摘要
为提升新能源汽车运行过程中的故障诊断能力,本文提出一种基于多传感器信息融合的故障诊断方法。通过采集电池、电机、控制器及车载环境等多源传感器信号,构建统一的特征融合框架,并且针对多源数据维度高、冗余性强的问题,引入改进卷积神经网络(CNN)与长短期记忆网络(LSTM)相结合的深度特征提取模型,实现时序与空间特征的联合学习。同时,采用注意力机制增强关键特征对故障模式识别的贡献度。试验结果表明,所提出的CNN-LSTM-Attention多传感器融合模型在动力电池、驱动电机及控制系统等典型故障工况下的综合诊断准确率达97.2%,精确率、召回率及F1-score分别为96.8%、96.5%和96.6%,模型在SNR=10 dB强噪声环境下仍保持92.2%的识别准确率较单传感器方法显著提高。进一步验证表明,该方法对未知工况具有较好的泛化能力和稳定性,为新能源汽车的安全运行与智能运维提供了技术支撑。
To enhance fault diagnosis capabilities during the operation of new energy vehicles,this paper proposes a fault diagnosis method based on multi-sensor information fusion.By collecting signals from multiple sensors including batteries,motors,controllers,and onboard environments,a unified feature fusion framework is constructed.To address the challenges of high dimensionality and strong redundancy in multi-source data,an improved deep feature extraction model combining Convolutional Neural Networks(CNN)and Long Short-Term Memory(LSTM)networks is introduced,enabling joint learning of temporal and spatial features.Additionally,an attention mechanism is employed to enhance the contribution of key features to fault pattern recognition.Experimental results demonstrate that the proposed CNN-LSTM-Attention multi-sensor fusion model achieves a comprehensive diagnostic accuracy of 97.2%under typical fault conditions involving power batteries,drive motors,and control systems.The precision,recall,and F1-score reach 96.8%,96.5%,and 96.6%,respectively.The model maintains a 92.2%recognition accuracy even under strong noise conditions(SNR=10 dB),significantly outperforming single-sensor approaches.Further validation demonstrates the method’s robust generalization capability and stability for unknown scenarios,providing technical support for the safe operation and intelligent maintenance of new energy vehicles.
作者
凌松
刘进福
蒋金伟
陈华奎
陈吉
LING Song;LIU Jingfu;JIANG Jinwei;CHEN Huakui;CHEN Ji(Changzhou Vocational Institute of Light Industry,Changzhou 213100,China)
出处
《农机使用与维修》
2025年第11期99-106,共8页
Agricultural Machinery Using & Maintenance
基金
江苏省自然科学基金(BK20220241)
江苏省高校哲学社会科学研究一般项目(2023SJYB1348)
江苏省教育科学规划课题(C/2023/02/25)
江苏省高职院校教师访学研修项目(2024TDFX004)
2024年江苏省青蓝工程优秀青年骨干教师项目
第四批常州市领军型创新人才项目(CQ2021079)
2025年江苏省高等教育教改研究课题(2025JGYB033)。
