Data Driven Prognostics

To better interpret the complex and high-speed data in guidance, targeting, or control systems, we propose an algorithm that can dynamically analyze the spatial/temporal error patterns to determine the data error. Our goals for data driven prognostics arevery specific: (1) develop a genetic algorithm as the building block of the prognostic library, (2) implement a wavelet neural network, (3) use FPGAs to demonstrate concept validity, (4) simulate ASIC design. Once the FPGA design is completely verified andthe WNN performs as expected, we can translate the FPGA design into ASIC design if desired. If ASIC is determined to be the best fit for selected applications, we will be able to create a dedicated ASIC design, and this design will be thoroughly simulatedbut not fabricated in Phase I due to time and cost constraints. Based on guidance from the MDA, in Phase II a complete ASIC module can be fabricated and tested. For Phase I, Williams-Pyro intends to focus on one subsystem for proof of concept: opticaltracking and video targeting system. Because of the sensitive nature and availability of the data, in Phase I design and testing, we will simulate video data with commercial digital video camera recordings. The initial market for the proposed high datarate system prognostics system comprises new ships and aircraft, as the data-driven prognostics tool must be built into systems of the vehicles and is not appropriate for retrofitting. Although the algorithm developed during Phase I can potentially bedeveloped into products useful in a variety of applications, such as condition-based maintenance of manufacturing equipment and vehicles, the initial targets-new ships and aircraft-will be the focus of our marketing efforts for the first few years. Thesevehicles have multiple high data rate systems that would benefit from the speed of data processing and interpretation, and WPI's relationship with companies such as Bath Ironworks, Lockheed Martin, and Boeing will be useful in developing a product suitedto these targets and in marketing the prognostics integrated circuit (IC) for implementation in new ships and aircraft. Preliminary research indicates that the initial market of new aircraft and ships is promising, with extensive market potential: from$6.06 million in 2006 to $531.002 million in 2010.