The paper discusses horizontal completion drilling technology that uses telesystem diagnostic parameters to monitor the condition of the bottom hole BHA in order to increase wellbore penetration speed and reduce critical vibration and shock loads on BHA elements. The technology is based on solving the dependencies of penetration speed on the parameters of the rotary drilling mode and BHA vibrations using artificial neural networks and, in particular, the «random forest» method. During the development of the technology, a justification was given for the failure of the methods of classical mathematics due to the difficulty or impossibility of formalizing part of the influencing factors, as well as the discreteness of the values obtained from the face. An array of events of the drilled well was used for the study, the data of the GTE, the bottomhole telemetry complex and the on-board recorder of the drilling machine are tied to each other in depth and time. Events were sampled to obtain the correct data array for further investigation with similar parameters of drilling modes, the correct operation of the solved dependencies was tested for a part of the same array, and statistically significant results were obtained. A universal model has been built, with the help of which, based on previously received events, it is possible to obtain the necessary parameters of the rotary drilling mode to reduce the risks of BHA failure and increase the drilling speed. The model is selflearning and improves its quality by replenishing the base of reference events with new events. Thus, even if the wellbore trajectory or geological conditions change during drilling, the model remains applicable.