Real-Time Implementation of Deep Learning Technique in Microcontroller-Based DC-DC Boost Converter- A Design Approach

This article proposes a design methodology to implement an online based deep learning (DL) algorithm in a microcontroller for DC-DC boost converter. This technique may pave a new futuristic dimension to prevent cyber-attacks in DC microgrid (DCMG) systems as well. The presence of constant power loads (CPL), sudden load changes, source or input variations makes the control system complicated and prone to being unstable. Hence, there's a necessity to develop a robust control methodology for dc-dc converters to achieve voltage stabilization with low ripples. This article proposes a model-free and data driven approach which efficiently substitutes the conventional controller methodologies and also eliminates the need for the usage of a current sensor. Active compensation technique (ACT) is designed to overcome the negative impedance effect produced by CPLs. The effectiveness of the proposed scheme is validated by conducting the real-time experiments and its results are explored.