About the intellectual system of prevention of complications in the construction of wells on land and at sea created at OGRI RAS

Alali Walid walidalali@gubkin.ru Gubkin University Moscow
Eremin N.A. Oil and gas research institute RAS Moscow
DOI: 10.24412/2076-6785-2023-1-27-32

The article describes an intelligent system for preventing complications during the construction of wells on land and at sea, created at the OGRI RAS. Intelligent systems for preventing complications when drilling wells are designed to warn the driller in advance about the possibility of violating the normal drilling regime. Intelligent systems to prevent complications during the construction of wells help to increase the productive time and economic efficiency of drilling oil and gas wells. Large volumes of geodata from geological and technological measurement stations during drilling vary from tens to hundreds of terabytes, respectively, on land and at sea. The creation of neural network modeling software components is aimed at identifying hidden patterns in big data sets from geological and technological measurement stations in real time.

Materials and methods
When creating the system, sets of big volumes of data from geological and technological measurement stations in Russia and abroad were used. For each type of complication, procedures were carried out for normalization and labeling of big volumes of geodata. The corrected historical Big geodatasets served as the basis for training neural networks on new geodatasets. An innovative approach to the collection of heterogeneous geodata was used. The main stages of the approach were as follows: the collection of big geodata obtained using sensors built into the drilling rig; formation of simulation data sets using
a drilling simulator; use of geological and geophysical data obtained during geological exploration; creation of test and training sets of geodata of drilling parameters; development of algorithms for cleaning Big sets of geodata using a pre-processing software module from noisy, missed geodata; clustering and visualization of large geodata (hook weight, penetration rate, drilling fluid consumption, torque, etc.). Python and Pandas libraries have become effective tools for building complex statistical models that allow you to efficiently and accurately predict, diagnose, analyze big geodata in order to improve well construction productivity.

artificial intelligence methods, artificial neural networks, well drilling, safe well construction, prevention of complications, geological and technological information, big geodata, intelligent system
Download article