This study investigates the economic and environmental impacts of solar-optimized EV charging and explores the feasibility of predicting long-term charger behavior from short-term test data. Using standardized testing procedures developed in the "Wallbox-Inspektion" project, two commercial charging systems were analyzed. A Bidirectional Gated Recurrent Unit (BiGRU) model, a specialized Recurrent Neural Network (RNN), was developed to predict five-hour charging behavior based on five-minute operational data. The results show that for systems with consistent control characteristics, long-term performance can be accurately predicted from short-term measurements. However, prediction accuracy decreases for chargers with irregular behavior. This approach offers the potential to significantly reduce testing time and associated costs, supporting faster development and certification of efficient, solar-optimized charging solutions.

This paper was presented at the 38th International Electric Vehicle Symposium and Exhibition (EVS38), Gothenburg, Sweden, 15–18 June 2025, and published in the official EVS38 Proceedings in the category "D: Charging Infrastructure and Grid Integration" (Paper ID: 448, no DOI assigned). This Zenodo record provides an open-access version of the publication.