基于噪声方差估计和模型参考的传感器动态补偿  

Dynamic compensation based on noise variance estimation and model reference for sensors

作　　者：刘清[1] 曹国华[2] 

机构地区：[1]南京师范大学计算机科学与技术学院,江苏南京210042 [2]南京师范大学电气与自动化工程学院,江苏南京210042

出　　处：《江苏大学学报（自然科学版）》2009年第6期601-605,共5页Journal of Jiangsu University：Natural Science Edition

基　　金：国家自然科学基金资助项目(60774060)

摘　　要：针对采用最小二乘法(least-squares,LS)无法得到传感器动态补偿器参数无偏估计的问题,提出了一种基于输入信号噪声方差估计和模型参考的无偏辨识算法.该算法先通过小波变换估计噪声的方差,再由估计得到的方差,通过偏差消除的递推最小二乘法,对补偿器的参数进行无偏辨识.同时,利用参考模型建立卡尔曼滤波器,消除高频噪声对测量精度的影响.在薄膜热电偶上进行了仿真研究,以验证该方法的有效性.结果表明,算法降低了对输入辨识信号的要求,即不再必须为白噪声,具有较强的实用性.To solve the problem that unbiased estimate of parameters of the sensor dynamic compensator can not be obtained by using the ordinary least-squares method,a new unbiased identification algorithm is proposed based on the noise variance of the input signal and mode reference.The noise variance was firstly estimated by using the wavelet transform,and then a recursive least-squares method with bias-elimination was applied to estimate the unbiased parameters of the compensator. At the same time,Kalman filter was constructed on reference mode to eliminate the effect of high frequency noise on measurement precision. The simulation experiment was implemented on film thermocouple to verify validity of this algorithm. The results show that the proposed algorithm lowers the request of input signal,so that the white noise is no longer needed,and this algorithm has better practicability.

关 键 词：噪声 动态测量 偏差消除 方差估计 卡尔曼滤波 

分 类 号：TP212.6[自动化与计算机技术—检测技术与自动化装置]