Chaotic Local Weighted Linear Prediction Algorithms Based on the Angle Cosine

Abstract This paper expounds the limitations of the Euclidean distance as the measure between points similarity. According to the limitations of the original algorithm presented, chaotic local weighted linear forecast algorithm based on the angle cosine is proposed, which replaces Euclidean distance by cosine in the measurement of the similarity between phase points. In the process of parameters identification in the linear fitting, replace the Euclidean distance by the module and angle of vector as the optimal object. This algorithm overcomes the disadvantages of chaotic local prediction algorithm based on the Euclidean distance, and has obtained good effect in power load forecasting which is sensitive to the climate.