基于动态光谱的血糖无创检测中有效信号的提取与建模方法研究

Effective Signal Extraction Method and Model Building in Noninvasive Detection of Blood Glucose Based on Dynamic Spectrum

摘要: 为提升无创检测的准确度，本文在动态光谱检测预处理方法中应用了新的信号提取方法，对接对数脉搏波的所有上升沿与下降沿，筛选有效沿，剔除粗大误差，提高了信噪比，消除了频谱重叠现象.在建模方法上，采用主成分分析方法对多样本的动态光谱值矩阵实现降维处理，采用GA与BP神经网络相结合取长补短的方法建立最终的数学模型，实现血液成分浓度预测的目的.实验表明，该方法能够提高血糖检测的准确度，有利于动态光谱在血糖无创检测方面的发展.

Abstract: In order to improve the accuracy of noninvasive detection,a new method of signal extraction from the dynamic spectrum was presented.With this method,spectrum overlap was eliminated effectively.In the model building method,principal component analysis was used to reduce the dimension of the dynamic spectral matrix,genetic algorithm and BP neural network was combined to establish mathematical model to predict the blood glucose.Experimental result shows that this method can improve the accuracy of blood glucose detection.