文档详情

分段线性插值法求插值.doc

发布:2017-05-29约5.29千字共9页下载文档
文本预览下载声明
分段线性插值法求插值 摘要 本文根据题目的要求,利用分段线性插值法对采样点和样本值进行插值计算。为了更好的评断模型的优化性,我们同时采用了最近点插值,3次多项式插值和3次样条插值法来处理同样的问题,作为分段线性插值方法的参考模型。根据插值函数计算区间内任意取样点的函数值。最后再利用所得函数值画出相应的函数图象,并与原函数g(x)的图象进行对比。 通过对本题四个问题的解答,并观察对比函数图象我们得到了如下两个重要 的结论: (1)在同一取样点,利用不同的插值方法可能会得到不同的函数值,所得函数值与原函数的标准函数值的误差大小决定了该插值方法的“好坏”。而最优化的插值方法往往依赖于被插值函数。本题中,在函数式g(x)对应X,Y的条件下,可以根据对比函数图象明显看出:分段线性插值方法和3次多项式插值方法优于3次样条插值和最近点插值。 (2)在插值计算中,取样点的多少往往会影响所得插值函数优化程度。一般情况下,取样点越多所得插值函数越优化,对应的函数值与标准函数值越接近。通过对本题四个问题相应对比函数图象的观察,我们也明显看出:在区间[-6 6]内,当取样点为21,41时,分段线性插值法进行插值计算得到的函数图象基本上与原函数g(x)吻合。 Abstract In this article ,we use piecewise linear interpolation to compute the sampling point and sample value according to the request of question. In order to judge the models quality in a better way, we use nearest interpolation, cubic interpolation and spline interpolation regarded as the model reference of piecewise linear interpolation to deal the question in the same way at the same time. Then draw the function picture by function value of any sampling point in the interval of interpolating function. Finally, we make a comparison between the original function g(x) image and the interpolating function image. At the base of analysing the final result and comparing the constrastive image . We can summarize two items of important conclusion as follows: At the same sampling point , different interpolating method can obtain different function value. Usually , the optimization algorithm depends on the size of error between the object function value . When processing interpolating compute , the number of the sampling point will make an effect on the quality of a model. Commonly, the more multitudinous the sampling points were used ,the more precise the interpolation model will be . 目录 问题的重述……………………………………………… 1 问题的分析……………………………………………… 1 三.问题的假设……………………………………………… 1 四.分段线性插值原理……………………………………… 2 五.问题的求解……………………………………………… 2 六.插值方法的优劣性分析………………………………… 5 附录…………………………………………………………… 6 一.问题的重述 已知,用分段线性插值法求插值,绘出插值结果图形,并观察插值误差。 1.在[-6,6]中平均选取5个点作插值 2.在[-6,6]
显示全部
相似文档