Power Generation through Hybrid Sources and Prediction Estimated Power for Future Using Deep Learning Architectures for INDIA

Abstract

Introduction: The power generation and maintaining its quality and also issues related to it in solar grid systems are major challengers and renewable energy is integrated in intermittent nature, these are mainly depending on environment conditions like weather and many other factors. Due to continuous variations of irradiance solar all over day creates problem energy produced through solar panels and same are integrated in solar power systems and installed in electrical networks. Due to continuous fluctuations, the voltages may loss and leads to damages the many appliances

Objectives: The proposed system uses the Renewable Energy Sources (RES), Solar Power Plants (SPP) and wind as power sources, the RES at Distributed level in Smart Grid has been extensively analysed in terms of power generation and estimation of power consumption in hourly basis

Methods: The Deep Learning based Neural Network (DLNN) has been used to calculate of generated power and estimate consumed power in the country. Based on estimated power required, the proposed system can produce a required amount of output power with good quality for reliability for the loads. In order to exact estimation required power, the standard data set has been classified using DL-long-short term memory (LSTM) network and it’s naturally sequential

 Results: Results show significant improvements: An LSTM is part of DLNN that can learn and test dependencies between time steps of sequence data. For efficient prediction of required power generation and estimation of energy that required to be produced with help solar power plant, XGBoot methods has been incorporated which will produce optimal and tuning results.

Conclusions: It introduces a robust day-ahead solar irradiance forecasting system using XGBoost, DLNN, and SDM, demonstrating superior accuracy and efficiency over existing methods. A web application was developed for interactive visualization using real-time data. The system holds promise for future enhancements in dataset quality and applications in power infrastructure maintenance.