Deep-learning based Buck-boost Converter for Laptop Applications

The growth of power electronics increase rapidly in the recent past compared to other technical work. A converter may be the heart of power electronics. It has applications like PV, home inverters, battery charging, etc. There are a lot of issues related to the converters and one of them is the output DC voltage without ripples. In other words, the early steady-state output is the requirement for the converter. As the early steady-state achieves, the converter output is close to the ideal one. In the literature, there were different models used to achieve it. But, unfortunately, the machine-learning technique literature is very few, as the working of machine-learning with power electronics is a newer one. In this paper, we propose a deep-learning-based buck-boost converter model that will achieve the buck level or boost level with early steady-state. The training of the model is achieved by using the PID controller by acquiring the data from the PID controller. The input voltage considered for the simulation is 12V. The output voltage is suitable for laptop charging when the model is working in boost mode and are suitable for the USB port of the laptop when working in buck mode. The simulation results validate our deep-learning model. The results were far better than the results achieved from the PID controller. Existing evaluation measures i.e settling time and overshoot are used to validate the proposed model. The results clearly show the achievement of an early-steady state and getting stability earlier with the proposed model. The problem of output voltage ripples is almost solved (in an ideal case, there are no output ripples and in our case, the ripples are less than the tenth part of one percent ripples. There is a table used to show the promising results compared to the PID controller results.