Abstract
This paper presents an enhanced digital twin framework for a single-stage three-phase grid-connected solar PV system, addressing limitations of existing digital twin approaches that rely on offline calibration or data-driven surrogates with limited interpretability. Prior studies often infer component aging indirectly and struggle to reconstruct internal electrical parameters under switching operation. To overcome these gaps, a unified digital twin architecture integrating real-time parameter estimation, control automation, and health monitoring is developed. A Predictive Drift-Free Maximum Power Point Tracking (PDF-MPPT) algorithm enables fast, oscillation-free power extraction without training data. Current regulation is achieved using a Delay-Compensated Predictive Deadbeat Model Predictive Control (DCP-DBC-MPC) strategy to mitigate delay effects and improve dynamic response. Internal parameter estimation is performed using a modified Coulomb-based Electromagnetic Field Optimization algorithm (C-EFOm), enabling online estimation of interfacing inductors, inverter switching resistances, ripple-filter elements, and DC-link capacitor parameters. To quantify accuracy, a Digital Twin Fidelity Score (DTFS) is introduced for waveform-level comparison between physical and virtual systems. Health monitoring is demonstrated through controlled degradation of inductors, capacitors, and switches using radar plots and waveform analysis. Experimental validation confirms fidelity and parameter tracking.
Funding Information
  • Anusandhan National Research Foundation (ANRF/ARG/2025/000645/ENS)

This publication has 23 references indexed in Scilit: