CONTROL OF INTERVAL CONSTRAINTS FOR REAL-TIME DRILLING

摘要

The present invention relates to a system and method for controlling a drill bit within a wellbore, utilizing Bayesian optimization with interval constraints. A computing device samples observed values for controllable drilling parameters, such as a bit weight (WOB) and a rotational drill bit speed (RPM), and evaluates a selected drill parameter. , such as a penetration rate (ROP), for values observed using an objective function. Interval constraints may be continually learned by the computing device as interval constraints change. Bayesian optimization, subject to range constraints and observed values, can produce an optimized value for the controllable drill parameter to achieve a predicted value for the selected drill parameter. The system can then control the drill bit using the optimized value to achieve the predicted value for the selected drill parameter.