基于向量值有理插值的最优预测算法研究及陶瓷需求量预测的应用

On Study of Optimum Predict Algorithm Based on Vector Value Rational Interplation and its Application on Predict of Ceramic Demand

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.

In this paper, A optimum predict algorithm based on vector value rational interplation is got through rational vector interplation functibn and error bounds of each component,through defect analysis of modeling and forecasting methods to overall expression of high-demensional data, It has overcome occurrence of the forecast deviation, because of overall expression method of vectors in other algorithms; In addition,the optimum predict value of predict samples vector are determined through vector value rational interplation function of trainning samlpes, error bounds of each component and iteration method.the examples show that the precision of the predicted values obtained by the algorithm is better than that by the other algorithms,and that error bound and iteration step to the effect on performance of algorithm is analyzed.