基于RSIFICA的行星齿轮箱故障诊断方法

Fault diagnosis method of planetary gear box based on RSIFICA

为了对行星齿轮箱进行故障检测和诊断,提出了一种基于共振稀疏快速独立分量的分析方法(RSIFICA).该方法首先采用共振稀疏分解对信号进行降维预处理,进行二次共振稀疏分解,构造虚拟通道增加传感器通道数目,同时引入牛顿-辛普森公式对快速独立分量分析方法进行改进.该方法减少包含瞬态冲击的宽带信号的影响,实现信号中振源信号数目的降维.同时,二次分解增加输入FastICA的通道数,解决了独立分量分析方法在提取行星齿轮箱故障特征频率过程中出现欠定盲源和收敛速度缓慢问题,同时提高了FastICA的运算效率.将该方法应用到行星齿轮箱的故障诊断中,包络谱分析结果表明,RSIFICA能准确地提取行星齿轮箱断齿故障特征频率,有效地解决了FastICA的问题,计算效率提高了21.49%.对比实验表明,相比于EMD-FastICA联合方法,RSIFICA能够对齿轮微弱故障特征进行更为有效的提取.

To detect and diagnose the planetary gearbox, a method based on resonance sparse improved fast independent component analysis(RSIFICA) was proposed. Firstly, the resonance sparse signal decomposition(RSSD) was used to reduce the dimensionality of the signal and the Newton-Simpson formula was introduced to improve the fast independent component analysis(FastICA). The method reduces the influence of wideband signals including transient impacts and the number of vibration source signals in the signal. Meanwhile, the secondary decomposition increased the number of input channels of FastICA, and solved the problems of underdetermined blind sources and slow convergence speeds in the process of extracting the characteristic frequency of the planetary gearbox by the independent component analysis method. Thus, the computing efficiency of FastICA was also improved. The method was applied to the fault diagnosis of planetary gearboxes. The analysis results of envelope spectrum show that RSIFICA can accurately extract the characteristic frequency of broken gear faults of planetary gearboxes, and the calculation efficiency of FastICA is improved by 21.49%. Comparative experiments show that RSIFICA can solve FastICA problems for accurately diagnosing planetary gearbox faults.