关键词: 新能源汽车；故障诊断；解决方案生成；GPT；自然语言处理

Abstract: With the increasing popularity rate of new energy vehicles in China year by year, the failure rate is also on the rise. In order to ensure the safe and reliable operation of new energy vehicles, this study aims to explore a method of rapid and accurate diagnosis and solution provision based on fault information. In this study, a natural language processing method based on the generative pre-trained transformer model (GPT) is adopted to realize the end-to-end automated processing process from fault information to solution by making full use of its powerful language model. In the experimental process of this study, we constructed two rich corpora to extract relevant information from users' natural language descriptions and scientific papers respectively to fully support the training and testing requirements of the model. Based on the pre-trained model of GPT, this study fine-tuned the model to improve its adaptability and performance. In order to evaluate the generation effect of the model objectively and comprehensively, two evaluation indexes BLEU and ROUGE are used in this study. At the same time, we also conduct a subjective and detailed analysis of the generated solution to demonstrate the advantages and practical effectiveness of the model. According to the analysis of experimental results, the model proposed in this study shows a higher level of semantic consistency, completeness and accuracy when generating solutions, which has significant advantages over the comparison model.

Key words: New energy vehicles; Fault diagnosis; Solution generation; GPT; Natural language processing