Acoustic stealth plays a critical role in the effective deployment of naval platforms. For in-service platforms, one approach is to select those systems known to have a low impact on the acoustic signature. However, systems degrade over time and some systems may not perform as expected. The problem is further compounded with the complex interaction between systems, where known quiet systems when coupled together result in a greater impact on the signature. In addition, given the large number of systems on most platforms, managing every system whilst maintaining operability is not practical. In this contribution we demonstrate an approach for selecting the optimal configuration of systems in terms of vibration levels whilst maintaining platform operability. In particular, a Genetic Algorithm is used to identify which systems should be selected to minimise the measured vibration levels at a number of hull positions. Measured platform data is used to demonstrate the approach; however, this data isn’t comprised of measurements across all possible system configurations. Therefore, an optimiser is assisted with a Machine Learning model to make vibration predictions for system configurations not previously observed. Practical constraints on which systems can operate in combination are implemented to ensure the optimiser produces system configurations that would maintain feasible platform operability. Results are presented for identified optimal system configurations showing vibration levels relative to those observed during normal operating configurations for an in-service platform.
