为提高分形插值的准确性，提出分段分形插值，建立多个迭代函数系。为验证分段分形插值准确性，采用全域随机湍流风模拟程序TurbSim对SMOOTH和Kaimal两种湍流风谱模型进行风场仿真，并随机选取点作为初始插值点，对初始点进行分段分形插值。结果表明：分段分形插值能有效提高垂直比例因子选取的准确性，相同时间段内，分段分形插值平均标准误差低于传统非分段分形插值，30 s时，分段分形差值的平均标准误差为0.910，而当时间段为50 s时，平均标准误差仅为0.695；分段节点作为初始插值点被使用了两次，间接地增加了初始点个数，提高了分形插值精度；分段分形插值降低了生成点个数，便于分析数据局部波动信息。

In order to improve the accuracy of fractal interpolation,a fractal interpolation is proposed,and several iterative functions are established.In order to verify the accuracy of segmental fractal interpolation,the turbulence model of SMOOTH and Kaimal are simulated by turbulim simulating program,and the initial points are taken as the initial interpolation points,and the initial points are segmented and interpolated.The results show that the fractal interpolation can effectively improve the accuracy of the vertical scale factor selection,and the average standard error of subsection fractal interpolation is lower than that of traditional non-subsection fractal interpolation in the same time period.At 30 s,the average standard error of subsection fractal difference is 0.910,while with the time period of 50,the average standard error is only 0.695.The segmentation node is used twice as the initial interpolation point,which indirectly increases the number of initial points and improves the fractal interpolation precision.The fractal interpolation reduces the number of generated points,and it is easy to analyze the local fluctuation information of the data.

TK83

国家自然科学基金（51676131，51176129，51875361）