基于实测冲击响应字典稀疏表示的齿轮系统侧隙故障特征提取

摘要

为了从齿轮系统振动信号中提取由于齿侧间隙故障产生的瞬态冲击成分,提出了基于稀疏表示的故障特征提取方法,将特征提取转化为稀疏字典的构造和稀疏表示系数的求解.根据侧隙故障瞬态冲击的响应特性,提出采用与其更相似的系统实测冲击响应构造稀疏字典,继而利用正交匹配追踪算法优化求解稀疏表示向量,获得表征瞬态冲击发生时刻的稀疏系数,并重构故障冲击序列.通过冲击成分的包络解调分析,进一步获得瞬态冲击的调制频率.仿真和实验表明,提出的方法能够有效提取齿轮系统齿侧间隙故障特征.%To extract transient impact components caused by backlash faults of a gear system from its vibration signals,a feature extraction method based on sparse representation was presented.With this method,the feature extraction was converted into constructing a sparse dictionary and solving sparse representation coefficients.According to impact response characteristics of backlash faults,adopting system measured impulse responses being similar to backlash faults was proposed to construct a sparse dictionary.Then the orthogonal matching pursuit algorithm was employed to find the optimal solution to sparse representation vectors.The sparse coefficients were obtained to characterize time instants of transient impacts and reconstruct fault impact time sequences.Furthermore,modulation frequencies of transient impact responses were obtained through the envelope demodulation analysis of impact response components.Simulations and tests showed that the proposed method is effective to extract backlash fault features of gear systems.