Notifications
Back

RESEARCH GOALS

RESEARCH

GOALS

RESEARCH GOALS

Multimodal Predictive Maintenance Framework

Our research aims to develop an innovative predictive maintenance system for Autonomous Production

Systems (APS) by leveraging multimodal data

integration for diagnostics, predictive, and

prescriptive maintenance, using advanced

AI methodologies.

Year 1

Data Generation

The focus is on developing data generation models through

Physics-Informed Generative Models (PIGM) and Large

Language Models (LLM). These models address

challenges such as data imbalance and scarcity

by generating high-quality synthetic data that

accurately reflects the operational conditions

of production systems. Additionally, we aim

to extract critical health indicators of production

equipment by integrating physical and data-driven models.

Year 2

Multimodal Diagnostics

and Prognostics

We will create diagnostic and prognostic algorithms using

multimodal sparse learning to enable comprehensive monitoring and prediction of system health.

This includes real-time tracking of state changes

(both spatial and operational) and predicting

remaining useful life based on performance models.

Year 3

Optimized Maintenance

System

The final phase focuses on developing an optimized

maintenance decision-making framework using

Graph Neural Networks (MR-GNN) and Large

Language Models (LLMs). This system will

provide root cause explanations of system

states and deliver real-time prescriptive maintenance recommendations. By integrating MR-GNN and LLMs,

the system will improve decision-making for

maintenance scheduling and repair interventions,

ensuring optimal system performance.

RESEARCH GOALS

Multimodal Predictive
Maintenance Framework
Our research aims to develop an innovative predictive maintenance system for Autonomous Production Systems (APS)
by leveraging multimodal data integration for diagnostics, predictive,
and prescriptive maintenance, using advanced AI methodologies.

Year 1

Data Generation

The focus is on developing data generation models through Physics-Informed Generative Models (PIGM) and Large Language Models (LLM). These models address challenges such as data imbalance and scarcity by generating high-quality synthetic data that accurately reflects the operational conditions of production systems.

Additionally, we aim to extract critical health indicators of production equipment by integrating physical and data-driven models.


Year 2

Multimodal Diagnostics

and Prognostics

We will create diagnostic and prognostic algorithms using multimodal sparse learning to enable comprehensive monitoring and prediction of system health.


This includes real-time tracking of state changes (both spatial and operational) and predicting remaining useful life based on performance models.


Year 3

Optimized Maintenance System

The final phase focuses on developing an optimized maintenance decision-making framework using Graph Neural Networks (MR-GNN) and Large Language Models (LLMs). This system will provide root cause explanations of system states and deliver real-time prescriptive maintenance recommendations.

By integrating MR-GNN and LLMs, the system will improve decision-making for maintenance scheduling and repair interventions, ensuring optimal system performance.


GRL

Tel. 02-2220-0473 | Fax. 00-0000-0000
Engineering Center 702-2, Hanyang Univ. 222 Wangsimni-ro, Seongdong-gu, SEOUL, KOREA

Business Registration Number. 206-82-07306