高精度椭圆阵列电流传感器误差优化算法研究  

Research on error optimization algorithm of high-precision elliptic array current sensor

作　　者：沈悦[1] 陆佳嘉 唐玥 权硕 褚子扬 SHEN Yue;LU Jiajia;TANG Yue;QUAN Shuo;CHU Ziyang(School of Automation,Nanjing University of Information Science&Technology,Nanjing 210044,China;School of Internet of Tings Engineering,Wuxi University,Wuxi 214105,China)

机构地区：[1]南京信息工程大学自动化学院,江苏南京210044 [2]无锡学院物联网工程学院,江苏无锡214105

出　　处：《传感器与微系统》2025年第2期39-43,共5页Transducer and Microsystem Technologies

基　　金：国家自然科学基金面上项目(41674036);无锡市科技发展资金资助项目(N20201011);江苏省自然科学基金面上项目(BK20211037)。

摘　　要：为了减小由于矩形导体偏心和倾斜导致的传感器测量误差,实现电流高精度测量,提出一种基于灰狼优化(GWO)算法优化反向传播(BP)神经网络的椭圆阵列霍尔电流传感器误差优化方案。首先,基于三维磁场模型确定导体偏心参数和倾斜参数,并进行电流反演,得出电流积分模型;然后,使用BP神经网络估计导体参数,并引入GWO算法优化BP神经网络的初始权值和阈值,构建导体状态参数估计模型,实现偏心与倾斜误差优化;最后,搭建实验平台,对本文提出的误差优化方案进行验证。实验结果表明:本文提出误差优化方案能够精确估计导体状态参数,进而有效减小偏心误差和倾斜误差,导体X向偏心产生的电流误差减小65.07%,Y向偏心产生的电流误差减小45.74%,偏离Z轴产生的倾斜电流误差减小76.15%。An elliptic array Hall current sensor error optimization solution based on the gray wolf optimization(GWO)algorithm optimized back propagation(BP)neural network is proposed in order to reduce the sensor measurement error due to the eccentricity and tilt of the rectangular conductor and to achieve the high-precision measurement of the current.Firstly,the conductor eccentricity and tilt parameters are determined based on the three-dimensional(3D)magnetic field model,and the current inversion is carried out to derive the current integral model.Then,BP neural network is used to estimate the conductor parameters,and the GWO algorithm is introduced to optimize the initial weights and thresholds of the BP neural network,to construct the estimation model of the conductor state parameters,and to achieve the optimization of the eccentricity and tilt errors.Finally,the experimental platform is set up to verify the proposed optimization scheme of errors.The experimental results show that the proposed error optimization scheme can accurately estimate the state parameters of X-direction eccentricity of the conductor,and then,effectively reduce the eccentricity error and tilt error,the current error generated by X-direction eccentricity of the conductor is reduced by 65.07%,the current error generated by the Y-direction eccentricity is reduced by 45.74%,and the tilt current error generated by deviating from the Z-axis is reduced by 76.15%.

关 键 词：椭圆阵列传感器 矩形导体 偏心误差 倾斜误差 灰狼优化算法 反向传播神经网络 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] TM933.14[自动化与计算机技术—控制科学与工程]