Modeling Degradation Systems with Latent States: A Framework with State Space Models and Recurrent Neural Network

##plugins.themes.bootstrap3.article.main##

##plugins.themes.bootstrap3.article.sidebar##

Published Oct 28, 2022
Md Tanzin Farhat Ramin Moghaddass

Abstract

To monitor the dynamic behavior of degrading systems over time, a generic hierarchical discrete-time state-space model (SSM) is introduced that can formulate the stochastic evolution of the latent states (discrete, continuous, or hybrid) of a degrading system and dynamic sensors while system is operating. This generic SSM is inspired by Bayesian hierarchical modeling and recurrent neural networks without imposing strong parametric and distributional assumptions and prior of the system dynamics. The proposed framework is supported by experimental results using multiple datasets including NASA C-MAPSS dataset.

How to Cite

Farhat, M. T., & Moghaddass, R. (2022). Modeling Degradation Systems with Latent States: A Framework with State Space Models and Recurrent Neural Network. Annual Conference of the PHM Society, 14(1). Retrieved from https://papers.phmsociety.org/index.php/phmconf/article/view/3401
Abstract 48 |

##plugins.themes.bootstrap3.article.details##

Keywords

degradation model, sensor-driven systems, state-space model, Bayesian network, Recurrent neural network, Reliability, System health monitoring

Section
Doctoral Symposium Summaries