Reciprocating membrane pump´s fault diagnosis based on chaos and wavelet packets

The paper studies the chaos characteristics of different situations in the reciprocating membrane pump´ s typical faults with wavelet packets technology. The relative growth rate of the largest Lyapunove exponent is put forward as a new parameter to identify the development of the wearing clearance and the fluctuation of the tank suction´ s pressure. The research shows different wearing places have different largest Lyapunov exponents. It is also found that the wearing faults´ characteristic frequency bands are high frequency ones, while the pressure´ s are low frequency ones. The results are useful to distinguish the pumps´ different faults. All these prove vibration monitoring is a new way for the pumps´ monitoring.