BOSS-D-RBFN: BOosted Salp Swarm optimization based Deep RBFN for MPPT under partial shading condition in photovoltaic systems

PhotoVoltaic (PV) systems play a promising role in renewable energy resources (REs) as it provides environmentally friendly energy without pollution. PV systems under Partial Shading conditions (PSc) can significantly drop off their productivity of power where the arrays are shaded for various reasons, which lead to the output power that has many peaks, so there is a chance for local minima. Maximum Power Point Tracking (MPPT) methods will reduce the loss created over the PSc. Even though many conventional and soft computing approaches are commonly employed for the MPPT problem, conventional approaches show inadequate efficiency owing to fixed step size, and most of the existing soft computing approaches are limited by complicated rules implementation and require reinitialization during changing environmental conditions. So in this work, Deep learning based Radial Basis Function Network (D-RBFN) is proposed for the MPPT method as this neural network does not oscillate nearby the maximum power point region and performs well in nonlinear and rapid changing condition. Also, RBFN is optimized with the proposed BOosted Salp Swarm optimization (BOSS) approach to attaining higher accuracy and convergence speed by removing the random weights. Moreover, this proposed BOSS-D-RBFN method uses a DC–DC boost converter for faster transient response. The proposed method is evaluated against the existing classical MPPT, neural network-based MPPT, and fuzzy logic-based MPPT method to validate the performance. The results from the simulation study proved that our proposed BOSS-D-RBFN performs better compared to other existing methods with respect to global MPPT and MPPT power efficiency.