Self-Optimizing Industrial Intelligence

A Multimodal AI Framework for Predictive Maintenance and U.S. Manufacturing Productivity

Authors

  • Chaitanya Nagabhyru 12-1006, Rahmath Nagar 2nd Line, Chilakaluripet, Guntur Pin: 522616, Andhra Pradesh, India

DOI:

https://doi.org/10.66838/sauc.6548

Keywords:

Predictive Maintenance, Multimodal Learning, Concept Drift, Adaptive Learning, Advanced Manufacturing, Fault Detection, Industrial Automation.

Abstract

Unplanned equipment downtime erodes the productivity of U.S. factories, and the machine-learning models built to anticipate it often decay when equipment is upgraded or operating conditions change. This paper presents a multimodal predictive-maintenance framework that fuses machine sensor streams, equipment alarm logs, technician maintenance notes and production data, monitors each data type for drift, and adapts the model when drift is confirmed. A design layer for retrieval-based work-order assistance with expert validation is described but not evaluated. Because plant data cannot be shared, the framework is evaluated on a transparent synthetic fleet of 160 machines over 900 days, in which a retrofit of 40% of machines, a heavier workload and new production mix are introduced at day 450. Fusing all four data types raises average precision for seven-day failure prediction from 0.32 for sensors alone to 0.46 (late fusion) and 0.47 (early fusion) after the change.....

Downloads

Download data is not yet available.

Global Statistics ℹ️

Cumulative totals since publication
7
Views
3
Downloads
10
Total

Downloads

Published

2025-11-10

How to Cite

Chaitanya Nagabhyru. (2025). Self-Optimizing Industrial Intelligence: A Multimodal AI Framework for Predictive Maintenance and U.S. Manufacturing Productivity. Street Art & Urban Creativity, 11(7s), 120–135. https://doi.org/10.66838/sauc.6548

Issue

Section

Research articles