Reservoir parameter prediction method and apparatus based on centroid method for model weight allocation

Abstract

The present application relates to the technical field of oil and gas development, and provides a reservoir parameter prediction method and apparatus based on a centroid method for model weight allocation. The method comprises: acquiring target logging data of a target oil and gas well; using the target logging data of the target oil and gas well as an input, and respectively obtaining a plurality of initial reservoir parameter predicted values of the target oil and gas well by means of a plurality of pre-trained reservoir parameter prediction models; and acquiring a model weight of each reservoir parameter prediction model, and on the basis of the model weight of each reservoir parameter prediction model, performing weighted summation on the obtained initial reservoir parameter predicted values, so as to obtain a target reservoir parameter predicted value of the target oil and gas well. The present application can compensate for the impact of missing nuclear magnetic resonance logging data on reservoir evaluation, can realize quantitative evaluation of the pore size and pore structure quality of a formation, and can overcome the problems of instability and low accuracy in reservoir parameter prediction based on a single machine learning algorithm.

Images