为提高利用振动信号检测汽油机爆震强度的精度,提出了一种基于非线性小波变换的爆震强度识别方法.首先,采用非线性小波变换对振动信号进行分解,提取出爆震特征.然后,在包含爆震特征的小波细节分量上计算一组能够表征爆震强度的时域统计特征参数.最后,将这些统计特征参数输入到人工智能分类器进行爆震强度识别.对某汽油机进行了爆震台架试验,并对提出的方法进行了验证.结果表明:非线性小波变换可以清晰地从缸盖振动信号中检测出微弱的爆震冲击特征;同时,采用支持向量机能够获得更优的爆震强度识别精度和泛化性能.%To increase knock diagnosis accuracy by the vibration signal measured from engine cylinder head,a knock intensity identification method based on nonlinear wavelet transform (NWT) was proposed.First,the vibration signals were decomposed by the NWT to extract the oscillations induced by knocking combustion.Then,a series of parameters with temporal statistical characteristics were calculated from the resultant sub-band wavelet details that contain knocking information.Finally,these temporal statistical parameters were fed into the classifiers to identify the knock conditions.Knocking combustion experiments were carried out on a gasoline engine and the proposed method was tested.Experiment results show that the NWT enables the detection of weak impacts induced by knocking combustion.At the same time,a better knock intensity identification performance can be achieved by using support vector machine (SVM).
展开▼