Vibration behavior is an important evaluation index in the friction and wear performance test of materials. However, the current method is not comprehensive enough in the data representation of vibration behavior, which easily leads to the loss of features. Therefore, this paper proposes a vibration behavior data representation method based on dual time–frequency analysis and multiple algorithms. Firstly, a test platform for friction and wear of micro-textured AlSiTiN-coated cemented carbide and titanium alloy was built. Subsequently, vibration signals and friction data were acquired. DBO (Dung Beetle Optimizer) was used to optimize VMD (Variational Mode Decomposition) to denoise the original signal. Secondly, the characteristics of time–frequency images of vibration signals STFT (Short-Time Fourier Transform) and CWT (Continuous Wavelet Transform) are explored through image morphological analysis. Based on the results of the analysis, the corresponding features are extracted. MIC (Maximal Information Coefficient) combined with the BP neural network is used to screen the extracted features. Bayesian optimization random forest algorithm is used to verify the feature screening results and provide a friction prediction model. Finally, based on the obtained vibration behavior characteristic values, the effects of preparation process parameters and properties of micro-textured AlSiTiN coatings on vibration were investigated. Optimal parameters for micro-textured AlSiTiN coatings that exhibit superior vibration behavior were identified as follows: h = 2.9 μm, P = 45 W, v = 1700 mm/s, n = 8 time, d = 50 μm, l = 130 μm. It provides a novel method for the digital characterization of vibration behavior and a new idea for improving the surface properties of cemented carbide.

中文翻译：

---

使用 STFT-CWT 方法分析微纹理涂层硬质合金和钛合金中的摩擦和磨损振动信号


振动行为是材料摩擦磨损性能测试中的一个重要评价指标。然而，目前的方法在振动行为的数据表示上不够全面，容易导致特征的丢失。因此，该文提出了一种基于双时频分析和多算法的振动行为数据表示方法。首先，搭建了微织构 AlSiTiN 涂层硬质合金和钛合金的摩擦磨损试验平台;随后，获取了振动信号和摩擦数据。DBO （Dum Beetle Optimizer） 用于优化 VMD （变分模态分解） 以对原始信号进行去噪。其次，通过图像形态学分析，探究振动信号STFT（Short-Time Fourier Transform）和CWT（Continuous Wavelet Transform）的时频图像特性。根据分析结果，提取相应的特征。MIC （Maximal Information Coefficient） 结合 BP 神经网络对提取的特征进行筛选。采用贝叶斯优化随机森林算法对特征筛选结果进行验证，并提供摩擦预测模型。最后，基于获得的振动行为特性值，研究了制备工艺参数和微纹理 AlSiTiN 涂层性能对振动的影响。表现出优异振动行为的微织构 AlSiTiN 涂层的最佳参数确定如下：h = 2.9 μm，P = 45 W，v = 1700 mm/s，n = 8 时间，d = 50 μm，l = 130 μm。它为振动行为的数字表征提供了一种新方法，并为改善硬质合金的表面性能提供了新思路。