Abstract: To solve the calculation divergent problems in the application of traditional genetic algorithm stochastic resonance, an adaptive stochastic resonance based on stability constraint was put forward. The stability analysis was conducted to the Langevin equation.Thus the constraint formula of frequency compression ratio R in consideration of input signals was obtained. And this stability constraint was applied in the optimization process, changing the unconstrained optimization problem into constrained optimization problem. The improved Adaptive Stochastic Resonance Algorithm method was applied in the early rotor rubbing fault detection. The analysis results show that the improved method ensures the stability of system output. The frequency compression ratio R is under the constraint value during the optimal process, which avoids the calculation divergent phenomenon. The method realizes the weak fault signals extraction under strong noise conditions. 

Key words: Rotor; Genetic algorithm; Constraint optimization; Stochastic resonance; Fault detection

摘要： 为解决应用传统遗传算法优化的随机共振(Stochastic resonance, SR)方法易出现的计算发散问题,提出一种基于稳定约束的自适应随机共振方法。对求解随机共振的Langevin方程进行了稳定性分析,得到了考虑输入信号的条件下,使系统输出稳定的频率压缩比R的约束公式。将该稳定性条件应用于遗传算法参数的寻优过程,将原来的无约束最优化问题转化为有约束最优化问题。将改进后的自适应随机共振方法应用于转子早期碰摩故障检测,分析结果表明,该方法确保了系统输出的稳定性,寻优过程中的频率压缩比R的取值均在约束值以下,避免了计算发散现象,实现了在强噪声条件下对微弱故障信号的提取。 

关键词: 转子;遗传算法;约束优化;随机共振;故障检测 