Year-end Sale % OFF 
Abstract
The problem of predicting the state of an Electric Submersible Pump during operation is considered. Downtime and shortages caused by pump failure lead to losses in oil pro-duction and require time to replace equipment. By predicting the condition of the equipment, it is possible to minimize pump maintenance costs and reduce well downtime. Expert systems and pre-dictive analytics methods are used to analyze the state of systems. The scientific work uses methods that are based on artificial neural networks. Purpose of research. Elaboration of the issues of fore-casting the technical condition of the pump through by using machine-learning models. Materials and methods. Equipment failure forecasting is carried out using time series analysis. The data was obtained from telemetric sensors of the monitoring system installed on an electric submersible pump. The initial data were taken at one-minute intervals. Initial data preprocessing was carried out. The data was cleared of values (peaks) that are clearly got out of normal operation and places where the phase voltage was equal to zero were removed. An artificial neural network with the LSTM neuron type is used to predict time series. Time series forecasting was carried out for five days. Evaluating system parameters over long periods allows you to assess the condition of its compo-nents and prevent equipment failure. Results. The possibilities of neural networks trained on the ba-sis of data from telemetric sensors of the monitoring system for predicting the values of vertical vi-bration of the pump are investigated. The use of a neural network model in the form of LSTM, which has shown good results in the analysis of time series, is justified. It was found that neural net-works capture the trend well within the time series, which indicates the possibility of using it together with the expert system. Conclusion. The proposed methods and models are tested on real data, which confirms the possibility of their use in the development of an intelligent information system for managing the technical condition of an Electric Submersible Pump during operation.
Highlights
The problem of predicting the state of an Electric Submersible Pump during operation is considered
The scientific work uses methods that are based on artificial neural networks
Equipment failure forecasting is carried out using time series analysis
Summary
Forecasting the time series for the 84 values. 6 видно, что нелинейный тренд предсказан достаточно точно, но предсказанные значения вибрации в некоторых точках сильно отличаются от реальных. Forecasting the time series for 3 days. 8. Изменение средней ошибки на временном интервале прогноза Fig. 8. Что процент ошибки довольно высок для точного прогнозирования значений, но при этом нейросети хорошо улавливают тренд внутри временного ряда, что говорит о возможности применения нейросетей совместно с экспертной системой, основанной на знаниях экспертов, представленных в виде набора правил. Time series forecasting using artificial neural networks methodologies: A systematic review / A. Jenkins Time series analysis forecasting and control / E.P. George. Time series forecasting of petroleum production using deep LSTM recurrent networks / A.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callMore From: Bulletin of the South Ural State University. Ser. Computer Technologies, Automatic Control & Radioelectronics
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
