Quaternion generative adversarial -driven Soc estimation using Tyrannosaurus optimizer for improving hybrid electric vehicles renewably powered energy management

The change in transportation efficiency in the last several years has seen several new engine technologies like EVs and HEVs being more prevalent. Integration of RES wind energy technique, solar photovoltaics, and bio-energies becomes a requirement during the transition from conventional houses to smart houses and from conventional cars for energy efficient electric or hybrid vehicles. The battery of an HEV can only be charged to a certain level and similarly it should not be discharged beyond a certain limit, hence the battery state of charge (SOC) in HEVs has to be supervised by a smart battery management system (BMS). However, the current method requires improvement of the performance of SOC estimate on HEVs. Therefore, development of new SOC estimation method with DL for safe renewable energy management (REM) framework for Hybrid EV (Electric vehicles) is the prime focus of this paper is developed as DLSOC-REM. For more accurate SOC estimate, the proposed approach employs a Quaternion Generative Adversarial Network (QGAN) model. When hyper parameter tuning, the prototype is invigorated employing the Tyrannosaurus optimization algorithm (TOA) to fine-tune SOC estimate outcomes of the QGAN model. Using the QGAN model simplifies the modeling process and gives a correct representation of the battery model's input-output relationship. The work's originality is demonstrated by the design of the TOA-based QGAN model for SOC estimation. The suggested approach shows excellent accuracy with few errors for various drive cycles and temperatures: for US06, the RMSE stabilizes at about 0.05%, the MAE drops to 0.1%, and the MSE reaches 0.0025%.