混沌振动信号的盲分离

Blind Separation of Chaotic Vibration Signal

飞机发动机是一个复杂的机械动力系统,存在许多非线性环节,其混沌振动综合反映各种非线性特性,因此,根据混沌振动信号来判断飞机发动机的状态及进行故障诊断就是一种有效的技术方法。由于非线性系统混沌振动信号对初始条件敏感、混沌振动频带较宽且与各种特征频率重叠、实际振动信号含有噪声和各种干扰等成分,混沌振动信号分离就成为依其判断状态和故障的关键。对仿真的混沌混合信号应用FastICA盲分离方法,即使在低信噪比情况下也可有效分离出混沌振动信号以及其他源信号,并将该方法应用于实际的飞机发动机的振动信号分析,分离出实际的混沌振动信号,这使通过混沌振动信号进一步判断发动机的状态成为可能。

The aero - engine is a complex mechanical dynamic system and has many non - linear parts. Its chaos vibration synthetically reflects each non - linear response. Therefore, judging the aero - engine condition and carrying out the fault diagnosis according to the chaos vibration signal is an effective technical method. As chaotic vibration signal is sensitive to initial conditions, chaotic vibration signal has a wide band frequency , it has a variety of over- lapping characteristics, and the actual vibration signals contain ingredients such as various kinds of interference as well as noises, the separation of chaotic vibration signals is the key to judging a state and fault diagnosis. The paper applies Fast ICA to separate the simulation signals of chaotic mixed - signal blindly, even if in low signal - to - noise ratio situation, the Fast ICA can separate the chaos vibration signal effectively from other source signals. As a method applying to the actual aero -engine vibration signal analysis, the method can separate actual chaotic vibration signal from the actual aero - engine vibration signals, which makes the further judgment of the state of engine possible according to the chaos vibration signals.