Forecasting and techno-economic optimization of residential PV-battery systems: A synergistic approach for system enhancement
Year: 2026
Type: article
Abstract: A grid-connected solar PV system does not fully contribute to the expected advantages of renewable energy. To address this limitation, the study proposed a method for estimating the optimal size of an off-grid solar PV system with BESS by forecasting a load profile utilizing a machine learning algorithm. To train the machine learning models used in this study, including Linear Regression (LR), Polynomial Regression (PR), Artificial Neural Network (ANN) and XGBoost, several weather, calendar and historical data are collected from the field, Dhaka Electric Supply Company (DESCO) and NASA’s POWER database. Based on comparative analysis, the XGBoost model demonstrated the highest forecasting accuracy for both solar generation and residential load demand and was used to determine the optimal sizing of a 150 kW PV system and a 300 kWh battery storage unit. The existing system had a capacity of 14.5 kW, which was found to be inadequate to meet the building’s energy needs. To assess the economic feasibility of the recommended sizing, a simulation was conducted in PVsyst. The results show that the system achieved a levelized cost of energy of 0.07 USD/kWh, significantly lower than the local grid tariff of 0.15 USD/kWh, with a payback period of 8.4 years and a performance ratio (PR) of 0.77. These outcomes underscore the potential of the proposed model as a powerful tool for accurate forecasting and data-driven design of standalone renewable energy systems in Urban residential settings.
Source: Next Materials
Author: Riad Mollik Babu
Institution: University of Asia Pacific
Language: English
Topic: Smart Grid Energy Management
Subfield: Electrical and Electronic Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,280
OpenAlex ID: W7140735583