Abstract: With the rapid growth in the ownership of new energy vehicles, the market scale of new energy used vehicles has continued to expand. However, the difficulty in identifying accidentdamaged vehicles has become a core bottleneck restricting the healthy development of the market. Traditional identification methods rely on manual experience, suffering from problems such as low efficiency, strong subjectivity, and inconsistent standards, which make it difficult to meet the needs of market standardization. This paper researches and designs an intelligent accident identification system for new energy used vehicles. By integrating body structure data, battery system data, electronic control system data, and historical maintenance data, the system adopts an improved deep learning algorithm based on convolutional neural network-long short-term memory (CNN-LSTM) to achieve accurate identification of accidents in new energy used vehicles. To verify the system performance, multi-dimensional actual data of 1 000 new energy used vehicles from mainstream brands were collected for model training and testing. The results show that the overall recognition accuracy of the system for accident-damaged vehicles reaches 96.8%, which is 8.2% higher than that of the unimproved CNN-LSTM model; the precision, recall, and F1-score reach 96.2%, 97.1%, and 96.5% respectively. This system provides an efficient and reliable technical solution for accident identification of new energy used vehicles and is of great significance for promoting the standardized development of the market.

Key words: new energy vehicles; used vehicles; accident identification; intelligent identification system; multi-data fusion; CNN-LSTM model

摘要： 随着新能源汽车保有量的快速增长和二手车市场规模持续扩大，事故车鉴定难题已成 为制约市场健康发展的核心瓶颈。传统鉴定方式依赖人工经验，存在效率低、主观性强、标 准不统一等问题，难以满足市场规范化需求。文章研究并设计了一套新能源二手车事故智能 鉴定系统，通过融合车身结构数据、电池系统数据、电控系统数据及历史维保数据，采用改 进的卷积神经网络-长短期记忆网络（CNN-LSTM）深度学习算法，实现对新能源二手车事故的 精准鉴定。为验证系统性能，采集 1 000 辆主流品牌新能源二手车的多维度实际数据进行模 型训练与测试，结果表明，系统对事故车的总体识别准确率达 96.8%，与未改进的 CNN-LSTM 模型相比准确率提升了 8.2%，精确率、召回率、F1 分数也分别达到 96.2%、97.1%和 96.5%。 该研究为新能源二手车事故鉴定提供了高效、可靠的技术方案，对推动市场规范化发展具有 重要意义。

关键词: 新能源汽车；二手车；事故鉴定；智能鉴定系统；多数据融合；CNN-LSTM 模型