Nowadays, every large enterprise is concerned about reducing CO2 emissions. Along with legislation, management, packaging, and transportation decisions, optimising the operation of automated systems in the industry is important. Overheating processes or large cooling systems of one machine during product assembly may seem minor but at the industry level it is quite significant. Either an optimisation of cooling systems or an intelligent machine control which will prevent heat strokes and allow the transition to passive cooling of the whole system is an important issue for improving machine tools efficiency and contributing therefore to CO2 reduction in the industry sector. This research is a transitional phase from the creation of a control system to solve the problems of resonance in the control of systems with parallel piezo kinematics, which were designed to automate the iterative process of non-circular drilling with a precise shape and the subsequent research on the implementation of smart control to optimise the cooling of industrial machines. The total dynamics of the example system in this research is unknown and consists of the dynamics of electrical converters, piezo kinematics, and mechanics. The control signal of this system is generated by the model of the system state with assumptions and simplifications in combination with machine learning techniques considering the previous errors of the transient characteristics with the possibility of re-drilling without damaging the workpiece and with possibility of further trainings to eliminate the iterative process in general. Algorithms for further training at different resonances with a drilling depth change for cylinders of internal combustion engines are offered. These algorithms are proposed for accurate transmission of the input signal amplitude even in resonant situations, power optimisation, increase the system efficiency, as well as reducing the carbon footprint when used in industry in specific applications.