Advanced Ship Motion Forecasting for Expanded Aviation Operations

An innovative approach to Ship Motion Forecasting Systems (SFMS) based on Neural Net (NN) technology is proposed. Preliminary results using NNs with limited amounts of recorded ship motions indicate superior performance over classical forecasting techniques. The proposed Phase I effort will develop improved versions of the NNs and will demonstrate their performance on a significantly larger set of recorded ship motion data. The ship motions will be representative of different types of naval ships operating in various sea states. The Phase I effort will also develop performance metrics and confidence levels. For example, the estimated pitch at 10 seconds has a 90% chance of being within 0.25 degrees of the actual pitch. A software and hardware specification for a real time SFMS will be written during the Phase I Option, and a PC-based prototype system will be built and verified during the Phase II effort. Specialized NN or DSP hardware will be used if required for real time operation. The prototype system will include a hardware interface for current generation Shipboard Visual Landing Aid Systems, and will include a user interface with textured 3D graphics.