Improving Distance Estimation in Wireless Power Transfer Using Symbolic Regression Techniques

Machine learning methodologies are becoming integral for the optimization of Wireless Power Transfer (WPT) systems. This work specifically investigates the use of Symbolic Regression (SR) for the critical task of estimating the distance between coupled coils. By processing both simulated and experimental datasets, our approach demonstrates that SR can derive explicit and physically meaningful mathematical expressions with high accuracy and low prediction error. A key advantage of this technique over conventional 'black-box' models is its interpretability; the resulting analytical formula provides direct insight into the system's underlying dynamics. The findings confirm that SR is a powerful tool, not only for achieving precise distance predictions but also for enhancing the physical understanding required for advanced WPT system design and control.