Weak Mechanical Fault Signal Detection And Feature Extraction Method Based On Duffing Oscillator

Rotating machinery is the key equipment in industrial field and has been widely used in the departments of electric power, petroleum, metallurgy, and aviation. But due to the poor manufacturing and assembly or because the system in high speed and high load condition, it is easy to occur all kinds of problem, if has not been timely diagnosed, it will cause secondary faults, which will cause significant influence to the normal operation of the machine. In the paper, focusing on weak signal detetion method based on double coupled duffing oscillator, and put forward two methods to complete the weak mechanical fault signal feature extraction, they are the combination of the double coupled duffing oscillator with EMD decomposition and the double coupled duffing oscillator stochastic resonance.This paper expounds the definition of chaos, studied the principle of weak signal detection based on duffing oscillator. For low sensitivity of the traditional duffing oscillator detection, this paper proposes a methed that using high-dimensional nonlinear restoring force replace the low-dimensional nonlinear restoring force of the traditional duffing oscillator system, because it has the higher detection sensitivity and higher stability. Due to the higher sensitivity in weak signal detection, it is more suitable for the detection of weak signal.For the noise resistance of duffing oscillator, this paper proposes weak signal detection method based on double coupled duffing oscillator. Double coupled duffing system has better antinoise performance, it can detect weak signal within the strong noise background. This paper studies detection method which the double coupled duffing oscillator to detect all parameters of the weak signal, and the feasibility of this method is verified by simulation.For the detection of weak signal in strong noise environment has a low accuracy, if only rely on the duffing oscillator system for detecting, the detection effect is not very ideal. in order to further improve this situation, this paper puts forward two improved methods: One is combined with the EMD decomposition. Decompose the weak signal into multiple intrinsic mode components(IMF), which contains original weak signal characteristics. Then make use of the double coupled duffing oscillator to detect IMF, Finally reach the purpose of detecting weak signal.This paper proposes a double coupled duffing oscillator combined with stochastic resonance signal feature extraction method, through the synergy of weak signal and noise signal, the system will generate stochastic resonance, which can detect the weak mechanical failure signals. For small parameter detecting limit of the double coupled duffing oscillator stochastic resonance method, however, usually mechanical failure signals has a small frequency, So the introduction of scale transformation method to realize the large frequency signal detection, through the scale transformation, the large frequency will become a small frequency signal which satisfy the requirements of the stochastic resonance frequency, and then after scale recovery,the weak mechanical fault signal will be detected.