Institute for Systems Researchhttp://hdl.handle.net/1903/43752017-11-16T05:11:39Z2017-11-16T05:11:39ZFramework for Knowledge-Based Fault Detection and Diagnostics in Multi-Domain Systems: Application to HVAC SystemsDelgoshaei, ParastooAustin, Markhttp://hdl.handle.net/1903/202072017-11-15T03:32:58Z2017-11-13T00:00:00ZFramework for Knowledge-Based Fault Detection and Diagnostics in Multi-Domain Systems: Application to HVAC Systems
Delgoshaei, Parastoo; Austin, Mark
State-of-the-art fault detection methods are equipment and domain specific and non-comprehensive.
As a result, the applicability of these methods in different domains is
very limited and they can achieve significant levels of
performance by having knowledge of the domain and the
ability to mimic human thinking in identifying the
source of a fault with a comprehensive knowledge of the system and its surroundings.
This technical report presents a comprehensive semantic framework for
fault detection and diagnostics (FDD) in systems simulation and control.
Our proposed methodology entails of implementation of the
knowledge bases for FDD purposes through the utilization of
ontologies and offers improved functionalities of such
system through inference-based reasoning to
derive knowledge about the irregularities in the operation.
We exercise the proposed approach by working step by step through
the setup and solution of a fault detection and diagnostics problem
for a small-scale heating, ventilating and air-conditioning (HVAC) system.
2017-11-13T00:00:00ZBat-Inspired Robot NavigationKuhlman, Michael JosephMcRoberts, Katehttp://hdl.handle.net/1903/201562017-10-13T02:34:06Z2009-08-01T00:00:00ZBat-Inspired Robot Navigation
Kuhlman, Michael Joseph; McRoberts, Kate
A key objective of Robotics is the autonomous navigation of mobile robots through an obstacle field. Inspired by echolocating bats, we developed a two-part navigation system consisting of obstacle detection through echolocation and motion planning. The first part relies upon a binaural sonar system, which emits ultrasonic pulses and then determines the interaural level difference (ILD) of the returning echoes to infer obstacle locations. Next, the Openspace motion planner computes the best direction of travel based on the locations of the target and the detected obstacles. We implemented this navigation system on a mobile platform, which repeatedly computes the safest direction of travel and moves accordingly, ultimately generating a real-time path to the goal.
2009-08-01T00:00:00ZOptimal replacement strategy for residential solar panels using monte carlo simulations and nonlinear optimization methodsPoissant, Andrewhttp://hdl.handle.net/1903/192292017-06-30T03:39:27Z2017-05-01T00:00:00ZOptimal replacement strategy for residential solar panels using monte carlo simulations and nonlinear optimization methods
Poissant, Andrew
The purpose of this analysis is to determine the optimal replacement strategy for a residential
photovoltaic (PV) array. Specifically, the optimal year and number of solar modules that should
be replaced on a residential solar panel system. This analysis aims at saving the stakeholder,
a homeowner with a residential PV array, money. A Monte Carlo simulation and nonlinear
mixed-integer programming are the analytic techniques used in determining the replacement
strategy. Localized cost of electricity (LCOE) is the objective function in these analyses. Modular, environmental, and market factors are all variables that can affect the LCOE. University
of Maryland’s LEAFHouse was the basis of these analyses because it is a house equipped with
an aging PV array and readily accessible data. Based on the findings in this report, it was
determined that 0 ± 0 solar modules should be replaced after 1.42 ± 0.32 years with a reference
year of initial installation being 2007. While the analysis results were not expected, they were
proven to be reasonable based on cost trends for solar panels and the calculated monetary value of the power production lost from the PV array.
2017-05-01T00:00:00ZOn The Number of Unlabeled Bipartite GraphsAtmaca, AbdullahOruc, Yavuz Ahttp://hdl.handle.net/1903/191862017-06-30T03:35:04Z2016-01-01T00:00:00ZOn The Number of Unlabeled Bipartite Graphs
Atmaca, Abdullah; Oruc, Yavuz A
Let $I$ and $O$ denote two sets of vertices, where $I\cap O =\Phi$, $|I| = n$, $|O| = r$, and $B_u(n,r)$ denote the set of unlabeled graphs whose edges connect vertices in $I$ and $O$. It is shown that the following two-sided equality holds.
$\displaystyle \frac{\binom{r+2^{n}-1}{r}}{n!} \le |B_u(n,r)| \le 2\frac{\binom{r+2^{n}-1}{r}}{n!} $
This paper describes a result that has been obtained in joint work with Abdullah Atmaca of Bilkent University, Ankara, Turkey
2016-01-01T00:00:00Z