
Digital-Twin and LLM Decision Support for Thickening–Dewatering
An operational workflow linking field measurements, condition recognition, a process digital twin, and controlled LLM tools.
PhD Candidate, Control Science and Engineering
2024–present
Northeastern University
M.Eng., Materials and Chemical Engineering
2021–2024
Anhui University of Technology
B.Eng., Automation
2017–2021
Anhui University of Technology
Northeastern University
Anhui University of Technology
Anhui University of Technology
Recent work across industrial intelligence, time-series modeling, and visual analysis.
A human-AI decision-support architecture integrating feedback-based retrieval, memory-augmented SQL generation, and meta-cognitive tool orchestration.
A differentiable Max-Min reachability objective that redirects training gradients toward connectivity bottlenecks in slender-structure segmentation.
A risk-gated Bayesian adaptation framework that uses open-ended scenario evidence to correct frozen specialist models without retraining.
An offline LLM-guided input factorization that lets each numerical window activate task-aware variable semantics before a standard forecasting backbone.
A pre-inference correction framework that uses measurement semantics to build independent references for locally inconsistent industrial signals.
Meta-contrastive adaptation for multi-step temperature forecasting across variable ladle-preheating tasks.

An operational workflow linking field measurements, condition recognition, a process digital twin, and controlled LLM tools.

A clinician-reviewed workflow that combines oral dermoscopy images, patient reports, and medical knowledge for structured analysis.

A closed-loop workflow connecting furnace imaging, slag-state analysis, bottom-blowing control, and PLC integration.

An end-to-end system for molten-iron level tracking, tapping-state recognition, industrial communication, and digital-twin visualization.