电动机轴承故障信号准确识别仿真

Accurate Recognition Simulation of Motor Bearing Fault Signal

研究电动机轴承故障信号准确识别问题,有利于电机设备的安全稳定运行。由于采集的电机轴承故障信号中故障特征频率往往被噪声淹没难以准确识别,而传统小波去噪方法中小波基函数选择困难、阈值选取不确定,传统EMD去噪方法不能有效保留有用信息,提出一种基于果蝇算法优化阈值的EMD去噪方法,并为了进一步平衡其收敛速度和全局搜索能力,提出一种改进的自适应步长的果蝇算法。提出的方法参照小波阈值去噪方法,结合仿生学的全局优化参数选取算法,以信噪比为目标函数,可以快速搜索到最佳阈值,最终实现良好的去噪效果。对正弦信号、blocks信号以及电机轴承外圈故障模拟信号进行去噪,仿真结果表明,改进果蝇算法优化阈值的EMD去噪方法与其它方法相比效果更优。

The problem of accurate identification of motor bearing fault signal is studied, which is beneficial to the safe and stable operation of motor equipment. As the fault frequency characteristic of the fault signal is difficult to be accurately identified by the noise flooding in the traditional wavelet denoising method, the wavelet basis function is difficult to choose and the threshold selection is uncertain, so the traditional EMD noise reduction method can not effectively keep the useful information. An EMD denoising method based on fruit fly algorithm optimization threshold is proposed. And in order to further balance its convergence speed and global search ability, an improved adaptive step size fruit fly algorithm is proposed. The proposed method is based on the wavelet threshold denoising method, combined with the bionic global optimization parameter selection algorithm, and the SNR is the objective function, which can search the optimal threshold quickly and finally achieve the good denoising effect. The denoising results of the sinusoidal signals, blocks signals and the motor bearing outer ring fault simulation signals show that the EMD denoising method based on improved fruit fly algorithm optimal threshold is better than other methods.