EEMD和TFPF联合降噪法在齿轮故障诊断中的应用
Application of Combined TFPF and EEMD Denoising Method in Gear Fault Diagnosis

为了消除噪声对齿轮传动系统故障特征提取的影响，提出了一种基于集成经验模态分解(ensemble empirical mode decomposition，简称EEMD)和时频峰值滤波(time frequency peak filtering，简称TFPF)相结合的降噪方法。针对TFPF算法在窗长的选择方面受到限制的问题，采用了EEMD方法对其进行改进，使得信号在噪声压制和有效信号保真两方面得到权衡;含噪声的信号经过EEMD分解后，得到一系列频率成分从高到低的本征模态函数(intrinsic modef unctions，简称IMFs)，计算出各IMFs间的相关系数，判断需要滤波的IMFs。对不同的IMFs选择不同的窗长进行TFPF滤波，把过滤后的IMFs和剩余的IMFs重构得到最终的降噪信号。用模拟仿真信号和齿轮齿根故障信号对该方法进行验证，可见EEMD+TFPF能有效地去除噪声，成功提取齿根裂纹故障特征。
In order to eliminate the influence of noise on fault feature extraction in gear transmission system, a method based on time-frequency peak filtering (TFPF) and ensemble empirical mode decomposition (EEMD) noise reduction method combining is proposed. In view of the TFPF algorithm being restricted in the window length selection problem, the balance in two aspects of the signal noise suppression and signal fidelity is improved by using the EEMD method. When the noisy signal is decomposed by EEMD, a series of intrinsic mode function (IMFs) is obtained, which is arranged from high to low according to frequency components. Through calculating the correlation coefficient between IMFs, the IMFs needed to be filtered is determined. Then , a different window length is chosen to filter different IMFs by using TFPF. At last, a reconstructive signal can be obtained by combining the filtered IMFs and the residual IMFs. The denoising method is applied to simulation signals and measured vibration signals, and the results show that the EEMD+TFPF method can effectively extract crack fault feature from intensive background noise.