Nonlinear wavelet compression methods for ion analyses and dynamic modeling of complex systems.

摘要

Chemometrics has been recognized as a powerful tool for identification, classification, and interpretation of data in modern laboratories. In instrumental analysis, chemometrics is often concerned with multivariate measurements for which several independent variables contribute to the measured responses. In particular, ion mobility spectrometry (IMS) furnishes sensitive, fast, portable, and inexpensive sensors that have a wide variety of potential applications. Data compression is useful in IMS when data is being continually collected and analyzed in real-time, or when the data is to be transmitted using wireless technology.; Nonlinear wavelet compression (NLWC) was developed and applied to ion mobility spectra from IMS instruments mounted in an unmanned aerial vehicle. To obtain a higher compression ratio, a two-dimensional nonlinear wavelet compression (2D-NLWC) was created and applied to the IMS data of chemical warfare simulants. However, modeling the data during the compression procedure itself is important. The algorithm of SIMPLe-to-use Interactive Self-modeling Mixture Analysis (SIMPLISMA) and alternating least squares (ALS) combined with 2D-NLWC, was designed to analyze the IMS data of chemical warfare simulants.; Wavelet compression is beneficial when applied to mass spectra. Mass spectral profiles retain information regarding ion peak shapes. This information can be used to differentiate spectra from isomers or other structurally similar compounds. During the traditional process of integrating raw mass spectra into histograms, some useful information regarding ion peak shapes may be lost. However, without such integration, it is difficult for analytical instruments to process the large quantity of raw data. Wavelet compression was used to reduce the large size of the mass profiles while at the same time retaining the useful information of ion peak shapes. This dissertation concluded that the mass profiles after wavelet compression provided better projected difference resolutions (PDRs) among xylene isomers than the histogram.