Ge Chen

Ge Chen

Assistant Professor, Great Bay University, Dongguan, China

Research

My recent work centers on AI-driven power and energy system decision-making, including dispatch-oriented modeling, uncertainty-aware optimization, and flexibility assessment based on real-world datasets.

1. AI-Driven Power System Modeling and Optimization

This line of research studies how to build dispatch models directly from measurements, improve their computational efficiency, and understand their theoretical performance. The goal is to support power-system optimization even when physical parameters are incomplete or hard to obtain, by learning the relationship among operating conditions, dispatch actions, objectives, and security constraints from data.

Representative work in this direction includes AI-driven dispatch model construction from complete node-level measurements, intelligent pruning for reducing redundant variables and constraints, self-supervised end-to-end dispatch learning, and theoretical upper-bound analysis for model generalization and optimization loss.

Research direction 1: AI-driven power system modeling and optimization
Research direction 1: AI-driven power system modeling, pruning, end-to-end dispatch learning, and generalization analysis.

Selected papers

2. AI-Driven Uncertainty Modeling and Risk-Aware Dispatch

This direction focuses on how to model uncertainty in renewable generation, load behavior, and operating conditions, and how to defend against the resulting dispatch risk. The core idea is to combine data-driven uncertainty representation with optimization techniques such as robust optimization and chance-constrained programming.

Representative studies include Gaussian-mixture and support-set based uncertainty modeling, risk-aware dispatch under distributional ambiguity, adversarially robust dispatch learning, and fast surrogate modeling for chance-constrained optimization under non-Gaussian uncertainty.

Research direction 2: AI-driven uncertainty modeling and risk-aware dispatch
Research direction 2: uncertainty distribution learning, support-set construction, and AI-enabled risk defense for secure dispatch.

Selected papers

3. Real-Data Vehicle-Grid Interaction and Flexibility Assessment

This direction studies how interpretable AI and real-world operational data can support system-level flexibility evaluation. Beyond model construction, the emphasis here is on understanding decision logic, quantifying flexibility potential in practice, and assessing the societal and engineering value of electrified transportation and real energy systems.

Representative studies include post-hoc interpretability analysis for AI dispatch models, flexibility assessment in practical regional systems, data-driven evaluation of EV charging and shared-fleet value, and application-oriented studies with demonstrated societal impact.

Research direction 3: real-data vehicle-grid interaction and flexibility assessment
Research direction 3: interpretable AI dispatch analysis and large-scale flexibility valuation built on real-world datasets.

Selected papers