基于小波神经网络的新型角度传感器补偿技术  

Compensation technique for novel angular sensor based on wavelet neural network

作　　者：赵伦 张靖 周路遥 季丹 尹蝶 普春芬 ZHAO Lun;ZHANG Jing;ZHOU Luyao;JI Dan;YIN Die;PU Chunfen(School of Physics and Information Engineering,Zhaotong University,Zhaotong 657000,China)

机构地区：[1]昭通学院物理与信息工程学院,云南昭通657000

出　　处：《传感器与微系统》2025年第4期96-100,共5页Transducer and Microsystem Technologies

基　　金：云南省科技计划项目(202303AA080041)。

摘　　要：针对现有霍尔角度传感器、光电编码器等存在编码难度较大、成本较高的问题,设计了一种新型的角度传感器,主要由激光器、摄像头、转盘和转轴组成,转轴带动转盘转动时,转盘上的激光器发出的激光束在摄像头中的成像上形成光斑,利用二阶矩法识别出光斑中心点坐标,并将其转换为对应的旋转角度值。利用傅里叶级数拟合法对旋转角度与光斑中心点坐标进行拟合分析,以确定传感器的非线性特性。并采用小波神经网络(WNN)对传感器存在的非线性问题进行补偿。实验结果表明:经WNN补偿后,传感器的非线性误差从2.39降低至1.56,测量精度达到了0.84°。Aiming at the problem of extremely difficult encoding and high cost associated with existing Hall angular sensor and photoelectric encoders,a novel angular sensor is designed.It primarily consists of laser,camera,rotating disk,and shaft.As the shaft drives the disk to rotate,the laser beam emitted from the laser emitter on the disk forms a spot on imaging of camera.The center point coordinates of the spot are identified using the second-order moment method and convert into corresponding rotational angle values.Fourier series fitting method is employed for fitting analysis of rotational angle and the spot center point coordinates,thereby determine the nonlinear characteristics of the sensor.A wavelet neural network(WNN)is utilized to compensate for the nonlinear issues inherent in the sensor.Experimental results indicate that after WNN compensation,the nonlinear error of the sensor is reduced from 2.39 to 1.56,measurement precision achieve 0.84°.

关 键 词：角度传感器 小波神经网络 二阶矩法 傅里叶级数 非线性补偿 

分 类 号：TP212[自动化与计算机技术—检测技术与自动化装置] O439[自动化与计算机技术—控制科学与工程]