Project Overview
Development of self-adapting models that autonomously integrate sub-system interactions to comprehensively explain the state of a ship at the system level.
Research Objectives
[Obj1]. Machine learning capable of predicting state, fault location, and remaining useful life of a single sub-system. [Obj2]. Multimodal multivariate feature integration across multi sub-systems using probabilistic latent state model. [Obj3]. Data-driven framework developed to be integrated with physics models for each equipment in the future.
Publications
[1] A. Ghavidel, H. Park, S. Kovacic, and A. Sousa-Poza. GMM-LSTM Integration for Predicting Latent State, RUL, and Fault Location in Rotating Machinery. International Journal of Computer Integrated Manufacturing (In Press) [2] F. Javadnejad, H. Park, S. Kovacic, and A. Sousa-Poza, 2025. A Novel BiGMM-HMM Integration Frame-work with Divergence-Based Sensor and Subsystem Analysis for Predictive Maintenance in Naval Ves- sel Propulsion Systems. Discover Oceans 2 (33) [3] A. Ghavidel, H. Park, S. Kovacic, and A. Sousa-Poza. A New Predictive Maintenance Approach: novel integration of GMM-LSTM for prediction of latent state and failure location of Rotating Machinery. 19th Annual System of Systems Engineering Conference, Tacoma, WA, June 23-26, 2024. [4]. F. Javadnejad, H. Park, S. Kovacic, and A. Sousa-Poza. A Novel BiGMM-HMM Framework for Predictive Maintenance in Naval Vessel Propulsion Systems: Capturing Multimodality and Optimizing Performance. 19th Annual System of Systems Engineering Conference, Tacoma, WA, June 23-26, 2024.