Enhancing PV power forecasting with deep learning and optimizing solar PV project performance with economic viability: A multi-case analysis of 10 MW Masdar project in UAE

Muhammad Faizan Tahir, Anthony Tzes, Muhammad Zain Yousaf

Research output: Contribution to journalArticlepeer-review


Renewable energy systems, particularly solar PV, are increasingly essential in addressing environmental and cost concerns associated with conventional energy sources However, accurately predicting PV generation power remains challenging due to its stochastic and volatile nature, coupled with the task of designing an optimal renewable energy system that balances efficiency and cost-effectiveness. This study proposes the use of bi-directional long short-term memory with Bayesian optimization to forecast PV power generation, conducting a comparative analysis with various machine learning algorithms such as artificial neural networks, Gaussian process regression, support vector machines, and ensemble of regression trees. Furthermore, the system advisor model software is employed for performance and financial analysis of the 10 MW Masdar PV project located in the UAE. The total energy yield computed by system advisor model is validated using PVsyst and PVGIS tools. This work comprises five case studies, with the initial three cases concentrating on the optimal design of solar PV panels. Each case evaluates various factors such as tilt angle, ground coverage ratio, tracker rotation, modules, and number of inverters to determine the most efficient and cost-effective design. Additionally, case 4 investigates the impact of replacing monofacial modules with bifacial modules, while case 5 explores integration of battery with various dispatch strategies like manual and automated dispatch, considering perfect look ahead and one day look behind options. Performance indicators like capacity factor, energy yield and economic indicators like internal rate of return and net present value are assessed to determine energy performance and economic feasibility. The results underscore the efficacy of 2-axis rotation tracking, the advantages of bifacial modules, and positive impact of battery storage on both performance and financial viability. These findings provide crucial insights for solar PV project optimization, highlighting the importance of advanced forecasting techniques and strategic system design in achieving sustainable and economically viable renewable energy solutions in the landscape of UAE.

Original languageEnglish (US)
Article number118549
JournalEnergy Conversion and Management
StatePublished - Jul 1 2024


  • Bayesian optimization
  • Bi-LSTM
  • Performance and economic analysis
  • PV prediction
  • System advisor model
  • United Arab Emirates

ASJC Scopus subject areas

  • Renewable Energy, Sustainability and the Environment
  • Nuclear Energy and Engineering
  • Fuel Technology
  • Energy Engineering and Power Technology


Dive into the research topics of 'Enhancing PV power forecasting with deep learning and optimizing solar PV project performance with economic viability: A multi-case analysis of 10 MW Masdar project in UAE'. Together they form a unique fingerprint.

Cite this