RESEARCH / PUBLICATION RECORDS
Methods behind
the operating questions.
Published work on agentic AI, manufacturing systems and the connection between evidence, decisions and execution.
Research offers a place to examine the mechanism, the evaluation and the limits.
Explore multi-agent control, policy-governed manufacturing integration and event-centric graph intelligence. Follow the publisher records for the original methods, reported results and publication details.
SELECTED PUBLICATIONS
Read the original work.
Each record links to the publisher and a related guide for readers approaching the topic from practice.
Journal-level citation indicators · Metric year: 2025 · Checked 13 September 2026. Select a metric label to view its source.
2025
PEER-REVIEWED RESEARCH
AI agent-driven process automation for dynamic production efficiency and intelligent equipment integration
The MAS-DME framework combines equipment control and resource allocation agents with reinforcement learning. The study examines dynamic manufacturing coordination, adaptive control and scheduling.
DOI: 10.1007/s10845-025-02706-1
READ WITH THIS QUESTION
How are local equipment decisions connected with wider production objectives?
2026
PEER-REVIEWED RESEARCH
An integrated framework featuring policy-governed agentic AI for closed-loop manufacturing control with multi-source sensor–MES–ERP
The PGAI-CLMC framework integrates sensor, MES and ERP data in a traceable manufacturing workflow. It examines anomaly detection, adaptive monitoring, digital-twin validation and human review using 25,275 manufacturing records.
DOI: 10.1007/s00170-026-17806-2
READ WITH THIS QUESTION
What evidence and policy checks connect a detected event with an authorized response?
2026
PEER-REVIEWED RESEARCH
Event-Centric Graph Intelligence for Industrial Information Integration: Early-Warning and Degradation Awareness in Manufacturing
The K-Event-GNN framework brings heterogeneous manufacturing data into a governed production-event graph. Using 25,275 production records consolidated into 13,926 events, the study examines early-warning forecasting and degradation awareness, while showing the limits of classifying equipment health from isolated events.
DOI: 10.1016/j.jii.2026.101171
READ WITH THIS QUESTION
How does connecting production events change what a model can detect—and what context is still needed before acting on an early warning?
A RESEARCH READING CHECKLIST
Look beyond the headline result.
Use these questions to examine how a reported finding relates to your own setting.
SETTING
Where was the system evaluated?
Examine the production context, data coverage, operating assumptions and the conditions represented in the evaluation.
MECHANISM
What connects the inputs to the action?
Trace the roles of the models, decision logic, enterprise interfaces and human review.
TRANSFER
What would need to be tested again?
Consider differences in process, data, constraints and responsibilities before adapting a framework.