You are here

Shipboard Intelligent Machinery Prognostic and Learning Environment (SIMPLE)

Award Information
Agency: Department of Defense
Branch: Navy
Contract: N68335-23-C-0454
Agency Tracking Number: N231-028-0135
Amount: $139,685.00
Phase: Phase I
Program: SBIR
Solicitation Topic Code: N231-028
Solicitation Number: 23.1
Timeline
Solicitation Year: 2023
Award Year: 2023
Award Start Date (Proposal Award Date): 2023-07-25
Award End Date (Contract End Date): 2024-01-24
Small Business Information
1410 Sachem Place Suite 202
Charlottesville, VA 22901-2559
United States
DUNS: 120839477
HUBZone Owned: No
Woman Owned: No
Socially and Economically Disadvantaged: No
Principal Investigator
 Jason Burkholder
 (434) 973-1215
 barron@bainet.com
Business Contact
 B. Eugene Parker
Phone: (434) 973-1215
Email: barron@barronassociates.com
Research Institution
N/A
Abstract

Hull Mechanical & Electrical (HM&E) Machinery Control Systems (MCS) for U.S. Navy combatant craft are highly complex systems often with hundreds of actuators and thousands of sensors spread across dozens of shipboard subsystems.  As described in SBIR topic N231-028, Artificial Intelligence / Machine Learning (AI/ML) applied to HM&E MCS offers the potential for improving robustness and survivability while simultaneously reducing operator cognitive burden and reducing manning requirements. The research team proposes to develop the Shipboard Intelligent Machinery Prognostic and Learning Environment (SIMPLE) using a highly flexible foundational learning technology that is equally adept at leveraging known physics-based models when available and using a purely data-driven approach when physics-based models are not available. Barron Associates, Inc. (BAI) has implemented and refined a suite of machine learning tools dubbed AURA (Algorithms for Uncertainty Representation and Analysis).  BAI’s AURA approach to machine learning is different from some other techniques and well-suited for the HM&E application due the ability to include prior physics-based models of any complexity (when available), explicitly include and quantify uncertainty, and optimize prediction accuracy with sparse training data while learning continually.  In many cases, even simple physics-based models that capture basic dynamics will reduce the amount of initial training data and training time that will be required for SIMPLE.  When physics-based models are not available, purely data-driven AURA will learn efficiently and still offer the primary benefits of the AURA methodology.     The research team will leverage its existing MCS land-based test environment (LBTE) for the Phase I simulation demonstration and also to generate additional training data as needed. The LBTE includes a fully functional MCS interfaced to a dynamic signal-level Plant Model Simulation of HM&E Systems. The ability of the MCS to control and monitor the ship propulsion and electric plants is dependent on the accuracy of the dynamic response analysis and the fidelity with which plant behavior is implemented in the LBTE plant model.  The comprehensive, complex, and accurate plant simulation model and the test platform built around it represent the collective efforts over the past decade to create an LBTE that facilitates successful shipboard integration.  This LBTE includes the ability to easily and efficiently generate and log realistic additional training data for the SIMPLE effort proposed herein. The Phase I work plan is designed to unequivocally establish that the proposed SIMPLE technology will lay a foundation for a highly-automated HM&E that features continuous self-learning and excels, in particular, in the presence of failures and/or damage. Phase I will culminate in a LBTE simulation-based demonstration of multiple realistic scenarios that cover a range of normal and degraded conditions.

* Information listed above is at the time of submission. *

US Flag An Official Website of the United States Government