Poteriailo L. Smart models for drilling performance optimization based on parameterized case databases

Modern technologies make it possible to measure, collect and store ever-increasing amounts of data. Companies are finding that the complexity and volume of data has made big data processing both time-consuming and costly. The problem of improving the efficiency of drilling process management in terms of data interpretation to support decision-making based on them is relevant to this industry and requires the construction of intelligent models. Intelligent drilling control involves processing the current values of the regime parameters and soil properties to obtain the values of the optimal regime parameters and the minimum cost of drilling. As a result of the conducted research is improved the definition of multifactor interdependences of the mode parameters used at decision-making of technological process of drilling of oil and gas wells; for the first time: - the expansion of the precedent method of finding the decision by means of reasoning on the basis of models containing the information on dependences between subject essences, processes, the phenomena is offered and reinforced by simulated cases; - a model has been developed that explicitly takes into account the parameters of the drilling process, which provides the definition of the optimal mode; methods of solving multi-criteria and multi-purpose problems have been further developed through the use of the method of relationship analysis, which determines the importance of goals and increases the efficiency of decision-making. The phases of the cycle of reasoning based on knowledge with a projection on the information cycle of drilling process control are revealed. The scheme of organization of information flows in the design and implementation of intelligent decision making is described, where artificial intelligence is considered as a means to avoid distraction to unnecessary data, creating improved visibility of the process, and thus improving security and efficiency of decision making. The possibility of applying machine learning methods to data analysis tasks related to the drilling process has been identified. The use of a combined approach to adapt the data used for knowledge-based decision-making is proposed. It was found that a critical issue in the process of learning an intelligent system that mimics the drilling process is to determine the patterns of the technological process, the difficulty lies in the limited number of precedents offered to the system of real oil and gas facilities. The possibility of using systems "Drilling simulators" to provide the necessary amount of adequate close to the real data on abnormal situations of the technological process of drilling, characterized by high risk, to model the optimization of drilling characteristics based on parameterized cases. The effectiveness of case identification due to the relationship with the root causes of complications of the drilling process is analyzed. The feasibility of using precedent-based considerations in the construction of a digital oil field and related production environments has been demonstrated and the impact of such an approach on improving asset reliability and avoiding downtime has been identified. The architecture of automation of technological process of drilling with reference to a pyramid of computer-integrated production is presented. In order to search for and identify precedents in historical data in the construction of intelligent models used classification methods, including cluster analysis - the process of segmentation of the original data set into sets (clusters or groups) of homogeneous records that form precedents. For cluster analysis, a metric was selected, which was used to calculate the distances between records. The process of formalizing the problem in modeling using an approach based on considerations of the cases was studied by analyzing information on problematic issues during the construction of wells at the fields of three drilling departments of PJSC "Ukrnafta", to demonstrate the case method used data obtained during drilling works on well №9 Mykulychynska. As a result of a set of research of theoretical material, processing of industrial geological data from wells, computer research, models based on the principles of artificial intelligence are presented, which can be used to build an adequate system that allows forecasting and decision support to operational engineering support. technological process of drilling. The ratio of the values of the corresponding set of technological indicators and the optimal values of the main regime parameters of the drilling process, which can be used by engineering personnel at existing oil and gas enterprises, is determined.