An Sand Plug of Fracturing Intelligent Early Warning Model Embedded in Remote Monitoring System

Abstract

In recent years, data mining technology has been widely used in different fields. For petroleum industry, the application of data mining method not only can analyze the data faster and more accurately but also can reduce the defects such as misjudgment and missed judgment caused by relying on artificial monitoring. Fracturing is one of the key techniques for increasing petroleum production. Due to the contingency of the abnormal variation of the fracturing construction curve, there is also a large lag and contingency by using the construction curve for sand block warning. Frequent false alarms and delayed alarms often occur. Therefore, the paper proposes an early-warning method of the sand plug of fracturing based on data mining. First, an early warning model of double logarithmic curve sand plug of fracturing is established, and the time series analysis algorithm is used to predict the oil pressure and casing pressure in the double logarithmic curve sand plug of fracturing risk warning model, so the early warning accuracy is improved. And then the general regression neural network (GRNN) algorithm is designed to optimize the prediction results of the time domain analysis. The improved affinity propagation (AP) clustering algorithm is used to cluster the monitoring data to help to improve the accuracy of the subsequent slope calculation, so as to improve the coincidence rate of the fracturing and sand plug risk. Finally, the risk warning model is applied and analyzed on site, and the validity and accuracy of the model are verified. The model is embedded in the remote monitoring system to realize online remote intelligent monitoring of risks by urban office workers.

Keywords

• AP clustering
• data mining
• early warning
• sand plug of fracturing
• GRNN algorithm