Forecast of water-cut at wells under design by machine learning methods

Abstract

A large amount of data is generated during the operation of oil fields. Such data can be both data already interpreted by a specialist, or "raw” data obtained directly from the devices, both structured and not structured, or locally structured (that is, allowing for local analysis, but in such form not allowing analyzing in conjunction with other types of data). To obtain from such a set of more informative data that will allow making decisions in the course of field operation, it is necessary to involve specialists from different areas of the oil industry. Therefore, it is possible and necessary to use non-deterministic methods for analyzing the data obtained. The article discusses the use of machine learning methods in the task of determining the initial water-cut based on well logging data.