摘要
为提高声学层析成像温度场重建性能及热点追踪能力,建立了基于改进的径向基函数神经网络(IMRBFNN)声学温度场重建模型,并提出基于温度梯度自适应的网格划分和布点方案。该方法在RBFNN隐藏层加入多项式作为修正项建立IMRBFNN重建模型。首先将测量区域均匀划分网格进行初始温度场重建,依据温度梯度信息创建概率分布数组,在区域内填充不同尺寸、相互之间无覆盖的圆,然后应用Voronoi多边形镶嵌法则对区域进行多边形划分,对分布点重建温度场。与经典最小二乘法、Tikhonov正则化方法对比,该方法可有效将温度场信息可视化并追踪热点位置,单热点温度场重建误差<2%。在自主研发的声波测温实验系统中,可检测到加热器移动0.25 m时的热点成像。仿真和实验均表明该声学方法能有效实现温度场热点追踪。
To enhance the capability of acoustic tomography for hotspot tracking within temperature fields,an improved radial basis function neural network(IMRBFNN)acoustic temperature field reconstruction model was developed,and a temperature gradient adaptive mesh partitioning and point layout scheme were proposed.This method established an IMRBFNN reconstruction model by adding polynomials as a correction term to the hidden layer of RBFNN.Firstly,divided the measurement area evenly into meshs and reconstructed the initial temperature field.Created a probability distribution array based on the temperature gradient information,and filled the area with non-overlapping circles of varying sizes.Then,applied the Voronoi polygon tessellation rule to divide the area into polygons and reconstructed the temperature field of the distribution points.Compared with classical least squares and Tikhonov regularization methods,this method can effectively visualize temperature field information and track hotspot locations,with a reconstruction error of less than 2%for a single hotspot temperature field.In the self-developed acoustic temperature measurement experimental system,hotspot imaging can be detected when the heater moves 0.25 m.Simulation and experiments have shown that this acoustic method can effectively track hotspots in the temperature field.
作者
王哲
孔倩
姜根山
赵义豪
刘月超
WANG Zhe;KONG Qian;JIANG Genshan;ZHAO Yihao;LIU Yuechao(Department of Mathematics and Physics,North China Electric Power University,Baoding,Hebei 071003,China;Hebei Key Laboratory of Physics and Energy Technology,North China Electric Power University,Baoding,Hebei 071003,China;School of Mathematics and Physics,North China Electric Power University,Beijing 102206,China)
出处
《计量学报》
北大核心
2025年第8期1118-1126,共9页
Acta Metrologica Sinica
基金
国家自然科学基金(12104151,12274122)
河北省自然科学基金(A2021502005)
中央高校基本科研业务费专项(2021MS118)。
