The Prediction Model Of Fault Diagnosis Of Rod Pumping System Based On Big Data

Due to the complexity of downhole working conditions,the failure of the rod pumping system has always been a concern of oil field staff.In the context of the current digital oil field construction,in order to be able to find the rod pumping system in the process of operation in a more timely manner.In order to deal with the fault problem in the.In this paper,the on-site recovery data of digital oilfields is used to establish a fault diagnosis and prediction model for rod pumping systems based on big data mining methods.In this paper,by studying the change rule of the pump power diagram eigenvalue of the rod pumping system to realize the prediction of fault conditions,a method of establishing a prediction model using data mining method of big data technology is proposed.Using the gray matrix algorithm,the invariant moment theory and the Freeman chain code method to extract the eigenvalues of the pump work diagram respectively,using the eigenvalues of the pump work diagram as the predictor variables,and at the same time studying the basic theory and algorithm of the time series method in data mining,Using ARIMA model(autoregressive differential moving average model),LSTM(long-short-term memory network)and gray model,three prediction models are used to establish a prediction model for the fault condition of the rod pumping system through MATLAB.Then use the evaluation indicators to evaluate the prediction of the model,and use gray correlation analysis to diagnose the prediction results of the fault condition.Finally,based on the comparison of the eigenvalues and fault diagnosis,the effects of different combinations of eigenvalues and prediction models on the prediction accuracy are explained respectively,and the prediction models with relatively high accuracy are selected.The results show that these prediction models can better simulate the changing trend of the working conditions of the rod pumping system,and the prediction accuracy based on the LSTM with the Freeman chain code characteristic values as the prediction variables is higher.This paper has conducted a comprehensive study on the fault diagnosis and prediction of the rod pumping system,which has great guiding significance for the practical prediction.