Aerodynamic measurement on horizontal axis wind turbines in the field is a challenging research topic and also an essential research method on the aerodynamic performance of blades in real atmospheric inflow conditions. However, the angle of attack is difficult to determine in the field due to the unsteadiness and unevenness of the inflow. To study the measuring and analyzing method of angle of attack in the field, a platform was developed based on a 100 kW wind turbine from the Institute of Engineering Thermophysics (IET) in China in this paper. Seven-hole probes were developed and installed at the leading edge to measure the inflow direction, static and total pressure at the near field. Two data reducing processes, sideslip angle correction, and induced velocity correction, were proposed to determine the effective angle of attack based on the inflow data measured by probes. The aerodynamic measurement platform was first validated by the comparison with wind tunnel results. Then some particular aerodynamic phenomenon in the field were discussed. As a result, the angle of attack varies quasi-periodically with the rotation of the rotor, which is caused by the yaw angle of the inflow. The variation of angle of attack induces dynamic response of a clockwise hysteresis loop in lift coefficient. The dynamic response is the main source of a dispersion of instantaneous lift coefficients with a standard deviation of 0.2.
Machine learning method has become a popular, convenient and efficient computing tool applied to many industries at present. Multi-hole pressure probe is an important technique widely used in flow vector measurement. It is a new attempt to integrate machine learning method into multi-hole probe measurement. In this work, six typical supervised learning methods in scikit-learn library are selected for parameter adjustment at first. Based on the optimal parameters, a comprehensive evaluation is conducted from four aspects: prediction accuracy, prediction efficiency, feature sensitivity and robustness on the failure of some hole port. As results, random forests and K-nearest neighbors’ algorithms have the better comprehensive prediction performance. Compared with the in-house traditional algorithm, the machine learning algorithms have the great advantages in the computational efficiency and the convenience of writing code. Multi-layer perceptron and support vector machines are the most time-consuming algorithms among the six algorithms. The prediction accuracy of all the algorithms is very sensitive to the features. Using the features based on the physical knowledge can obtain a high accuracy predicted results. Finally, KNN algorithm is successfully applied to field measurements on the angle of attack of a wind turbine blades. These findings provided a new reference for the application of machine learning method in multi-hole probe calibration and measurement.
Operation and maintenance (O&M) costs can account for 10%–20% of the total costs of electricity for a wind project. Structural health monitoring method is a form of preventive maintenance consisting of regular monitoring of the wind turbine components to detect potential faults. The strain measurements on the blades are typically used for load monitoring and damage detection, but only obtain a small amount of concentrated load information. In this work, a strain distribution monitoring method was proposed and validated based on aerodynamic measurements in the field. Then the contribution and evolution of the strain distribution from the aerodynamic load, gravity load, inertial load, and centrifugal load in the complex start-up process of the wind turbine were analyzed. The results shows that the mean value of the synthetic strain is mainly determined by the aerodynamic load and the centrifugal load, while the fluctuation amplitude is mainly determined by the gravity and the aerodynamic load. Furthermore, the fatigue damage of the blade root was evaluated based on strain extrapolation, rainflow cycle-counting algorithm, Goodman diagram and Miner’s linear superposition principle. It is found that the fatigue damage is generally greatest near the pressure side and suction side, and least near the leading edge and trailing edge. The strain distribution monitoring method can capture the location of maximum stress and the most severe fatigue damage on the blade, which are helpful for damage detection and load control, and further reduces O&M costs.