中文摘要：

机械设备故障的发生往往伴随着振动现象，通过对故障振动信号进行有效的分析是机械设备故障诊断的关键。最近提出的稀疏分解算法具有多分辨率、稀疏性和冗余的特点，但是也存在着原子库构造困难和分解算法计算量大的问题，为了更好将稀疏分解算法应用于机械故障诊断中，提出在正交匹配追踪算法的基础上，采用具有良好时频特性的Gabor原子，利用量子遗传算法快速求解多参数全局最优解的优点，从振动信号中快速和准确地提取出故障特征信息。通过数值仿真信号分析证明了所提的方法无论在特征提取的准确性上还是减小计算时间上都优于传统的正交匹配追踪算法，另外在轴承故障诊断实际应用中的实例分析中，相比传统的频谱分析方法更能有效地提取出故障特征信息，有效降低了背景噪声和杂质频率的干扰。

英文摘要：

The occurrence of mechanical equipment fault is often accompanied by vibration phenomenon. Therefore, the effectiveanalysis of fault vibration signals is the key to the mechanical equipment fault diagnosis. Although, the new sparse decomposition algorithm has the advantage of mutil-resolution, sparsity and redundancy. In order to better apply the recentlysparse decomposition algorithm to mechanical fault diagnosis, a new method based on orthogonal matching pursuit algorithm is proposed.The Gabor atom, s is introduced which have a good time-frequency characteristic. Meanwhile, the quantum genetic algorithm is utilized due to that it can quickly get the global optimal solution of multiple parameters for rapidly. Thus, the proposed method can accurately extracting fanlt characteristic information from the vibration signal. It is superior to the traditional orthogonal matching pursuit algorithm on the accuracy and reducing the computation time through analyzingnumerical simulatedsignals. The results of application on bearing fault diagnosis show that it is more effective than traditional spectrum analysis method in extracting fault characteristic information and diminishing influence of background noise and unrelated frequency.