Design of a multiblock general regression neural network for wind speed prediction in Algeria

Abstract

In this work, we investigate a new design of a multiblock general regression neural network applied to wind speed prediction in Algeria. The idea in our proposed method is to minimize the error of the prediction for wind speed in such a way as to minimize the quantity of training samples used, and thus to reduce the costs related to the training sample collection. For this reason, we propose to select the most significant sample among a large number of training samples by using multiblock general regression neural network (MBGRNN). This paper presents experimental results on six different real wind speed measurement stations in Algeria namely, Alger, Djelfa, Bechar, Oran, Sétif and In Aménas. The wind speed data covers a period of ten years between 2001 and 2010.