基于Reflected Sigmoid径向基函数插值的温度场重建算法
Temperature field reconstruction algorithm based on Reflected Sigmoid radial basis function interpolation
作者：董晨龙(四川大学电子信息学院)；周新志(四川大学电子信息学院)；白兴都(四川大学电子信息学院)；张若彬(四川大学电子信息学院)
Author：dongchenlong(College of Electronics and Information Engineering，Sichuan University)；zhouxinzhi(College of Electronics and Information Engineering，Sichuan University)；baixingdu(Sichuan University)；zhangruobin(Sichuan University)
收稿日期：2018-08-22 年卷（期）页码：2019,56(5):0851-0856
期刊名称：四川大学学报: 自然科学版
Journal Name：Journal of Sichuan University (Natural Science Edition)
关键字：声学测温；最小二乘法；温度矩阵；Reflected Sigmoid径向基函数；温度场重建
Key words：Acoustic pyrometry；Least square method；Temperature matrix；Reflected Sigmoid radial basis function；Temperature field reconstruction
基金项目：国家973计划项目（2013CB328903-2）
中文摘要
声学法测温在特殊的温度场环境中有良好的应用,主要是利用有限的超声波传播路径上的飞行时间重构出连续分布的温度场.现有的温度场重建算法中最小二乘法是最常用的方法,但其重建后的温度场会出现边缘信息缺失的现象.针对这一问题,提出在最小二乘法确定温度矩阵的基础上,结合Reflected Sigmoid函数进行插值,实现了二维平面温度场的无缺失重建.通过两种典型的单峰温度场模型的重建结果及误差分析表明,在补全温度场边缘的条件下,单峰对称温度场的均方根百分误差在1.6%,单峰偏斜温度场的均方根百分误差在3.5%,取得了很好的重建效果.
英文摘要
The acoustic thermometry has a good application in the special temperature field environment. That is, the continuous distribution of temperature field can be reconstructed using the flight time spent on the finite ultrasonic propagation path. The least square method is one of the acoustic temperature field reconstruction algorithms, and the reconstructed temperature field will have the phenomenon of missing edge information using this method. In order to solve this problem, an algorithm based on Reflected Sigmoid radial basis function interpolation is proposed to reconstruct the two dimensional distribution of temperature without information loss. In this algorithm, the least square method is adopted to determine the temperature matrix and the Reflected Sigmoid function is used to interpolate the temperature field. The reconstruction results and error analysis of two typical single peak temperature field models show that the root mean square percentage error of the symmetrical single peak temperature field is 1.6% and the root mean square percentage error of the skewed single peak temperature field is 3.5% after complementing the edge of the temperature field.
【关闭】
