | = | Pitch gear ratio – |
| Ib | = | Blade mass moment along the longitudinal axis kg-m2 |
| Im | = | Hydraulic motor mass moment of inertia kg-m2 |
| ωsmax | = | Maximum value of pitch rate deg/s |
| tpmax | = | Time to reach the maximum value by the pitch rate s |
| Qhm | = | Hydraulic motor flow rate m3/s |
| Vhm | = | Hydraulic motor volumetric displacement m3/rad |
| Ps | = | Supply pressure Pa |
| ωm-max | = | Maximum value of shaft speed at the hydraulic motor rad/s |
| Tmax | = | Maximum torque at the shaft of the hydraulic motor Nm |
| Qp | = | Flow rate of the hydraulic pump m3/s |
| Vp | = | Hydraulic pump volumetric displacement m3/rad |
| ηpv | = | Volumetric efficiency of the hydraulic pump – |
| ωem | = | Angular speed of the electric motor rad/s |
| Qo | = | Minimum flow rate of an orifice m3/s |
| Cd | = | Discharge coefficient – |
| N | = | Number of valves in DFCU – |
| Qv1 | = | Flow rate of orifice at valve 1 in DFCU m3/s |
| Qv2 | = | Flow rate of orifice at valve 2 in DFCU m3/s |
| Qv3 | = | Flow rate of orifice at valve 3 in DFCU m3/s |
| Qv4 | = | Flow rate of orifice at valve 4 in DFCU m3/s |
| Qv5 | = | Flow rate of orifice at valve 5 in DFCU m3/s |
| Av1 | = | Area of orifice at valve 1 in DFCU m2 |
| Av2 | = | Area of orifice at valve 2 in DFCU m2 |
| Av3 | = | Area of orifice at valve 3 in DFCU m2 |
| Av4 | = | Area of orifice at valve 4 in DFCU m2 |
| Av5 | = | Area of orifice at valve 5 in DFCU m2 |
| ρf | = | Density of the hydraulic fluid kg/m3 |
| ∆P | = | Pressure difference across the hydraulic orifice Pa |
| Sgen | = | Generates states – |
| Sp | = | Predicted states – |
| βref | = | Predicted reference pitch angle deg |
| ψ | = | Tracking errors – |
| βD | = | Desired pitch deg |
| ψDE | = | Desired pitch error deg |
| βg | = | Generated pitch deg |
| ωe | = | Generator speed error rad/s |
| wb | = | Weight of the neurons – |
| n | = | Incoming connections – |
| yo | = | Output – |
| fb | = | Bias – |
| T(x) | = | Tangent sigmoidal function – |
| a*u | = | Control function – |
| µ | = | Combination coefficient – |
| Γ | = | Error function – |
| α | = | Learning constant – |
| I | = | Identity matrix – |
| Jk | = | Jacobian matrix – |
![Supercharge Your Next Research Paper: Research and write your next paper with Jenni AI.]({"type":"image" , "src" : "/sda/112445/Skyscraper-econ.png"})