The objective of the present PhD study refers to the development of smart grid optimization algorithms, using machine learning and automated control. Recently, the major need to increase renewable energy resources penetration into distribution grids, brings certain challenges regarding balanced and smooth operation of the overall system. The present PhD shall focus on complicated distribution grids, comprising of distribution substations and respective automations (such as capacitor bank controllers, On-Load Tap changer controllers, battery controllers, etc.), distribution grids and respective automations (such as line capacitors, grid reconfigurations, etc.), distributed energy production, and end-consumers having flexibility capabilities, power generation and demand forecasts, etc.