Compressed Prediction of Time Series Three-dimensional Fluorescence Spectroscopy
YU Shao-hui1,XIAO Xue2,DING Hong1,XU Ge1
1. School of Mathematics and Statistics, Hefei Normal University, Hefei, Anhui 230601, China
2. Key Laboratory of Environmental Optics & Technology, Anhui Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Hefei, Anhui 230031, China
Abstract：As the time series spectroscopy data is usually redundant and it is a non-stationary process, it is firstly compressed by the wavelet transform along both the time mode and the spectroscopy mode. The distortion rate of two-dimensional reconstruction is smaller than 0.1 and the similarity is lager than 0.99. As for the fluorescence region, the distortion rate of one-dimensional reconstruction is also smaller than 0.2 and the similarity is lager than 0.9. All this demonstrates the effectiveness of the compression by wavelet transform. After the compression, the prediction of time series spectroscopy data is completed by ARIMA(p,d,q) model and six different tests are discussed. Furthermore, it is compared with wavelet neural network prediction model. Numerical results show the effectiveness of compression and the fast and accurate ability of the model.
［1］Cohen E, Levy G J, Borisover M. Fluorescent components of organic matter in wastewater: efficacy and selectivity of the water treatment ［J］. Water Research, 2014, 55:323-334.
［2］杨丽丽, 王玉田, 鲁信琼. 水中石油类污染物的光纤荧光在线测量技术研究［J］. 计量学报, 2014, 35(4):393-397.
［3］Santacruz P, Garcíareiriz A. Application of third-order multivariate calibration algorithms to the determination of carbaryl, naphthol and propoxur by kinetic spectroscopic measurements［J］. Talanta, 2014, 128:450-459.
［4］Yin X L, Wu H L, Gu H W, et al. Second-order calibration method applied to process three-way excitation-emission-kinetic fluorescence data: A novel tool for real-time quantitative analysis of the lactone hydrolysis of irinotecan in human plasma［J］. Chemometrics and Intelligent Laboratory Systems, 2015, 146:447-456.
［5］吴希军,田瑞玲,杜德琴. 荧光谱及平行因子分析在芝麻油与芝麻香精鉴别中的应用［J］. 计量学报, 2016, 37(6): 657-662.
［6］谷政, 褚保金, 江惠坤. 非平稳时间序列分析的WAVELET-ARMA组合方法及其应用［J］. 系统工程, 2010, 28(1):73-77.
［7］周扬, 戴曙光, 吕进, 等. 近红外光谱小波投影寻踪回归定量分析方法研究［J］. 计量学报, 2013, 34(2):184-186.
［8］徐科, 徐金梧, 班晓娟. 基于小波分解的某些非平稳时间序列预测方法［J］. 电子学报, 2001, 29(4):566-568.
［9］贺国光, 马寿峰, 李宇. 基于小波分解与重构的时间序列预测法［J］. 自动化学报, 2002, 28(6):1012-1014.
［10］Wang W C, Chau K W, Xu, D M, et al. Improving Forecasting Accuracy of Annual Runoff Time Series Using ARIMA Based on EEMD Decomposition［J］. Water Resources Management, 2015, 29(8):2655-2675.
［11］Tran H D, Muttil N, Perera B J C. Selection of significant input variables for time series forecasting［J］. Environmental Modeling & Software, 2015, 64:156-163.
［12］孙永明, 吴谨, 刘劲, 等. 基于DWT的高频系数压缩感知图像融合［J］. 红外技术, 2014, 36(9):714-718.
［13］岳著风, 王颖. 基于小波包变换的三维荧光光谱数据压缩方法［J］. 北京化工大学学报(自然科学版), 2013, 40(6):100-105.
［14］王洪斌,王世豪,籍冰朔,等. 基于改进多阈值小波包的去噪算法及应用［J］. 计量学报, 2016, 37(2): 205-208.
2017, Vol. 38 
