Results of the summarizing process for the arXiv paper: 2303.0004v1

The optimization of electrical circuits is a complex and time-consuming process that requires the expertise of professionals. However, with the increasing sophistication of algorithms, this task can now be performed by AI techniques. In this paper, the authors present a reinforcement learning (RL) approach to optimize an LLC converter at multiple operation points corresponding to different output powers at high efficiency and different switching frequencies. During a training period, the RL agent learns a problem-specific optimization policy that enables optimizations for any objective and boundary condition within a pre-defined range. The results show that the trained RL agent can solve new optimization problems based on LLC converter simulations using Fundamental Harmonic Approximation (FHA) within 50 tuning steps for two operation points with power efficiencies greater than 90%. Therefore, this AI technique provides the potential to augment expert-driven design processes with data-driven strategy extraction in the field of power electronics and beyond. The described use case targets resonant converters using LLC resonant circuits, which are important in galvanically isolated switch-mode power supplies and inductive power transfer systems in industrial, automotive, and avionic applications. The LLC resonant converter uses sinusoidal oscillations between the inductances Lr, Lm and the resonant capacitance Cr offering advantages such as zero-voltage switching in applied semiconductors, size reduction of passive components by selecting high switching frequency, and advantageous utilization of parasitic elements (stray inductances). To provide engineers with a good starting point for detailed design processes without performing CPU-intensive simulations using accurate models of semiconductor devices, they used a fast static approximation algorithm called Fundamental Harmonic Approximation (FHA). This algorithm takes input voltage Vin, equivalent load resistance RL, switching frequency fi, inductions Lr and Lm, resonant capacitance Cr and coupling factor k as input to compute output power and corresponding efficiency at load resistance RL.

• Electrical circuit optimization is a complex and time-consuming process that requires professional expertise.
• AI techniques can now perform this task with increasing sophistication.
• The authors present a reinforcement learning (RL) approach to optimize an LLC converter at multiple operation points.
• During training, the RL agent learns a problem-specific optimization policy that enables optimizations for any objective and boundary condition within a pre-defined range.
• The trained RL agent can solve new optimization problems based on LLC converter simulations using Fundamental Harmonic Approximation (FHA) within 50 tuning steps for two operation points with power efficiencies greater than 90%.
• This AI technique provides the potential to augment expert-driven design processes with data-driven strategy extraction in the field of power electronics and beyond.
• The described use case targets resonant converters using LLC resonant circuits, which are important in galvanically isolated switch-mode power supplies and inductive power transfer systems in industrial, automotive, and avionic applications.
• The LLC resonant converter uses sinusoidal oscillations between the inductances Lr, Lm and the resonant capacitance Cr offering advantages such as zero-voltage switching in applied semiconductors, size reduction of passive components by selecting high switching frequency, and advantageous utilization of parasitic elements (stray inductances).
• Engineers used a fast static approximation algorithm called Fundamental Harmonic Approximation (FHA) to provide them with a good starting point for detailed design processes without performing CPU-intensive simulations using accurate models of semiconductor devices.

Reinforcement Learning for Optimization of LLC Converters

The optimization of electrical circuits is a complex and time-consuming process that requires the expertise of professionals. However, with the increasing sophistication of algorithms, this task can now be performed by AI techniques. In this paper, the authors present a reinforcement learning (RL) approach to optimize an LLC converter at multiple operation points corresponding to different output powers at high efficiency and different switching frequencies.

Background

The use case targets resonant converters using LLC resonant circuits, which are important in galvanically isolated switch-mode power supplies and inductive power transfer systems in industrial, automotive, and avionic applications. The LLC resonant converter uses sinusoidal oscillations between the inductances Lr, Lm and the resonant capacitance Cr offering advantages such as zero-voltage switching in applied semiconductors, size reduction of passive components by selecting high switching frequency, and advantageous utilization of parasitic elements (stray inductances).

Fundamental Harmonic Approximation (FHA)

To provide engineers with a good starting point for detailed design processes without performing CPU-intensive simulations using accurate models of semiconductor devices, they used a fast static approximation algorithm called Fundamental Harmonic Approximation (FHA). This algorithm takes input voltage Vin, equivalent load resistance RL, switching frequency fi , inductions Lr and Lm , resonant capacitance Cr and coupling factor k as input to compute output power and corresponding efficiency at load resistance RL.

.

Reinforcement Learning Approach

During a training period ,the RL agent learns a problem - specific optimization policy that enables optimizations for any objectiveand boundary condition withina pre - defined range . The results show that the trained RL agent can solve new optimization problems based on LLC converter simulations using Fundamental Harmonic Approximation ( FHA ) within 50 tuning steps for two operation points with power efficiencies greater than 90 % . Therefore , this AI technique provides potential to augment expert - driven design processes with data - driven strategy extraction in fieldofpower electronicsand beyond .