Multi-objective optimization and decision-making for renewable energy production and enhancement of wind turbine selection in Morocco

Selecting an appropriate wind turbine for a specific location requires the simultaneous evaluation of energy output, cost, and site-specific deployment constraints. This research introduces a comprehensive framework that integrates simulation, multi-criteria decision-making (MCDM), and hybrid deep learning to enhance the selection of turbines for various sites. One year of European Centre for Medium-Range Weather Forecasts Reanalysis version 5 (ERA5) reanalysis data and geographical information from satellite and elevation datasets was used to model the energy output of 40 commercial turbines in 15 Moroccan cities. A weighted scoring procedure was developed to evaluate turbines based on their capacity factor, annual energy production (AEP), capital and operating expenditures (CAPEX/OPEX), topographical constraints, and proximity to infrastructure. Two hybrid models, convolutional neural network-long short-term memory (CNN-LSTM) and convolutional neural network-Transformer (CNN–Transformer), were trained on structured inputs that included meteorological, topography, and spatial data to enhance decision assistance. The CNN–Transformer model outperformed the CNN–LSTM model in predictions and closely aligned with the expert rankings derived from simulations. Constraint-aware filtering and cost-performance trade-offs were employed to enhance the final turbine suggestions. The findings demonstrate the efficacy of integrating physics-based simulation with machine learning for large-scale energy planning at specific locations. This framework supports future upgrades to renewable energy systems, including AI-based digital twins, real-time monitoring, and use across more locations.