APPLICATION OF MACHINE LEARNING METHODS FOR PREDICTING WELL DISTURBANCES

In the process of field exploration, along with regular flooding, a significant part wells is flooded prematurely due to leakage string and outer annulus. an effort intensify flow oil bottom in conditions, specialists often try solve this problem by using various technologies that change reservoir characteristics formation. Any increase pressure exceeds strength rocks compression or tension leads rock deformation (destruction cement stone, creation new cracks). Moreover, repeated operations under pressure, as rule, lead water cut appearance behind-the-casing circulations. For reason, important condition for maintaining their efficient operation timely forecasting such negative phenomena behind-casing cross casing leakage. The purpose work efficiency well interventions workover machine learning algorithms predicting disturbances. Prediction based on methods, regression analysis, identifying outliers data, visualization interactive processing. data allow training model and, its basis, determine presence absence disturbances wells. As result, forecast showed high accuracy Based this, candidate can be selected further work. each specific well, optimal set studies planned, repair isolation addition, course work, scientific technical solutions was developed algorithms. This approach will without stopping it.