基于果蝇优化算法的自适应随机共振轴承故障信号检测方法

Adaptive stochastic resonance method for bearing fault detection based on fruit fly optimization algorithm

针对传统自适应随机共振系统只能单参数优化,而基于群智能算法的自适应随机共振系统存在优化算法参数选取困难、收敛速度慢的缺陷,提出了基于果蝇优化算法的自适应随机共振方法。该方法以双稳随机共振系统输出信噪比作为果蝇优化算法的味道浓度,结合二次采样技术,自适应选取随机共振系统的结构参数,实现周期信号的特征增强。数据仿真与轴承内圈故障数据分析表明,该方法简单易行,收敛速度快,能有效的检测特征信号,实现轴承故障诊断。

The traditional adaptive stochastic resonance method can only realize one-parameter optimization, moreover the swarm-aptitude optimization algorithms need to choice appropriate parameters and the convergence speed will slow down with the increase of population.In order to avoid the disadvantages,a new adaptive stochastic resonance method based on the fruit fly optimization algorithm（FOA）was proposed.The output signal to noise ratio of a bi-stable system was taken as a fitness function in FOA,and the parameters were selected adaptively.The analysis of the simulation data and the real fault data of a bearing shows that the new adaptive stochastic resonance method can effectively realize the characteristic signal detection and early fault diagnosis effectively.