Predictive maintenance of naval equipment using text mining
Abstract
The Navy’s maintenance operations each year are wide-ranging and complex due to the nature and scope of its mission. As a result, the prioritization of resources and prevention of potential future problems is paramount. In this research project, our objective is to utilize a student-driven team overseen by qualified researchers to address this problem. We propose to develop a predictive maintenance mechanism for the Navy to detect potential maintenance concerns before items of equipment become faulty, allowing decision-makers to allocate resources efficiently. To do so, we will use Big Data analytic techniques to analyze large volumes of naval maintenance logs and build a predictive model. As these maintenance logs will be written by naval personnel, we will first utilize text mining techniques to determine “smoke terms,” or words and phrases that can be used to predict which maintenance logs indicate a likely maintenance concern that requires addressing. We will integrate the results of our text mining model with other variables, such as equipment type, time since last servicing, equipment utilization level, and equipment age to create a more holistic predictive maintenance model. The anticipated outcome of our research is to create a unique and viable Big Data technique that the Navy could implement to guide resource allocation for its future maintenance efforts. Utilizing a predictive maintenance approach to efficiently allocate maintenance resources would allow for large financial savings as well as improving naval readiness. Furthermore, as our project will involve substantial involvement of talented University students, the project will allow the students to grow their capabilities and to pursue future Navy-related careers.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Apr 09, 2025
- Source ID
- N001742010026
Entities
People
- David Goldberg
Organizations
- Salk Institute for Biological Studies
- United States Navy