Machine learning applications for photovoltaic system optimization in zero green energy buildings

Abstract

In this paper, the energy supply of a zero-energy building with 220 square meters is considered using optimized nanocomposite solar panels with respect to maximum efficiency. An optimized hybrid machine learning method plays a key role in presenting solar panel modeling with over 0.99% accuracy. Predicting the properties of the nanomaterial solar cell in four different seasons is performed by efficient support vector machines (SVM), and k-nearest neighbors (KNN) machine learning algorithms. In addition, the KNN algorithm is optimized by the Particle Swarm Optimization (PSO) method to improve the capabilities of KNN and reveal the best performance criteria for the photovoltaic modeling characteristics. The parameters of the nanocomposite cells are optimized using the proposed novel Multidisciplinary Optimization Method (MDO) to increase the efficiency of the solar panel by up to 170%. Optimization of solar cell performance with nanocomposite material under energy consumption constraints is carried out to propose the best construction of cells with 3 layers. The presented approach as a solution and indicator for the next generation of commercial and residential buildings can increase the potential values of solar cells to at least 70%.

Keywords

• Zero energy buildings
• Machine learning
• Optimization
• Photovoltaic systems
• Solar panel
• Nano-composite material