Abstract: The fault monitoring and early warning of large pumps is significantly important to the safety of LNG operation. Traditional monitoring methods mainly rely on vibration alarm, but some faults will not cause vibration over-limit. And once the vibration exceeds the limit, it will usually cause interlock shutdown, which will directly affect production. In this paper, a fault early warning algorithm model based on axis orbit feature identification was constructed for Tangshan LNG seawater pumps. The monitoring signals of existing vibration sensors were used in the model to analyze the distribution features of the pump axis orbit. Base on this, the normal running feature index was constructed. When the axis orbit index deviates, a fault alarm will be issued to the operator to start an inspection process. Practices have proved that this method can make full use of the monitoring data to achieve early identification of seawater pump failures and provide effective support for reducing unplanned outages of seawater pumps.

Key words: seawater pump, axis orbit, feature recognition, fault early warning