偏最小二乘法在传感器误差补偿中的应用
2021-08-19徐建忠张彦超李永生于广浩苏奎
徐建忠 张彦超 李永生 于广浩 苏奎



摘 要:在实际测量中由于湿度温度等不确定性使得传感器数据出现无法避免的随机性误差。为了减小误差,一种快捷简便的处理方式是使用最小二乘法对数据进行线性回归修正。线性补偿的方式能解决很多传感器测量对于噪声等随机现象出现的误差,但通常测量数据的实际函数本身都是非线性的,用线性函数来模拟非线性的测量数据往往会出现精度不足的问题。为此,本文设计了一种增加自变量阶次及变量系数的方式来提高补偿精度,由于增加的变量系数可能导致多重相关性等问题,改用单因变量偏最小二乘法来建立补偿模型。
关键词:传感器数据;数据补偿;偏最小二乘
中图分类号:TP391 文献标识码:A DOI:10.3969/j.issn.1003-6970.2021.02.022
本文著录格式:徐建忠,张彦超,李永生,等.偏最小二乘法在传感器误差补偿中的应用[J].软件,2021,42(02):075-077
Application of Partial Least Squares in Sensor Error Compensation
XU Jianzhong, ZHANG Yanchao, LI Yongsheng, YU Guanghao, SU Kui
(Mudanjiang Medical University, Mudanjiang Heilongjiang 157011)
【Abstract】:In practice, uncertainties such as humidity and temperature cause unavoidable random errors in the sensor data. In order to reduce the error, a quick and easy way to deal with it is to use the least squares method to correct the data by linear regression. The linear compensation method can solve many sensor measurement errors for random phenomena such as noise, but usually the actual function of the measurement data itself is nonlinear, and using a linear function to simulate nonlinear measurement data often results in a lack of accuracy. For this reason, this paper designs a way to increase the order of independent variables and variable coefficients to improve the compensation accuracy, and because the increased variable coefficients may lead to problems such as multiple correlations, a single dependent variable partial least squares method is used instead to build the compensation model.
【Key words】:sensor data;least square;data compensation
由于传感器本身属性或者湿度、温度、噪声等影响,传感器在测量中不可避免的发生随机误差[1-2]。现今传感器的数据补偿方式主要分为硬件补偿与软补偿(数字补偿)两大类[3-4]。前一种方式为改进传感器工艺、提高精度或通过对测量电路与软补偿各自的优势进行综合以达进行优化来达到对传感器测量数据补偿[5]。后一种方式则通过智能算法-包括数值分析或神经网络学习等方式对采集数据进行回归[6]。亦可以综合硬件补偿到提高精确测量的目的[7]。
基于最小二乘方法的回归补偿由于其可适用范围广、建模简单、操作方便等特性现已广泛应用于各式传感器的软补偿方式中[8-10]。……
