Numerical empirical study on structural approximation analysis of neural network for blade of wind turbine

Structural approximation analysis is important for the design optimization of wind turbine blades. First, the method of structural approximation analysis of the neural network for the wind turbine blade is introduced, and the neural network for the blade is briefed. The empirical and relevant practical studies on the structural approximation analysis of the neural network is introduced, and the influence of the patterns of the wind turbine blade's performance on the analysis is investigated. The numerical experiments of different learning rates to construct the model of the neural network for the structural approximation analysis are made. According to the experimental results, the number of patterns of the blade should be enough to describe the relationship between the performance and the blade's parameters. It is concluded that the accuracy of the structural approximation analysis of the neural network is higher with a greater number of patterns of the blades. Based on the experiments, the large learning rate is helpful for obtaining a better model of the neural network. The empirical study is helpful for reducing the expensive cost of the design optimization of the wind turbine blade using the structural approximation analysis of the neural network. ©, 2015, Nanjing University of Aeronautics an Astronautics. All right reserved.