Stochastic Characterizations for Signals & Systems Associated with Quasi-Periodic Systems.

摘要

The goal of this funded research was to complete the development of a fault diagnosis system for rotating machinery, which was initiated under a previous AFOSR grant (F49620-95-1-0090). The focus of the proposed research was strongly linked to the accomplishments achieved to date. Because the value of such a system is ultimately gauged by it's applicability and utility in real world settings, a major opponent of the research results contained herein center on real world issues which violate many of the assumptions used to develop advanced a signal analysis methods. Specific technical outcomes related to this research include: (a) Characterization of the nature of 'real-world' sinusoids associated with rotating machinery (b) Development of improved periodic time/ frequency analysis tools for quasi-periodic processes (c) Implementation of theoretical statistics derived under previous funding in the development of a multi-resolution, intelligent methodology for estimating the spectral information associated with mixed random processes. We have also aggressively pursued establishment of research ties with U.S. and foreign industries and academic institutions to evaluate the utility of the above results and to guide the direction of future research and funding opportunities. These efforts are documented in this report. We believe that they are valuable not only for transitions of research results obtained under this grant, but also for securing future support from other agencies.