【摘要】 针对基于欧氏距离的混沌局域预测算法在高嵌入维情况下预测精度迅速下降的问题,提出了基于夹角余弦的局域加权线性预测算法。该算法使用夹角余弦取代欧氏距离作为判别相点间相关性的标准;同时将相点间相关性大小通过加权的方式作用于预测模型;并将相点视为向量,以向量的模和夹角为优化目标,进行预测参数识别。在详细论述改进算法理论基础的同时,使用南方某城市电力负荷进行了算例验证。算例结果表明,改进算法的预测精度明显优于原算法,验证了改进算法的有效性。更多还原

【Abstract】 The forecasting precision of chaotic local forecasting algorithm based on Euclid distance decreases rapidly when the embedding dimension is high.Aiming at above problem,we proposed a local adding-weight linear forecasting algorithm based on included angle cosine.The algorithm uses the included angle cosine as criterion to judge correlation between phase points instead of Euclid distance;at the same time,and the values of correlation are acted on chaotic series forecasting model by means of adding-weight.Moreover,regarding phase points as vectors,we identified forecasting parameters through optimization module and angle of vectors.While introducing the theoretical foundation of the improved algorithm at length,the uses power load of a southern city is taken for a forecasting example.The forecasting results indicate the precision of the algorithm is obviously higher than the original one,demonstrating effectiveness of improved algorithm.更多还原