Enabling HH-60G Predictive Maintenance via Computational Fluid Dynamics (CFD) Artificial Intelligence (AI) Rotorcraft Development and Modeling (CARD-M)

View Video Presentation: https://doi.org/10.2514/6.2023-0422.vid 4G accelerometer signal dropout results in significant H-60 maintenance (Mx) downtime due to overly conservative requirements to compensate for corrupted 4G flight data. When the data is in fact corrupt or unavailable, worst case scenario conditions must be assumed to reduce risk; this causes significant impacts to H-60 fleet readiness, cost, and system level impact. One tactic to address this issue is to partially or fully reconstruct the corrupt signal using stacked modeling with a combination of Artificial Intelligence (AI)/Machine Learning (ML) in conjunction with a semi-empirical synthetic signal leveraging CFD. To support and inform the separate AI/ML modeling, and aid in the reconstruction of these signal dropouts, a CFD model of the HH-60G is developed using CREATE-AVTM Helios to simulate flight performance during predetermined maneuvers. That being said, CFD simulations can be exceedingly computationally expensive and as such high fidelity methods are compared to medium fidelity methods in order to reduce the computational burden and accelerate data generation. The results of the CFD simulation are post-processed to generate synthetic force and acceleration signals corresponding to aerodynamic and inertial loads on the vehicle during flight, along with off-body complex flow visualization for additional insight into flight characteristics. Using heritage flight data, the Joint AI Center (JAIC) Predictive Maintenance (PMx) technical guidance, and the Army Engineer Research and Development Center (ERDC) High Performance Computing (HPC) computational resources, these modeled synthetic signals are compared to scripted flight data for Verification and Validation (V&V) against baseline flight conditions of interest.