Neural-Gradient Maximum Power Point Tracking Method for Solar Panel

A popular method used to maximize the power output of solar panels is Maximum Power Point Tracking (MPPT). This work attempted to improve MPPT by utilizing an artificial neural network (ANN) algorithm and gradient perturbation of the duty cycle, dubbed the Neural-Gradient MPPT method. The ANN takes in as input, irradiance, and temperature, whereas its outputs are the power and current at the MPP. We compared our proposed algorithm to the popular traditional MPPT algorithm, Perturb and Observe (P&O). Our proposed MPPT technique has a faster response to abrupt changes in solar irradiance and temperature and less steady-state error than the baseline P&O algorithm.