Intelligent Self-tuning Active EMI Filtering for Electrified Automotive Power Systems Using Reinforcement Learning

The increasing electrification and intelligence in modern automotive systems necessitate robust electromagnetic compatibility, reliability, and adaptability in power electronics. High-frequency switching converters in electric and autonomous vehicles generate electromagnetic interference (EMI), which can adversely affect safety-critical functions, in-vehicle communication, and overall system efficiency, particularly under dynamic operational conditions. Traditional passive EMI filters, while reliable, often present limitations in terms of size, weight, and adaptability, leading to increased volume and energy losses. This research proposes an intelligent self-tuning active EMI filtering methodology specifically designed for electrified automotive power systems, leveraging reinforcement learning (RL). The EMI mitigation challenge is conceptualized as a Markov decision process, enabling an RL agent to continuously adjust

filter parameters in response to varying interference characteristics. To enhance robustness and generalization in complex and non-stationary environments, a variational autoencoder is utilized for efficient state representation, complemented by a noise-based exploration mechanism to improve learning efficiency and prevent suboptimal convergence. The efficacy of this approach is validated through a MATLAB/Simulink co-simulation framework, employing experimentally measured EMI spectra from an automotive electric drive unit.

Message: Hi FE, I need help with “Intelligent Self-tuning Active EMI Filtering for Electrified Automotive Power Systems Using Reinforcement Learning” in “Electronics & Communication Engineering”