Artificial Intelligence–based control techniques and Power Electronics

Rizwan Ahmed, Student, RVR & JC College of Engineering

Abstract

This review surveys recent advances in artificial intelligence (AI)–based control techniques and their application to power-electronics systems (power converters, inverters, motor-drive interfaces, EV chargers, and microgrid converters). We present a taxonomy spanning classical AI methods (fuzzy logic, neural networks), modern deep-learning architectures, model-based approaches enhanced with learned components (ML-MPC, NN-MPC), reinforcement learning (RL) and safe-RL, and optimization/metaheuristics for controller tuning. Emphasis is placed on works (2019–2025) that demonstrate closed-loop implementation, hardware acceleration, or rigorous safety/robustness analysis. We summarize benchmarking practices, implementation challenges (real-time computation, dataset scarcity, distribution shift), and open research directions including verifiable safe learning, real-time neural MPC on FPGAs/ASICs, and standardized benchmarks for power-electronics control. Key literature and surveys are cited throughout. 

Keywords: Artificial Intelligence,Power Electronics, AI Techniques

References

Bahrami, M., Review of Machine Learning Techniques for Power Electronics Control and Optimization (arXiv:2310.04699).

Yu, Z., Review on Advanced Model Predictive Control Technologies for High-Power Converters and Industrial Drives, MDPI Electronics (2024).

P. Yu et al., Safe Reinforcement Learning for Power System Control: A Review (arXiv:2407.00681 / 2024).

Alfred, D., Reinforcement Learning-Based Control of a Power Electronic Converter (MDPI, 2024) — applicative RL paper with comparisons to PID.

Survey and reviews on AI techniques in power electronics and related applications (various 2024–2025 reviews).