Date

October 2005

Document Type

Thesis

Degree Name

M.S.

Department

Dept. of Computer Science and Engineering

Institution

Oregon Graduate Institute of Science & Technology

Abstract

The goal of this research is to improve the performance of the Maximum Daily Salinity regressor used in the fault detection mechanism deployed in the Columbia River estuary (CORlE). The Center for Coastal and Land-Margin Research is developing an Environmental Observation and Forecasting System. The goal of the CORlE project is to gain a better understanding of the estuary. The team has deployed sensors in the estuary to measure salinity, temperature, pressure, and velocity. Of these sensors, salinity sensors are subject to bio-fouling, an event that results in data loss over time. Previous work in fault detection helped prevent data loss. Our work improves the performance of the regressor used as part of the detector architecture. We looked at temperature measurements as inputs for the salinity regressor. We used the Gaussian Mixture Model to build a new salinity regressor. In addition to the Gaussian Mixture Model, we attempted to include historical information into our regressor, explored the use of single-layer neural networks, and considered incorporating measurements from nearby stations to improve regressor performance. We also considered incorporating numerical predictions for salinity from SELFE, a numerical model of the estuary developed by the CORIE team. We show a performance comparison of the original and new regressors.

Identifier

doi:10.6083/M41V5BWF

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.