Centrifugal Pump Bearing Fault Diagnose Based on Time-Frequency Domain Analysis and BP-Neural Network

Abstract Fault refers to the loss of function of equipment during operation. When a fault occurs, the equipment would usually show some surface features, which can be also called fault signals. Thus, the faults of the equipment can be diagnosed by deep analysis. Among the mechanical and electrical products, the centrifugal pump is one of the most typical ones, which has the majority characteristics of equipment in various kinds of ordnance. Through the study of centrifugal pump fault diagnose, the route of diagnose of mechanical and electrical products and the establishment of diagnose system can be clarified. In the study, the composition and function of centrifugal pump is firstly introduced. So is the fault diagnose system. Then the concept of the fault diagnose experiment is given. The data gathered by the diagnose system are processed in time and frequency domain. The BP- neural network is also applied to analyse the pump’s fault which is injected during the experiment. The analysis result reveals that the BP- neural network can effectively estimate the condition of centrifugal pump and distinguish the different fault modes.