Physics-informed learning-based fault-tolerant control for robust wireless power transfer in electric vehicles.

Area of Science:

• Electrical Engineering
• Artificial Intelligence
• Automotive Technology

Background:

• Electric vehicles (EVs) offer a sustainable alternative to internal combustion engine (ICE) vehicles, reducing global carbon emissions.
• Wireless Power Transfer (WPT) in EVs promises extended range, contactless charging, and enhanced safety.
• Complex WPT systems are susceptible to power converter failures and coil misalignment, impacting EV performance and reliability.

Purpose of the Study:

• To develop a robust fault-tolerant control (FTC) system for WPT-based EVs.
• To enhance the resilience of WPT systems against power converter failures and coil misalignment.
• To ensure uninterrupted and stable operation of WPT-based EVs.

Main Methods:

• Implementation of a Physics-informed Neural Network (PINN) to monitor power converter signals.
• Integration of an Active Fault-Tolerant Controller (AFTC) with a Fault Detection and Isolation (FDI) block.
• Real-time signal observation and fault management using the PINN-based observer.

Main Results:

• The proposed framework achieves accurate current tracking with a reduced settling time of 0.22s.
• The system maintains residual signals within thresholds despite parameter variations.
• Steady-state error is minimized to 0.3 A, outperforming existing control methods.

Conclusions:

• The PINN-based FTC ensures the resilience and reliable operation of WPT-based EVs.
• The developed system effectively manages faults, including power converter failure and coil misalignment.
• This approach contributes to the advancement of dependable WPT technology for electric transportation.