Fault diagnosis of suck rod pumping system via extreme learning machines

Abstract

Fault diagnosis of suck rod pumping system is an important research subject of oil extraction engineering. This paper presents a research using Extreme Learning Machine (ELM), which is a simple and useful pattern recognition method, to handle downhole dynamometer card auto recognition problems in a suck rod pumping system. An ELM associated with a set of reasonable dynamometer card features is constructed to recognize faults of the system automatically. The model we proposed is trained and tested by the real data acquired from Yanchang oil fields, China. Finally, we conclude based on the experiment results that ELM model has excellent generalization performance and is applicable to the automatic fault diagnosis of suck rod pumping system.