The Diagnosis Research of Electric Submersible Pump Based on Neural Network

Abstract

There are many down-hole failures of electric submersible pump which are difficult to diagnose in the process of oil production. And the fault diagnosis has become the focus to study at present. In the oil field production, the diagnosis of electric submersible pump has important significance to assuring the equipment working efficiently and saving production cost. The method of neural network pattern recognition and data acquisition is presented in the paper. What’s more, the software which can distinguish the operation mode and draw the behavior graph and trajectory characteristic graph is developed based on this method. And then study the feature extraction by the method of time series model according to the different current curve on the current cards. Moreover, it can also form a characteristic repository of the current cards and continuously perfect it. The diagnosis range and diagnosis accuracy will be improved greatly by this method, which is an extension of traditional methods. The practice shows that, this technology has very wide application prospect.