Prognozowanie mocy fotowoltaicznej z wykorzystaniem modeli głębokiego uczenia się 
i modelowanie zachowań użytkowników pojazdów elektrycznych z wykorzystaniem 
niemieckiego badania mobilności
[Solar PV power forecasting using deep learning models and EV user behavior modelling 
using the German mobility survey]

Vishnu Suresh (KPEE, PWr, Wrocław)


The talk will be focused on two areas. One regarding the forecasting of PV power from 
solar panels in the very short-term forecast horizon of 1-hour ahead using deep 
learning models that are LSTM based. The data for these models are taken from the solar 
panels of the Electrical Engineering Faculty and is a comprehensive dataset containing 
irradiation measurements, temperature, wind speed and output power.

Secondly, I will talk about understanding EV user behavior in Germany. It is a dataset 
of 200 cars where numerous activities of the EV users are modeled based on the German 
mobility survey using emobpy in Python. The dataset features different variables such 
as distance traveled per day, activities carried out in a day such as shopping/working, 
charging schedules. Also, based on this, the total load demand due to charging was 
calculated for a given cluster of 200 cars. The goal of this talk is to further discuss 
which of these variables can be predicted, and to delve into an area of forecasting 
that is not as popular as PV power forecasting.