论文部分内容阅读
通过对高维数据整体表达式建模预测方法和分区间等预测算法的缺陷分析,提出基于向量值有理插值的最优预测算法,通过有理向量插值函数和各分量的误差限得到向量之间的相似性,克服了其它很多算法利用向量的整体表达式方法而产生预测的偏差;另外,通过向量的误差限与训练样本所得向量值有理插值函数及迭代仿真方法来确定预测样本向量所对应的最优预测值.通过实例,算法所得预测值的精度比其他算法更高,并且分析了误差限和迭代步长对算法性能的影响.
Based on the analysis of the defects of the whole-body modeling and forecasting method of high dimensional data and the inter-partition prediction algorithm, an optimal prediction algorithm based on vector valued rational interpolation is proposed. Through the vector interpolation function and the error limits of each component, Similarity, and overcome many other algorithms using the vector of the overall expression of the method to generate the bias of the prediction; In addition, through vector error limits and training samples obtained vector interpolation rational function and iterative simulation method to determine the best prediction vector corresponding to the sample Through the example, the accuracy of the prediction value obtained by the algorithm is higher than other algorithms, and the influence of error limit and iteration step size on the performance of the algorithm is analyzed.