2026 ASEE Annual Conference & Exposition

This work-in-progress paper presents the design and early implementation of a Digital Twin (DT) framework to support active, inquiry-based learning of photovoltaic (PV) systems within renewable energy education. The DT is directly linked to a laboratory-scale physical PV system through time-synchronized electrical and environmental measurements. This enables students to engage with authentic system behavior rather than idealized data and investigate how PV performance responds to changing operating conditions.

The framework integrates a hybrid DT architecture that combines a physics-based single-diode PV model with an AI-enhanced residual learning layer implemented using a Histogram Gradient Boosting regressor. This approach preserves physical interpretability while improving predictive accuracy by learning systematic discrepancies between modeled and measured performance. Using real sensor data, the hybrid DT achieved substantially improved agreement with measured power output (MAE = 0.0262, RMSE = 0.0525, R² = 0.9337) compared to the physics-only model (MAE = 0.1203, RMSE = 0.1415, R² = 0.5175), providing early technical validation of the modeling approach.

The DT is deployed through an interactive, browser-based dashboard that serves as the primary learning interface. Instructional integration includes pre-lab preparation, guided dashboard-based exploration, and post-lab reflection activities designed to emphasize conceptual reasoning, model comparison, and data literacy rather than algorithmic detail. Preliminary qualitative evidence from post-lab surveys suggests that students were able to distinguish physics-based and hybrid models, reason about model accuracy and limitations, and engage meaningfully with model–data comparisons.

This work contributes a replicable DT-enabled instructional framework that tightly couples real experimental data, hybrid modeling, and inquiry-driven learning activities, advancing digital twin pedagogy for renewable energy education and aligning with the Data Science and Artificial Intelligence (DSAI) Committee’s emphasis on data-informed and AI-enabled engineering learning.