摘要
针对基于汽车发动机振动信号处理的故障诊断问题,首先,利用截断矩阵奇异值分解方法对采集信号进行降噪预处理,以获取较为纯净的振动信号;然后,通过希尔伯特-黄变换(HHT)信号处理理论对采集信号进行分解与时频分析,提取出分量信号能量特征与边际谱区域变化特征两种参数作为汽车发动机故障诊断与识别的依据,并对比分析不同故障状态下的特征融和结果;最后,使用径向基(RBF)神经网络对故障样本特征进行训练,并进行多种实测故障数据的训练与识别。实际故障数据处理结果表明,上述特征参数可有效表征故障信号的时频域变化特点,可以作为汽车发动机故障的诊断依据。
To solve the fauh diagnosis problem of automobile engine based on vibration signal processing, the paper firstly pretreats noise reduction of collected signal with truncated singular value decomposition method to obtain purer vibration signal. Then, it conducts decomposition and time frequency analysis on collected signal with signal processing theory of HHT （ Hilbert-Huang transform） and extracts component signals and marginal spectrum region characteristics as the basis of fault diagnosis and identification of automobile engine, and makes comparative analysis on feature fusion result in different fault conditions. Finally, it trains fault samples with RBF （ radial basis function） neural network and carries out training and recognition of many kinds of fault data. The actual fauh data processing result shows that the characteristic parameters can effectively represents the time-frequency domain variation characteristics of fault signal, which can be used as the basis for the fault diagnosis of automobile engine.
作者
韩博
HAN Bo(Information Science and Engineering Institute of Circuits and Systems, Lanzhou University, Lanzhou 730000, China Unit 69019, Urumqi 830017, China)
出处
《军事交通学院学报》
2017年第3期47-52,共6页
Journal of Military Transportation University
