FPGA-Based Digital Twins for Real-Time Monitoring of Power Converters: A Review

Power converters are critical components in modern power electronics, requiring robust real-time control to maintain stability under disturbances and random faults. Conventional cloud-based digital twins (DTs) offer advanced monitoring and analysis but face challenges such as high latency, privacy risks, and potential data loss due to large data transfers between the physical asset and the cloud. This review paper explores an edge-based alternative by implementing digital twins directly on field-programmable gate arrays (FPGAs) for real-time condition monitoring of power converters. The approach leverages FPGA’s inherent parallelism, low latency, and enhanced data security to achieve effective monitoring and fault detection. Two case studies—a flyback converter and a DC-DC boost converter—illustrate the methodology. For the flyback converter, MATLAB/Simulink models were converted to HDL and deployed on FPGA to integrate PI control, PWM generation, ADC interfacing, and DT error calculation. In the boost converter, state-space modeling and discretization enabled real-time digital model execution on FPGA, while particle swarm optimization (PSO) was used to estimate health-related parameters. Comparative analysis between the DT and physical converters demonstrated close behavioral matching, low-latency performance, and the capability to detect sensor faults through error signal monitoring. The results confirm the feasibility of FPGA-based embedded DTs as a tool for smart, privacy-preserving monitoring and event detection in power electronics. However, the current implementations employ simplified models and are limited to basic voltage regulation tests. Future work aims to enhance predictive maintenance, expand diagnostic capabilities, integrate machine learning, and apply the framework to more complex converter systems. This review highlights FPGA-based DTs as a promising pathway toward intelligent, reliable, and secure power converter monitoring.