摘要：Accurate prediction of dew point pressure is essential to ensure the efficient development of condensate gas reservoirs.In recent years,big data technologies such as data mining and artificial intelligence have gradually become research hotspots.In this paper,based on intelligent algorithms and machine learning,a dew point pressure prediction model (GWO-LSSVM model) combining gray wolf algorithm (GWO) and least square support vector machine (LSSVM) is proposed.At the same time,based on optimization theory and applied statistics,we propose a non-parametric regression model of condensate gas reservoir dew point pressure determined by alternating conditional expectation transformation (ACE),and obtain calculation formula of dew point pressure of condensate gas reservoir with statistical characteristics.The results show that the prediction accuracy of GWO-LSSVM model and ACE model is higher than that of RBF and BP neural network models.The average absolute relative error (AARD) of the GWO-LSSVM model and the ACE model are 2.13% and 2.82%,respectively.Finally,according to the Leverage method,all data anomalies were detected.This study provides an effective method for dew point pressure prediction of condensate gas reservoirs.
关键词：Condensate gas reservoir;dew point pressure;prediction;alternating conditional expectation transformation;gray wolf algorithm;least square support vector machine
