项目名称: 基于太赫兹时域光谱的农药定性和定量分析若干关键技术研究
项目编号: No.61307127
项目类型: 青年科学基金项目
立项/批准年度: 2014
项目学科: 无线电电子学、电信技术
项目作者: 颜志刚
作者单位: 杭州电子科技大学
项目金额: 26万元
中文摘要: 太赫兹时域光谱(THz-TDS)技术由于其独特的光谱特性,正受到各领域学者越来越多的关注。本项目将利用THz-TDS技术分析浓度为ug/ml级的农药溶液在0.5-1.5THz的吸收谱特性及其与溶液浓度的关系,提出采用THz-TDS与全光谱偏最小二乘-最小二乘支持向量机相结合的方法,从理论上分析影响测量精度的误差来源;采用小波变换深入研究了THz时域光谱的信噪比。将离散小波多层分解和重构技术、小波变换的软阀值滤波和硬阀值滤波技术应用于THz时域光谱的噪声去除,提高THz时域光谱的信噪比;分别采用LS-SVM, Naive-Bayes 和 BP-ANN建模,实现不同品种农药的分类识别。同时,采用离子群算法和遗传优化算法优化了LS-SVM参数的求解;在建立物质THz吸收谱数据库的基础上,采用主成分分析法实现物质的在线识别检测;最终开发一套基于THz-TDS的农药残留无标记检测原型装置。
中文关键词: 太赫兹;光谱分析;定性分析;定量分析;农药
英文摘要: Terahertz time-domain spectroscopy (THz-TDS) is gaining more interests from various research fields due to its uniqueness. with THz-TDS,the 0.5-1.5 THz absorbance spectra of pesticide solutions with the concentration in the level of several ug/mls are analyzed, as well as their relationship to the solution concentrations.And a new method based on the THz-TDS technique and the combination of partial least square(PLS) and least-squares support vector machine(LS-SVM) modeling mehods is proposed for the detection of pesticide solutions. According to the model building and prediction preocesses, sources influencing the measurement precision are analyzed theoretically as well. The signal to noise ratio of the THz time domain spectroscopy was in-depth analyzed by Wavelet. Multi-layer decomposition and reconstruction of discrete wavelet,soft-threshold filtering and hard-threshold filtering of wavelet transform were applied respectively in THz time domain spectroscopy, and then the signal to noise ratio of the THz time domain spectroscopy was detail analyzed and improved. Least Squares Support Vector Machines, Naive Bayes and Back Propagation Artificial Neural Network were applied to achieve Multi-class classification of these different kinds of pesticides, and the classification results of three algorithms were analyzed
英文关键词: Terahertz;Spectral analysis;Qualitative Analysis;Quantitative Analysis;Pesticides