DoD’s Joint Artificial Intelligence Center Leverages ML-Powered Data Mastering to Improve Preventative Maintenance and Mission Readiness

Parts nomenclature data ontology serves as the foundational program for preventative maintenance global project roll-out

Image

Overview

The U.S. Department of Defense leads the world in military assets with more than 40,000 armored vehicles, 13,000 aircraft and nearly 500 ships. The maintenance programs to ensure each asset is combat-ready and costs the Defense Department an estimated $71 billion annually. The sustainment for just one program -- the MH 60r Seahawk -- is projected to cost $750 million.

The DoD needs a way to better manage and scale the asset preventative maintenance program by reducing the cost of resource readiness. Turning to the Joint Artificial Intelligence Center (JAIC), which serves as the Department of Defense’s Artificial Intelligence Center providing expertise to help the Department harness the game-changing power of AI. The mission of the JAIC is to transform the DoD by accelerating the delivery and adoption of AI to achieve mission impact at scale.

To help operationally prepare the department for AI, the JAIC integrates technology development with the requisite policies, knowledge, processes and relationships to ensure long-term success and scalability in accordance with the overarching DoD data strategy.

The data strategy outlined 7-Goals also known as “VAULTIS” that must be achieved to become a data-centric DoD:

1. Make Data Visible
– Consumers can locate the needed data.

2. Make Data Accessible
– Consumers can retrieve the data.

3. Make Data Understandable
– Consumers can recognize the content, context and applicability.

4. Make Data Linked
– Consumers can exploit data elements through innate relationships.

5. Make Data Trustworthy
– Consumers can be confident in all aspects of data for decision-making.

6. Make Data Interoperable
– Consumers have a common representation/comprehension of data.

7. Make Data Secure
– Consumer data is protected from unauthorized use/manipulation.

For this project, the JAIC focused on part nomenclature ontology management for improved data context (goal# 3), integration linkage in an ontology (goal #4) and data interoperability (goal #6).

Leveraging Tamr’s data mastering platform, the JAIC integrated machine learning data mastering into the MH60r helicopter predictive maintenance program by disambiguating the parts nomenclature and creating a uniform ontology to be leveraged as a global standardized reference set. This product also enables goals 2 and 5 by bringing historical data to the surface and increasing accessibility using common nomenclature.

The Challenge:

The DoD needed a way to reduce the cost of asset and resource combat-readiness. A major cost factor is the parts and service required as part of the department’s preventative maintenance programs.

The challenge

A term referring to a single component on the MH60r helicopter can range within each of the DoD’s service, code, shop and departments. The lack of nomenclature alignment between the various maintenance environments causes bottlenecks, delays in maintenance activity, supply chain inefficiencies and a lack of weapons systems readiness.

A common parts nomenclature ontology will serve as the core foundational element for the rest of the AI-powered preventative maintenance systems in place.

Image

 

Name: .CASE,TURBINE, ROTOR

NSN: 2840-01-287-1315

Part Number: 6038T92G09

 

T700

Name: TURBINE ROTOR CASE

NSN: 2840-01-332-3113

Part Number: 6055T36G07

 

T701C/D

Tamr’s OUTCOMES:

NCMS selected Tamr for its ability to quickly and easily scale with the needs of the department. Tamr’s data mastering platform uses machine learning guided by subject matter experts rather than rules; a key differentiator as the project scales beyond one service.

The solution

(NCMS) for which Tamr Government Solutions is an important sub-contractor, published this recent status report: NCMS’s Data Ontology Changes Maintenance Paradigm on a Global Scale Announcement.

NCMS selected Tamr for its ability to quickly and easily scale with the needs of the department. Tamr’s data mastering platform uses machine learning guided by subject matter experts rather than rules; a key differentiator as the project scales beyond one service.

Image

The JAIC’s data ontology as depicted in the above graphic enabled the DoD to create a data standard for a common interchange format for all maintenance data, records, requests and test results. The standard format can be expanded to serve as the generic helicopter parts hierarchy (referred to as portable digital twin) that creates a generic outline of maintainable components for the H60 helicopter. Furthermore, the parts variant hierarchy (digital twin) amplifies the power of predictive analytic frameworks and data combination capabilities.

A unified H60 helicopter part ontology aggregates homologous real world parts into a singular part type and merge cross-variant references into a single group to power AI/ML. Tamr’s machine learning flexibly integrates new datasets in a longitudinal process and finds complex and inexact homologies automatically using patented AI & ML technologies.

Image

Leveraging Tamr as the core machine learning platform for the parts ontology framework, the DoD can:

  • Create a force multiplier by maximizing the population size of data and therefore increasing analytical power and insights

  • Enable broader standardization of methodologies, processes, solution development

  • Promote quality consistency and automation for efficiency

  • Enable scalability by decreasing the development time of AI-based solutions

  • Allow reuse across a broader set of systems and components

The Results:

The JAIC delivered a scalable preventive maintenance program with the common nomenclature ontology at its core that allows the department to set the future gold standard for maintenance data, process documentation and much more. 

The Results

The JAIC delivered a plan to the DoD with the goal to reduce the cost of resource readiness for test and evaluation of preventative maintenance. To solve the problem of disparate maintenance, sustainment, and sensor data enterprises, there is a need for a common understanding of terms and meanings. See the results at this public site.

Following the 7-goals known as VAULTIS as established in the DoD’s data strategy, the JAIC delivered a scalable preventive maintenance program with the common nomenclature ontology at its core that allows the department to:

  • Set the future gold standard for maintenance data and process documentation

  • Deliver repeatable and scalable creation of adaptive digital twin

  • Force multiplier for data analysis for improved readiness & cost-avoidance

  • Allow ontology-enabled investigations

In February 2021, a large 60+ person overview of Tamr’s work along with LMI, a federal management advisory firm, showed the progress to date and how the work helps enable the DoD Data Strategy. The JAIC will continue to leverage Tamr as a critical partner in the success of this AI- and ML-powered preventative maintenance project by inviting feedback from DoD PMx stakeholders for iterative improvement of the ontology and expanding to additional weapon systems.

To find out how Tamr can help you better manage, scale, and reduce cost, schedule a meeting now.