Collaborate with GE Vernova’s Controls & Electrical Systems teams and MIT partners to design and implement innovative control strategies for variable grid strength and dynamic load demands. Intern 1 (Grid/Controls – load flow, parameter estimation, scenarios) - Grid modeling (in GE PSLF)and studies: Build and validate a power system model; run ≥10 load-flow and dynamic scenarios (varying SCR, renewables 20–80%, AI DC step/ramp loads). Outcome: quantified voltage profiles, congestion, losses, and stability margins. - Parameter estimation: Identify key plant/control parameters (governor, AVR, PSS) from operational/PMU data or simulations with ≤5–10% error (NRMSE ≥0.9). Deliver a calibrated model with uncertainty bounds. - Control strategy prototype: Design and simulate an adaptive/predictive control update that improves frequency/voltage recovery time by ≥20% and limits overshoot to ≤5% across scenarios. Scenario playbook: Produce a controls adaptation playbook for weak-grid and high-ramp AI load events, with trigger conditions, setpoint/limit adjustments, and rollback criteria. Intern 2 (ML/System ID – data-driven modeling, predictive control) - Data pipeline: Curate a clean, labeled dataset (signals, events, load profiles); document data quality metrics (missingness, latency, drift). - System identification/modeling: Develop and validate a predictive model (e.g., state-space/N4SID, ARX, or ML) with ≤5–10% MAPE on key outputs (e.g., bus voltage, frequency, torque) and calibrated time constants. - Predictive control prototype: Implement an MPC or equivalent controller using the learned model; demonstrate closed-loop stability and constraint handling with control latency ≤100 ms in simulation/HIL. - Performance gains: Show ≥15–25% reduction in oscillations or curtailment during AI load ramps and improved stability margin (e.g., damping ratio +10%). Explainability and governance: Provide feature importance/ablation results, model risk/assumptions, and retraining criteria for production handoff. Project Outcomes: Deliver validated models and control prototypes that address electrical and mechanical challenges from renewables and AI data center integration on power plants and grids, with clear evidence of stability/resilience gains. Intern1 Deliverables: Model files and scripts, study report with KPIs, control design rationale, and a reproducible runbook. Intern2 Deliverables: Trained models, evaluation notebooks, controller prototype, and concise documentation. Shared outcomes (both interns): - Communication: Midterm and final presentations to internal stakeholders and MIT collaborators; 8–10 page tech note plus a 1-page executive summary. - Integration: Joint demo where ML model informs scenario/control tuning in the grid model; side-by-side baseline vs. improved results. - Documentation and handoff: Versioned artifacts, how-to guides, and issue/risk log to enable continuity after the internship. - Stakeholder impact: Clear mapping from findings to operational decisions (setpoints, limits, protection coordination) and next-step recommendations.