"Nonlinear Systems Identification for Modelling and Analysis of Biological, Physical, and Industrial Processes"

点击翻译按钮获取中文摘要

Objectives of the proposed research are: (1) to develop techniques for nonlinear system identification and time-series analysis, creating methods capable of extracting useful information, e.g., significant frequency content, from short noisy experimental records. (2) to apply such methods to important problems in biology, and to physical and industrial systems. Scientific Approach: Regarding system identification, the proposed work involves representation of nonlinear systems by functional expansions and kernel estimation, parallel cascade identification (PCI) employing dynamic linear and static nonlinear elements, and use of other nonlinear models such as difference equations. Other means of representing nonlinear systems will continue to be actively explored. Regarding time-series analysis, proposed work includes estimation of sinusoidal, exponential, and exponentially-decaying sinusoidal series models from brief noisy time-series that may be unequally spaced. Regarding application of these methods in biology, the applicant intends to concentrate much activity on the interpretation of gene expression profiles and of tissue microarrays, such as for diagnosis and for prediction of treatment response. Physical and industrial applications will include reducing errors inherent in using (micro-electro-mechanical-system) MEMS-based inertial navigation systems (INS) and their fusion with a global positioning system (GPS).