多光谱温度测量反演算法研究  

Research on Inversion Algorithm of Multi-spectral Temperature Measurement

作　　者：刘睿辰 孙鹏[1] 王高[1] 王彬翰 杜秀昊 LIU Ruichen;SUN Peng;WANG Gao;WANG Binhan;DU Xiuhao(State Key Laboratory of Dynamic Measurement Technology Jointly Built by the Province and Ministry,North University of China,Taiyuan 030051,China)

机构地区：[1]中北大学省部共建动态测试技术国家重点实验室,太原030051

出　　处：《计算机测量与控制》2025年第1期293-300,共8页Computer Measurement &Control

基　　金：国家自然科学基金项目(62105305);山西省基础研究计划项目(20210302123068,202203021212141);山西省高等学校科技创新项目(2022L004)。

摘　　要：多光谱测温方法相较于传统单光谱测温方法而言,其具有非接触测量,响应速度快,鲁棒性好等优点,被广泛地用于科学研究,工业生产等领域;然而,光谱发射率的求解是多光谱测温方法中无法避免的问题;通过引入一种参考温度模型可将光谱发射率的求解转换为带约束的极小值优化问题;对于此问题,采用了两种启发式算法和3种其他混合算法对其进行求解,经实验测试实现了对算法性能的检验和评估;结果表明,遗传算法总体精度较高,粒子群算法速度较快,3种混合算法在一定程度上平衡了两种算法的性能优势;实际测量数据的反演结果表明,5种算法的平均相对误差均小于1.17%,满足了实际测量需求;此外,通过对比分析了不同算法在不同模型下性能,为不同测试场景和优化目标下的算法的选择提供了依据。Compared with traditional single spectral temperature measurement methods,multi-spectral temperature measurement methods have the characteristics of non-contact measurement,fast response time,high accuracy,and strong robustness,which is widely used in scientific research,industrial production and other fields.However,it is an inevitable issue to solve spectral emissivity in multi-spectral temperature measurement methods.By introducing a reference temperature model,the solution of spectral emissivity can be converted into a constrained minimum optimization problem.For this problem,two heuristic algorithms and three hybrid algorithms are used to solve it,and the performance of the algorithm is verified and evaluated by experimental testing;The results show that the genetic algorithm has a high overall accuracy,the particle swarm algorithm has fast speed,and the three hybrid algorithms balance the performance advantages of the two algorithms to a certain extent;The inversion results of actual measurement data show that the average relative errors of the five algorithms are all less than 1.17%,which satisfies the actual measurement requirements;In addition,the comparative performance of different algorithms under different models is analyzed,which provides a basis for the selection of algorithms under different measurement scenarios and optimization objectives.

关 键 词：多光谱温度测量 发射率 参考温度模型 启发式算法 算法分析 

分 类 号：TP751[自动化与计算机技术—检测技术与自动化装置]