Thesis
Smart simulation techniques for the evaluation of parametric uncertainties in black box systems
Master of Science (MS), Washington State University
2005
Handle:
https://hdl.handle.net/2376/331
Abstract
When parameters of complex processes are uncertain, it is often necessary to perform exhaustive simulations to characterize the outputs of these processes. If simulations are computationally intensive, characterization of outputs through exhaustive simulations may be infeasible. In such cases, intelligent approaches for choosing simulations based on probabilistic descriptions of uncertainties may be valuable. The Probabilistic Collocation Method is a probabilistic technique that can model the deterministic relationship between the uncertain parameters and an output of interest with a small set of simulations. In this thesis, we review PCM, provide a new generalization of PCM for systems with multiple correlated uncertain parameters and also present an order selection algorithm for the technique. Although we tout PCM as a very economic technique, the number of simulation points nevertheless grows exponentially with the number of uncertain parameters. To overcome this difficulty, we develop some Information theory-based techniques that can allow us to apply PCM using only a subset of the uncertain input parameters. We conclude our analytical development by discussing the possibility of applying PCM to solve optimization problems. In the penultimate chapter we illustrate a possible PCM application in Computer Science (specifically, in queueing theory) and also develop a larger electric power system example. We conclude the thesis by summarizing our results and discuss future directions. In our work and previous work, PCM has been used in such diverse areas as global climate evolution studies, chemical engineering applications and power systems analysis, which indicates the versatile nature of this algorithm.
Metrics
1 File views/ downloads
10 Record Views
Details
- Title
- Smart simulation techniques for the evaluation of parametric uncertainties in black box systems
- Creators
- Dinesh Ramamurthy
- Contributors
- Sandip Roy (Degree Supervisor)
- Awarding Institution
- Washington State University
- Academic Unit
- Electrical Engineering and Computer Science, School of
- Theses and Dissertations
- Master of Science (MS), Washington State University
- Identifiers
- 99900525100601842
- Language
- English
- Resource Type
- Thesis