基于局域波和混沌的转子系统早期故障诊断

Early fault diagnosis of rotor system based on local wave and chaos

针对转子系统早期微弱故障诊断问题,提出了一种基于局域波分析和混沌相结合的故障诊断新方法.分析了Duffing混沌振子的混沌运动,说明混沌振子的非平衡相变对微弱信号的敏感性和对白噪声的免疫力.可以通过混沌振子由混沌运动到大周期运动的相变识别微弱信号的特征频率成分.由于实际检测信号为多分量信号,若直接输入Duffing振子达不到检测识别目的.为了消除其他成分的干扰,利用局域波分解,任何复杂的信号都可以分解为有限的并且具有不同的基本模式分量,每个分量是单一成分信号,实现了信噪分离.将局域波分量输入所设计的混沌振子,通过混沌振子系统行为由混沌状态变为大周期运动状态,表明检测信号中含有特征成分,实现了利用混沌振子对低信噪比微弱信号的检测识别.对转子系统早期不对中故障信号进行检测结果证明了方法的有效性.

In order to accurately make out early fault diagnosis of rotor system with weak stgnat, a new fault diagnosis method based on local wave analysis associated with chaos oscillator was presented. The chaos motions of Dulling oscillator were analyzed. It is stated that the oscillator phase transform is sensitive to the weak signal and immune against the random noise. The characteristic frequency can be identified by the oscillator phase transform from chaos to large-scale periodic states. As the actual signals involve more than one components, it can＇t be identified by direct input. Local wave method is proposed to avoid component interference, and weak fault signal is divided into finite local wave components with different simple-intrinsic modes, so that signal was separated from noise. The components were inputted into chaos oscillator, and were transformed from chaotic states to large-scale periodic states, so that weak signal can be identified under low signal-noise ratio through distinguishing the characteristic frequency of signal. Fault diagnosis result shows that the method is feasible, and can accurately make out early fault diagnosis of rotor system.