基于多稳态随机共振的轴承微弱故障信号检测

针对双稳态随机共振模型无法有效处理调制信号的缺点,提出了一种以包络信号为输入信号的自适应多稳态级联随机共振(adaptive multi-stable cascaded stochastic resonance,简称AMCSR)信号强化方法。首先,对振动信号进行包络解调,依据包络信号分布特点,选用与信号分布相匹配的多稳态随机共振模型;然后,以故障特征频率的频谱幅值为指标,采用蚁群算法自适应地优化随机共振模型参数;最后,以噪声为强化源和驱动信号,通过级联随机共振方法对包络信号中的故障特征频率进行逐级强化,获得故障特征成分的强化信号。对实测轴承振动信号的验证结果表明,该方法能够增强故障特征频率成分,有效地提取被其他频率成分淹没的微弱故障信号。

Bearing Weak Fault Signal Detection Based on Adaptive Multi-stable Stochastic Resonance

In light of the low efficiency of the bistable stochastic resonance model in modulated signal processing, an extraction and detection method for weak signals was proposed based on adaptive multi-stable cascaded stochastic resonance (AMCSR). First, the vibration signal was demodulated, and the multi-stable stochastic resonance model was adopted, according to the distribution feature of the upper and lower envelope signal. Second, the stochastic resonance model parameters were optimized adaptively through the ant colony algorithm, which adopted the fault feature frequency amplitude as the optimization index. Finally, the fault feature frequency of the envelope signal was progressively strengthened through the optimal cascaded stochastic resonance system with noise as the energy source, then the enhanced output signal was obtained. Experimental results of the bearing vibration signal analysis show that, through the AMCSR method, the fault feature frequency components can be strengthened, and the weak fault signal can be effectively extracted from the mixed signal.