Research :: Nonlinear time series analysis
MCEC Nonlinear Time Series Analysis Software (NTSAS) in MATLAB

My graduate research assistantship was sponsored by the Measurement and Control Engineering Center at the University of Tennessee, Knoxville, through the CANDIES[Control/Chaos and Nonlinear Dynamics for Industrial Engineering Systems] project that focused on applying chaos theory to industrial systems with the goal to improve control and to facilitate fault diagnosis and process monitoring. Principal Investigators for the project were Duane D. Bruns and C. Stuart Daw. MCEC is a Industry/University .

As a part of the project, C. E. A. Finney and I developed sophisticated sophisticated GUI based software for nonlinear time series analysis in MATLAB. The software, called NTSAS (patent pending) contains state-of-the-art tools for nonlinear time series analysis and symbolization. The software is copyrighted to MCEC. In 2002, I administered a Chaos Short Course Workshop held at UT Knoxville for the industrial members of MCEC as a part of CANDIES project. Representatives from Exxon Mobil, Dow, etc. attended the workshop.

The software also has educational value and was used successfully in a 600-level course in nonlinear dynamics and chaos at UT Knoxville.

See a poster that has screenshots of the software. Note that this is how the software looked in 2001, and we have had two more releases in 2002 and 2003.

The documentation for most functions has an excellent Reference Manual authored by Charles Finney. I wrote the user's guide for NTSAS and two training sessions using industrial and experimental data to demonstrate the usefulness of the package as well as to present an analysis pathway given nonlinear (continuous or discrete) time series.

The most curious part of this development was that when I started out in early 1999 to make a sophisticated front-end interface, there was little or no documentation available on how to do that in MATLAB. Even now (as of 2005), I am told, documentation is sparse --and although simple tasks have been made accessible, adding functionality is not easy.

The CANDIES project posters were adjudged the best 6 times in MCEC meetings held in 1999 through 2003.