Vibration State Forecast Research Of Electrical Submersible Pump Unit Based On Gray Model

Electric submersible pump (ESP) system is one of the main artificial equipments at domestic and overseas.It's characterized by big capacity, high lift, convenient operation and management. ESP high comprehensive failure rate, however, brought heavy loss to oil field. Engineering practice was proved that one of the reasons for ESP failure was due to the excessive vibration.Therefore, monitoring and forecasting of ESP vibration could predict the fault of the unit in advance to nip it in the bud. The advanced management and predictive maintenance may decrease the loss caused by ESP failure and reduce maintenance cost.As a new and high practical theory, grey forecasting theory was widely used for mechanical vibration forecasting.This paper described the gray system theory and the most common used gray forecasting model-GM(1,1) forecasting model (static GM(1,1) forecasting model).However, the GM (1,1) Model requires high stationary of the original data and concave data lines trend which has big gray plane.Considering the above shortcomings, this paper presented moving average smoothing, trend concave transforming, equal dimension dynamic GM (1,1) model in order to improve its forecasting precision and enlarge its application area. Static and dynamic gray forecast model were successfully applied to vibration status forecast for one ESP operating well.The results showed that the precision of two models was more than 98%,forecasting error was less than 5%. It is proved that the two types of gray forecast model were successfully applied to vibration status forecast for electrical submersible pump and the effect of dynamic gray forecast model was better than the GM (1,1) forecast model for it added the new information timely, increased the white grade of gray interzone, and had the traceable feature. In addition, a real-time monitoring system software of ESP vibration on the basis of dynamic grey system theory has been designed and realized in this paper via the development tool of Kingview 6.53 and Matlab7.1.The testing data for a production well showed that the software is effective and applicable to be as a useful tool for production management of ESP wells.