Customer Demand Prediction for Wind Turbines Based on Bayesian Regularized BP Neural Network

To meet the demands of wind turbine customers, a neural network model was established to predict customer demands. Considering the shortcomings of the backpropagation algorithm of the BP neural network, such as slow convergence speed, long training time, and easy falling into local optimum, the BP neural network was optimized by the Bayesian regularization (BR) algorithm, the Levenberg Marquardt (LM) algorithm, and the Scaled Conjugate Gradient (SCG) algorithm, respectively. An index system of influencing factors of customer demands was established. Through the comparative analysis of the prediction results of the wind turbine customer demands of a certain wind turbine manufacturer, it was found that the prediction effect of the BR algorithm was better, effectively compensating for the large deviation of the product demand prediction for the second half of the year only from the business perspective.