Control of the productivity and the reservoir energy state by processing and analysis of permanent downhole gauges data using machine learning

The issues of high-quality hydrodynamic studies of wells are relevant at any stage of the development of an oil field. The reliability and reliability of the results depend on the significant technological indicators of the development of deposits, deposits and the parameters of the operation of individual wells. This paper presents a new approach to the analysis of hydrodynamic studies of wells by the method of steady-state sampling, based on cluster analysis. The steady-state sampling method is a significant and most effective tool in analyzing measurements of continuous monitoring systems, which allows you to obtain more reliable data for further use in solving various development tasks. The new clustering-based method improves the quality of analysis and reduces the negative impact of the human factor on the success of data operations. This is achieved by automatically dividing the data into groups or clusters, which allows you to more accurately determine the characteristics of steady-state modes. The article provides a detailed description of the new method, its advantages and possibilities of application in the practice of analyzing hydrodynamic studies of wells based on data from continuous monitoring systems. The obtained results of testing the presented approach to the analysis of large amounts of information in a single research plane allow us to say about the high level of its relevance and the possibility of further improvement of algorithms, which will reduce the level of uncertainty in the implementation of digital diagnostics of well operation. Keywords: well testing; permanent downhole gauges (PDG); telemetry systems; inflow performance relationship (IPR); reservoir pressure; clustering; machine learning methods; multilayer systems.