Performance Verification of a Novel Neural Network-Based P&O MPPT Control Method Using EN50530 Dynamic Test Procedure

This paper proposes a novel neural network (NN)-based perturb & observe (P&O) maximum power point tracking (MPPT) control method. The proposed approach introduces a feedforward neural network (FNN)-based PV operating point (PV-OP) estimator, which outputs a score indicating the relative position of the current PV operating point. This score is used within the P&O algorithm to dynamically adjust the perturbation step size, thereby improving tracking performance under rapidly changing irradiance conditions. Unlike conventional NN-based MPPT methods that directly determine the control variable, the proposed method decouples the NN output from the control action, providing only a reference signal. This approach reduces the sensitivity to training uncertainty and enhances control robustness. Moreover, the modular and independent features of the PV-OP estimator can be applied to various MPPT strategies. Irradiance step response tests and EN 50530 dynamic test procedures were conducted using MATLAB/Simulink simulations and hardware-in-the-loop experiments to validate the effectiveness of the proposed MPPT control method.