You are here

Characterization of Store Trajectory Dynamics Released from Internal Cavities Using Machine Learning, Artificial Intelligence and Other Advanced Data Analysis Techniques


TECH FOCUS AREAS: Artificial Intelligence/Machine Learning


TECHNOLOGY AREAS: Information Systems; Air Platform


OBJECTIVE: This topic's objective is to develop analysis techniques via application of machine learning, artificial intelligence and/or other advanced data analysis techniques to evaluate and characterize large amounts of trajectory data generated for stores released from an internal cavity weapons bay. The goal would be to utilize such techniques to identify and subsequently exploit potential linkages between flow conditions in the cavity at and after the time of release with the disparity of the store trajectories observed due to variation in release time.


DESCRIPTION: A large dataset consisting of approximately 100 cases is currently being generated via high-fidelity CFD simulating the trajectories of small, light-weight stores being released from internal weapons bays (cavities) at high speeds. The simulation in this dataset primarily consists of the store configuration being held in carriage for some period of time and then released using a prescribed ejector profile, with the release time being the only variation in the simulations. It has been shown that the time of release of the store has a significant impact on its subsequent trajectory due to the unsteady flow-field in the cavity. The existing CFD dataset consists of high-frequency integrated force/moment components acting on the store, two-dimensional flow-field representations at various spanwise locations and heights in the cavity, and pressure time histories at various positions on the cavity walls/ceiling and the store prior to release as well as during the trajectory.


Additional data could also be collected during subsequent simulations as needed to develop appropriate analysis techniques. This rich data set will be provided as a training set in order to use various AI/ML or other analysis techniques to attempt to determine if there is some predictable cavity flow-field and/or force/moment state either 1) at the time of release and/or 2) after release while the store is traversing the cavity, shear layer and/or free-stream that leads to specific trajectory states. Of particular interest are the states associated with “bad” releases, defined as the distance between the store center of gravity and aircraft hardware not monotonically increasing or the store entering the free stream with high rates of pitch and/or yaw.


PHASE I: Phase I efforts will determine the scientific and technical merit and feasibility of application of AI, ML and/or advanced analysis techniques to determine root causes for a specified store to reach a particular state when released from an internal store configuration. High-fidelity, unsteady CFD of 6DOF trajectories generated for a particular store released at various times will be provided as GFE.



Tangible outcomes for the Phase I effort would be the demonstration of a practical process to relate particular states of the cavity to specific trajectory behaviors. The envisioned main deliverable for Phase I would be a report documenting the process with sufficient detail to allow evaluation by the government and example(s) of its application on the dataset provided. Identification of the overall plan to mature the concepts into a useable tool along with plans to generate additional data needed to support development/expansion of method to additional configurations should also be reported.


PHASE II: Further develop the approach to demonstrate its ability to identify conditions in the cavity (including the shear layer) related to trajectories. This identification should be probabilistic in nature, where certain flow features and/or force/moment states produce, bad trajectories are observed to exist in some statistically significant number of cases.


Extension of approach to data from other stores and/or other cavity configurations would be encouraged. Tangible outcomes and expected deliverables for the Phase II effort would include stand-alone software that would take in high-fidelity unsteady CFD data and produce output that could identify release points or flow states/flow-field features associated with problematic trajectories. A stretch goal would be the inclusion of surrogate modeling of key cavity environmental features that would permit reduced order evaluation of configurations beyond the training data set.


PHASE III DUAL USE APPLICATIONS: Phase III efforts will focus on transitioning the developed technology to a working commercial or warfighter software/processes. Solutions developed will be immediately relevant to precision airdrop, cargo and weapons release, among a whole range of commercial and military applications. If a viable approach to identify conditions associated with bad trajectories are identified, this would allow potential flow-control solutions to be investigated to "fix" these conditions and diminish problematic releases. They would be in a position to supply future software/processes to the Air Force, and other DoD components to facilitate future weapons bay designs that would improve separation characteristics.


NOTES: The technology within this topic is restricted under the International Traffic in Arms Regulation (ITAR), 22 CFR Parts 120-130, which controls the export and import of defense-related material and services, including export of sensitive technical data, or the Export Administration Regulation (EAR), 15 CFR Parts 730-774, which controls dual use items. Offerors must disclose any proposed use of foreign nationals (FNs), their country(ies) of origin, the type of visa or work permit possessed, and the proposed tasks intended for accomplishment by the FN(s) in accordance with section 5.4.c.(8) of the Announcement and within the AF Component-specific instructions. Offerors are advised foreign nationals proposed to perform on this topic may be restricted due to the technical data under US Export Control Laws. Please direct questions to the Air Force SBIR/STTR Help Desk:



  1. Brunton, S. L., Noack, B. and Koumoutsakos, P., "Machine Learning for Fluid Mechanics", Annual Review of Fluid Mechanics, Vol. 52, pp 477-508, 2020 (doi:10.1146/annurev-fluid-010719-060214);
  2. Sun et al., "Resolvent Analysis of Compressible Laminar and Turbulent Cavity Flow", AIAA Journal, Vol .58, No. 3, pp. 1046-1055,(doi:10.251.4/J058663)


KEYWORDS: Artifical Intelligence; Machine Learning; Store Separation; Cavity; Computational Fluid Dynamics; Six-Degree-of-Freedom Trajectories

US Flag An Official Website of the United States Government