Intellectualization of the Drilling Process Control System with an Uncertain Workpiece Material

Increasing production productivity through automation is an urgent task of modern science. The introduction of hybrid additive-subtractive production complexes will become only one of the stages of this automation for modern times. Cur- rently, many tasks arising as a result of attempts at industrial application of such complexes are unresolved. Such tasks include the creation of a module control system for machining, which works out the loss of stability because of uncertainty in the mechanical properties of the material caused by the deposit of layers of the latter. The present work is aimed at investigating the possibility of using machine learning methods to adapt control contours during drilling to the uncertainty of the properties of the workpiece material, due to the difficulty of using traditional control methods, due to the complexity of nonlinear laws in the contact zone of the tool with the workpiece. The paper presents a mathematical model of the drilling process; using a series of numerical experiments, the possibility of the model to qualitatively describe the processes in the contact zone is confirmed. The description of the data set process for training machine learning models is given, and the effectiveness of their use for predicting the internal parameters of the system is confirmed. As a result of the performed con- structions, the paper presents a control system that meets the task, the effectiveness of which has been proven by numerical experiment. The presented control system identifies the mode of loss of stability of the object according to the signal from the force-moment sensor between the carrier and the tool and returns to the system the cutting parameters adjusted relative to the data predicted by a bunch of machine learning models for which the stability of the control object is maintained. The practical significance of the obtained results is determined by the effectiveness shown in the work of using machine learning methods in the development of control systems for machining. Further development of such systems can be aimed at solving related problems related to increasing the response time of the control circuit to the loss of stability of the machining process.