基于压缩感知弱匹配追踪算法的信号特征提取

针对滚动轴承故障诊断中的特征提取问题,提出一种基于压缩感知弱匹配追踪算法的特征提取方法。针对轴承故障信号特征特点构建了一个由傅里叶字典和冲击时频字典组成的联合字典,作为弱匹配追踪算法中的过完备冗余原子库。进而利用改进的简化粒子群寻优算法在联合字典原子库中寻找最能匹配轴承故障信号特征的原子，实现故障信号的快速高效稀疏分解。在信号重构阶段提出了一种改进的阈值降噪策略,解决了软阈值降噪存在恒定偏差以及硬阈值降噪的不连续问题。对CWRU（Case Western Reserve University）轴承数据中心所提供的标准轴承故障信号和某钢厂滚动轴承实测信号进行了仿真，仿真结果验证了该方法的优越性。

Feature Extraction Based on Compressed Sensing Weak Matching Pursuit Algorithm

A new feature extraction method was proposed to solve the feature extraction problem in rolling bearing fault diagnosis, which was based on compressed sensing weak matching pursuit algorithm. A joint dictionary which was made up of Fourier dictionary and impulse time  frequency atom was built as complete redundant atom library of weak matching pursuit algrithom. And then a improved simplified particle swarm optimization algorithm was presented for searching the atom which can match mostly the bearing fault signal characteristics to realize the sparse decomposition process quickly and efficiently. A improved threshold denoising method was introduced to solve the problem of bad signal consecutiveness and  low degree of approximating the real signals. The simulation results of using standard bearing fault signals of CWRU(Case Western Reserve University) and the measured signals of a certain steel company demonstrate the proposed new method is of advantages.