Applications of Machine Learning Methods for Positioning in Scanning Electrochemical Microscope

A typical scanning electrochemical microscope (SECM) consists of a three-axis positioning system, usually with stepper motors and potentiostat. Working at a micro and nanoscale, the most consistent problem is vibrations, which an antivibration table can solve. However, this is not a solution for some devices placed in small environments. Therefore, we suggest reducing vibrations by machine learning methods, making SECM more modular. In this paper, we compare and evaluate the use of neural networks within similar systems that have specialized architecture requirements, the use of samples that require a specialized approach to attribute extraction concerning feature fidelity for learning algorithms, control-machine learning combination methods for static and dynamic applications in electrochemical systems as well as application of neural network functions as stand-alone packages. The results were taken for the application of prototyping a tangible and portable electrochemical scanning device system. Potential improvements, such as micro stepping instead of time delay between steps, are considered. The proposed methods in this paper are aimed at countering the disturbances caused by transient processes during actuation and control tuning using non-evasive methods for cheaper solution alternatives.