Loading...
Presentation Type
Presentation
Full Name of Faculty Mentor
Nathan DeBardeleben, Los Alamos National Laboratory; William Jones, Computing Sciences
Major
Computer Science
Presentation Abstract
Our funding sponsor, Los Alamos National Laboratory (LANL), is interested in automatic anomaly detection and classification applied to highly instrumented flight shock and vibrational data for the purpose of providing insight into operational safety. In this work, we leverage recent advancements in machine learning (ML) by applying convolutional neural networks (CNNs) to a publicly available motor vibrational data set that serves as a proxy to the actual LANL data. We successfully train a CNN to classify anomalous motor states using the dataset, and use this model to simulate real-time anomaly detection and event classification. By extending our prior work in this area, we are able to achieve higher model accuracy, precision and recall in a variety of experimental configurations.
Start Date
12-4-2023 12:40 PM
End Date
12-4-2023 1:00 PM
Disciplines
Computer Sciences
Recommended Citation
Mclewee, Grace, "Online Classification of Shock and Vibrational Data Using Convolutional Neural Networks" (2023). Undergraduate Research Competition. 41.
https://digitalcommons.coastal.edu/ugrc/2023/fullconference/41
Online Classification of Shock and Vibrational Data Using Convolutional Neural Networks
Our funding sponsor, Los Alamos National Laboratory (LANL), is interested in automatic anomaly detection and classification applied to highly instrumented flight shock and vibrational data for the purpose of providing insight into operational safety. In this work, we leverage recent advancements in machine learning (ML) by applying convolutional neural networks (CNNs) to a publicly available motor vibrational data set that serves as a proxy to the actual LANL data. We successfully train a CNN to classify anomalous motor states using the dataset, and use this model to simulate real-time anomaly detection and event classification. By extending our prior work in this area, we are able to achieve higher model accuracy, precision and recall in a variety of experimental configurations.