Abstract:This paper proposes PE-GPT, a custom-tailored large language model uniquely adapted for power converter modulation design. By harnessing in-context learning and specialized tiered physics-informed neural networks, PE-GPT guides users through text-based dialogues, recommending actionable modulation parameters. The effectiveness of PE-GPT is validated through a practical design case involving dual active bridge converters, supported by hardware experimentation. This research underscores the transformative potential of large language models in power converter modulation design, offering enhanced accessibility, explainability, and efficiency, thereby setting a new paradigm in the field.

What problem does this paper attempt to address?

The problem this paper attempts to address is: how to utilize large language models (LLMs) to optimize the modulation design process of power converters, improving design efficiency, accuracy, and interpretability. Specifically, power converters play a crucial role in renewable energy systems, converting renewable energy into the form of electrical energy required by the grid. A good modulation strategy is essential for ensuring the efficient and reliable operation of power converters. However, the traditional modulation strategy design process is complex and time-consuming, requiring electrical engineers to have deep expertise and relying on a large amount of experimental data. These factors limit the practicality and efficiency of modulation design. To address these issues, the paper proposes a customized large language model named PE-GPT, specifically for the modulation design of power converters. PE-GPT is optimized in the following two aspects: 1. **Semantic Customization**: Through in-context learning technology, GPT-4 is trained in a specific domain, enabling it to better understand and handle professional terminology and technical details in the field of power electronics. 2. **Physical Customization**: Integrating two levels of Physics-Informed Neural Networks (PINNs), used for switch-level and converter-level modeling, respectively. This not only improves the accuracy of modulation design but also reduces the amount of training data required. The paper validates the effectiveness of PE-GPT through a practical case—the phase-shift modulation design of a Dual Active Bridge (DAB) converter. Experimental results show that PE-GPT can complete the design process in less than 10 seconds, and compared to traditional methods, its design accuracy is improved by 63.2%. Additionally, hardware experiments also verify the superior performance of the modulation strategy designed by PE-GPT in practical applications.

PE-GPT: A Physics-Informed Interactive Large Language Model for Power Converter Modulation Design

PE-GPT: A New Paradigm for Power Electronics Design

Data-Light Physics-Informed Modeling for the Modulation Optimization of a Dual-Active-Bridge Converter

How Can Large Language Models Help Humans in Design and Manufacturing?

Parameter Estimation of Power Electronic Converters With Physics-Informed Machine Learning

Enabling Large Language Models to Perform Power System Simulations with Previously Unseen Tools: A Case of Daline

Chattronics: using GPTs to assist in the design of data acquisition systems

OMPGPT: A Generative Pre-trained Transformer Model for OpenMP

Temporal Modeling for Power Converters With Physics-in-Architecture Recurrent Neural Network

Advancing Building Energy Modeling with Large Language Models: Exploration and Case Studies

How understanding large language models can inform the use of ChatGPT in physics education

Performance Modeling for Power Converters with Light Gradient Boosting Machine

Assessing Large Language Models in Mechanical Engineering Education: A Study on Mechanics-Focused Conceptual Understanding

Enhancing user experience in large language models through human-centered design: Integrating theoretical insights with an experimental study to meet diverse software learning needs with a single document knowledge base

Can Large Language Models Learn the Physics of Metamaterials? An Empirical Study with ChatGPT

LLM-based Frameworks for Power Engineering from Routine to Novel Tasks

A Generic Modeling Approach for Dual-Active-Bridge Converter Family Via Topology Transferrable Networks

SpeechGPT: Empowering Large Language Models with Intrinsic Cross-Modal Conversational Abilities

ModelGPT: Unleashing LLM's Capabilities for Tailored Model Generation

Large Foundation Models for Power Systems