基于摩擦振动信号的缸套-活塞环磨损状态识别研究

Research on Wear State Identification of Cylinder Liner–Piston Ring Based on Friction Vibration Signal

基于摩擦振动信号的多重分形特征,提出利用多重分形去趋势波动分析(multi-fractal detrended fluctuation analysis,MF–DFA)和支持向量机(support vector machine,SVM)相结合的方法识别缸套—活塞环磨损状态。在Bruker UMT—3摩擦磨损试验机上进行了柴油机气缸套—活塞环的摩擦磨损模拟试验。应用总体模态分解(ensemble empirical mode decomposition,EEMD)方法对测取振动信号进行降噪处理,获取能够表征摩擦副表面接触特性的摩擦振动信号。通过MF–DFA方法计算得到不同磨损状态下摩擦振动信号多重分形谱,由多重分形谱构造特征向量,通过差分进化(differential evolution,DE)算法对SVM参数进行优化,识别不同摩损状态。试验结果表明,正常磨损状态识别准确率为100%,磨合磨损状态和急剧磨损状态识别结果存在轻微混淆。所提方法可以实现缸套—活塞环不同磨损状态的识别。

Based on the multifractal characteristics of friction vibration signals,a method combining multi-fractal detrended fluctuation analysis(MF–DFA)and support vector machine(SVM)was proposed to identify the wear states of cylinder liner and piston ring.The simulation tests were carried out on the Bruker UMT—3 friction and wear testing machine.The vibration signals were denoised by the ensemble empirical mode decomposition(EEMD)method to obtain the friction vibration signal which could characterize the contact characteristics of the friction pairs.The multifractal spectra under different wear states were calculated by MF–DFA method.The feature vectors of different wear states were constructed from the multifractal spectrum,and the parameters of SVM were optimized by differential evolution(DE)algorithm to identify different wear states.The test results showed that the recognition accuracy of normal wear state was 100%,and there was slight confusion between the recognition results of running-in wear state and sharp wear state.Therefore,the proposed method can identify the different wear states of cylinder liner and piston ring.