High fidelity simulations of complex systems still pose a challenge to the scientist trying to understand its physical characteristics. The challenge is in finding useful bits in terabytes of data that directly relate to the nature of time-varying, multivariate data. An intelligent data mining capability is needed that has both knowledge (descriptive physics) and foresight (cognitive model of users needs). Concurrent Agent-enabled Feature Extraction (CAFÉ), from 21st Century Systems, Inc. and Brigham Young University, will address this challenge. CAFÉ features 21CSIs intelligent agent technologies that leverage BYUs expertise. CAFÉs innovative intelligent agent structure and evidential inference engine will allow concurrent data-mining, making it possible for multiple analysis methods to work together to improve the data-mining performance. This phase implements a bottom-up clustering algorithm to help tune feature extraction and predict features well before the simulation has converged. The agent design allows direct collaboration between data-mining algorithms and scientist. CAFÉ allows the scientist to observe and correct data-mining of simulations without wasting valuable research time. With our impressive track record of transitioning technology (100th percentile DoD commercialization index) and our strong team, we are the right team to provide intelligently guided concurrent data-mining for high fidelity fluid dynamic simulations.
Benefit: Current costs to develop a new aircraft engine are more than $2 billion and 10 years. While the Air Force and other services dont develop these components internally, the development of a new aircraft engine or airframe can have significant military application. University researchers and commercial contractors perform the lions share of this type of work. A better understanding of the complexities of the component and its interactions could have a significant impact in extending the envelope of current aircraft and in the development of new aircraft. CAFÉ offers the capability to perform intelligent concurrent data-mining for high fidelity fluid dynamic simulations. The CAFÉ tool is able to extract and display accurate, near real-time patterns from massively large data sets by searching for physics-related events in complex large scale simulations. With CAFÉ, the potential exists to go even further through the interaction of concurrent, on-the-fly, queries and responses among the CAFÉ agents as well as with the human operator. The primary target customer will be the CFD researcher at the research institution or commercial entity. Initially, the CAFÉ tool will be tailored to simulations of fans and compressors of gas turbine engines. In order to minimize overall technical risk while at the same time reducing physical testing costs, aircraft manufactures continuously search for computational design tools to remain competitive. They rely heavily on continual advancement of the technological frontier for faster solutions to more complex problems, more accurate results with improved performance, enhanced safety, environmental acceptability, and less time involved to develop new products. 21CSI recognizes these challenges facing aircraft manufactures. The computational models developed under CAFÉ will significantly accelerate the testing cycle allowing more efficiency in the design processes used to develop various types of flight vehicles. Using these tools, aircraft engineers and designers can address a wide range of design challenges including airfoil and geometry selection, wake vortex alleviation, flutter aero-elastics, load design, stability and control, and high-speed wing design. Businesses will be easily able to justify the CAFÉ software expense with the time saving that achieve.
Keywords: Turbulence, Feature Extraction, Data-Mining, Artificial Intelligence, Evidential Reasoning, Data Inf