基于数字孪生和VMD-CNN-Transformer的永磁同步电机故障诊断研究OA
Research on fault diagnosis of permanent magnet synchronous motors based on digital twin and VMD-CNN-Transformer
永磁同步电机凭借结构紧凑、功率密度大和运行效率高等优势,被广泛应用于新能源汽车、工业制造和航空航天等领域,其运行状态监测与故障诊断的准确性直接影响系统可靠性.针对传统诊断方法依赖人工经验、难以有效处理复杂非线性故障,以及实际工业场景中故障样本不足等问题,提出了一种基于数字孪生和VMD-CNN-Transformer的智能诊断新框架.该框架融合了物理机理建模、数据驱动分析与深度学习技术,可实现高精度故障识别.首先,构建多维度协同的电机数字孪生模型,涵盖物理映射、运行机理和数据驱动等维度,以实现对电机全生命周期运行状态的动态仿真与数据采集.然后,构建VMD-CNN-Transformer混合诊断模型,先通过变分模态分解(variational mode decomposition,VMD)算法对数字孪生数据集进行预处理,以有效分离故障特征与噪声干扰,再结合卷积神经网络(convolutional neural network,CNN)的局部特征提取能力与Transformer的长序列依赖关系捕捉优势,实现多尺度特征提取与故障分类.最后,集成数字孪生模型和VMD-CNN-Transformer模型,并将其部署至Unity3D平台,以实现电机故障的可视化监测与智能诊断.实验结果表明,VMD-CNN-Transformer模型的故障诊断准确率达99.51%,在诊断精度和鲁棒性上显著优于对比模型.研究结果为永磁同步电机故障智能诊断提供了新的技术路径.
Permanent magnet synchronous motors are widely used in fields such as new energy vehicles,industrial manufacturing,and aerospace due to their compact structure,high power density,and high operating efficiency.The accuracy of motor operation status monitoring and fault diagnosis directly affects overall system reliability.To address the problems of traditional diagnostic methods relying on manual experience and having difficulty in effectively handling complex nonlinear faults,as well as the insufficiency of fault samples in actual industrial scenarios,a new intelligent diagnostic framework based on digital twin and VMD-CNN-Transformer is proposed.This framework integrates physical mechanism modeling,data-driven analysis,and deep learning techniques to achieve high-precision fault identification.Firstly,a multi-dimensional collaborative digital twin model of the motor was constructed,encompassing physical mapping,operation mechanism,and data-driven dimensions,to achieve dynamic simulation and data acquisition throughout the motor's full lifecycle operation status.Secondly,a hybrid VMD-CNN-Transformer diagnostic model was developed.The variational mode decomposition(VMD)algorithm was applied to preprocess the digital twin dataset,effectively separating fault features from noise interference,and then the local feature extraction capability of convolutional neural network(CNN)and the long-sequence dependency capturing advantage of Transformer were combined to achieve multi-scale feature extraction and fault classification.Finally,the digital twin model and the VMD-CNN-Transformer model were integrated and deployed on the Unity3D platform,enabling visualized monitoring and intelligent diagnosis of motor faults.Experimental results demonstrated that the VMD-CNN-Transformer model achieved a fault diagnosis accuracy of 99.51%,outperforming comparison models in diagnostic accuracy and robustness.The research results provide a new technical pathway for intelligent fault diagnosis of permanent magnet synchronous motors.
江鹏;石宇强
西南科技大学 制造科学与工程学院,四川 绵阳 621000西南科技大学 制造科学与工程学院,四川 绵阳 621000
信息技术与安全科学
永磁同步电机数字孪生故障诊断变分模态分解卷积神经网络Transformer
permanent magnet synchronous motordigital twinfault diagnosisvariational mode decompositionconvolutional neural networkTransformer
《工程设计学报》 2026 (4)
512-521,10
评论