Adaptive Particle Swarm Optimization for Non-Invasive Parasitic Parameter Estimation in DC-DC Boost Converters

The growing demand for high-reliability power electronics, driven by electrified systems, necessitates accurate system modeling that accounts for non-ideal parasitic elements and thermal variations. This work presents a comprehensive framework that employs Adaptive Particle Swarm Optimization (APSO) for the non-invasive estimation of key parasitic parameters—including inductor series resistance, capacitor ESR, and MOSFET on-resistance—in a DC-DC boost converter. We developed an enhanced analytical system model that rigorously incorporates the combined influence of multiple operating points (duty cycles) and temperature-dependent resistance effects, ensuring the estimated parameters are valid across the full operational envelope. The estimation is formulated as a multi-objective optimization problem where APSO minimizes a loss function calculated from the discrepancy between the model's simulated signals and reference signals in both the time and frequency domains. The adaptive nature of the algorithm dynamically adjusts search parameters to promote efficient exploration and precise exploitation of the solution space, successfully mitigating the risk of premature convergence. Simulation results demonstrate the framework's superior accuracy and numerical stability, showing rapid convergence and precise parameter identification. The final model accurately replicates the converter's dynamic behavior, confirmed by the close correlation of phase-plane trajectories and harmonic spectral content across various duty cycles. This methodology provides a robust, generalized foundation for developing accurate digital twin representations and supporting advanced, model-based control strategies.