提出一种高速列车万向轴动不平衡车载动态检测的新方法。该方法的核心是:对万向节安装机座的振动信号进行谐波小波包分解,提取前19个小波尺度能量形成特征向量,分别统计动平衡、动不平衡万向轴在不同转速条件下振动信号的特征向量而构成主元分析的训练样本,以累积方差贡献量大于0.8作为主分量个数的确定准则,将特征向量从19维降低到2维,应用第一主元和第二主元联合区分万向轴的动不平衡状态。应用台架试验数据对该方法的有效性进行了测试验证,结果表明:该方法能够有效辨别万向轴的不平衡状态,其准确性高,具有一定的工程适应性。
A new method of detecting dynamic imbalance with cardan shaft is proposed applying wavelet scale energy primary element analysis. The vibration signals of gimbal installed base were decomposed through harmonic wavelet packets to form a feature vector by extracting the first 19 wavelet scale energy. The training samples of the principal component analysis were constructed through respectively counting the feature vector of the vibration signal under the different condition of balance and imbalance. The cumulative variance contribution greater than 0. 8 was as the guidelines to determine the number of principal components, the dimension of the feature vector was reduced from 19 dimensions to 2 dimensions. The first principal component and the second principal component jointly distinguish eardan shaft dynamic imbalance. The effectiveness of way has been proved through the bench test data, with high accuracy and a good engineering applicability.