A new combined extreme learning machine variable steepest gradient ascent MPPT for PV system based on optimized PI-FOI cascade controller under uniform and partial shading conditions

Abstract This paper presents a combined extreme learning machine variable steepest gradient ascent (ELMVSGA) maximum power point tracking (MPPT) for a photovoltaic (PV) system. A pioneering attempt is made to develop the ELMVSGA technique which combines the tracking ability of variable steepest gradient ascent and the accuracy of a noniterative artificial neural network. An effort is made for the first time to apply a new proportional integral fractional order integral (PI-FOI) cascade controller in the MPPT. Controller gains and other parameters are optimized using a new metaheuristic salp swarm algorithm (SSA). The proposed ELMVSGA along with SSA optimized PI-FOI cascade controller is simulated using MATLAB/Simulink under various climate conditions with gradual and step change in irradiance and temperature, partial shading condition (PSC). Comprehensive analysis with perturb and observe (P&O) and two recently MPPT algorithms P&O-PI, fractional order incremental conductance technique have been carried out. The result has shown the proposed algorithm provides better tracking ability than others. Furthermore, to access its effectiveness in tracking the proposed controller is tested with real time seasonal climate data conditions. Under PV uncertainties it has shown a better tracking ability to track the maximum power point with greater accuracy.