Comparative Machine Learning-based Techno-Economic Feasibility Analysis of Hybrid Renewable Energy Systems

This article provides a thorough analysis of the financial and technological impacts of four hybrid renewable energy system (HRES) setup with the goal of supplying Quantanics Techserv Pvt.Ltd, Tamil Nadu, India, with renewable electricity. The investigation emphasizes grid-dependent hybrid energy systems configured optimally to ensure uninterrupted power supply to workplaces. Initially, a techno-economic analysis is conducted using HOMER software to determine the economic feasibility and technical viability of the energy systems. The integrated configuration of solar PV, grid, diesel generator, battery, and converter show the lowest levelized cost of electricity (LCOE) at ₹3.05million per kWh and the lowest overall system performance among all the systems studied, according to the economic analysis ₹4.36 million is the net present cost (NPC). Nevertheless, technical examination reveals that the configuration made up of grid, solar PV, converters, and batteries renders the 29.9% renewable fraction. Following that, five machine learning algorithm like artificial neural network, decision tree, random forest, linear regression and XG-boost assess the different configuration and balance the techno-economic (TE) factors to cross-verify and predict system performance based on evaluation metrics to achieve sustainable and efficient energy usage. This research contributes to the growing field of sustainable energy solutions by integrating advanced modelling and predictive tools. The study's findings reveal that the ANN regression model outperforms other models, showcasing lower root mean square error and higher R2 values in techno-economic assessments of the commercial buildings considered. These results render valuable insights in to the setup and optimal system of HRES for on-grid applications in urban commercial building, considering their technical and economic aspects.