An Unstructured Mesh Approach to Nonlinear Noise Reduction (Technical Paper with Briefing Charts)
Abstract
In any type of data acquisition, the event of gathering undesirable noise along with desirable data is inevitable. To denoise signals originating from smooth, chaotic attractors, the Air Force Research Laboratory (AFRL) adapted the time-delay embedding theory of Takens' Theorem (1981) and the causation-detecting method of Convergent Cross Mapping (CCM) to develop a grid-based denoising technique. Given a clean signal from such a dynamical system, AFRL's technique attempts to denoise a corrupted signal observed from the same system. To improve this grid-based method, we implement an unstructured mesh based on triangulations and Voronoi diagrams that better distributes data over mesh cells and improves the accuracy of the reconstructed signal. Our method achieves statistical convergence with known test data and reduces synthetic noise on experimental signals from Hall Effect Thrusters (HETs) with greater success than the grid-based strategy.
Document Details
- Document Type
- Technical Report
- Publication Date
- Sep 23, 2020
- Accession Number
- AD1128505
Entities
People
- A. Kirtland
- Craig J. Johnson
- Daniel Eckhardt
- J. Botvinick-greenhouse
- M. Debrito
- M. Osborne
- Richard K. Martin
Organizations
- Air Force Research Laboratory