模型参考和误差白化的传感器动态补偿算法

通过模型参考的系统辨识方法建立传感器的动态补偿器, 常用的系统辨识采用均方误差(mean square error, MSE)为评判标准. 但是, 由于测量噪声存在和补偿器的频率特性, 使得补偿器的输入/输出信号存在严重的噪声干扰, 采用MSE为评判标准的系统辨识方法, 无法得到最优的补偿器的参数. 为此, 本文研究了一种以误差白化(error whitening criterion, EWC)为评判标准, 设计传感器动态补偿器的系统辨识方法. 该方法无需事先已知系统的动态特性, 根据传感器以及参考模型对

Reference-model and error-whitening in dynamic compensation for sensors

The dynamic compensator for a sensor can be modeled with a reference model identified by the MSE(mean square error) criterion. However, only the biased estimates of the parameters can be obtained due to the contamination of high level noises in the output/input signals and the limited frequency response range of the compensator. To tackle this problem, the error whitening criterion(EWC) is proposed by which a dynamic compensator can be designed without knowing the dynamic characteristics of sensor. Based on the step-input responses of the sensor and the reference model, the compensator parameters are estimated by the EWC. Finally, experimental results show that this approach is effective.