Levenberg-Marquardt Neural Network-based Unit Vector Template PWM Control Strategy for Current Harmonics Mitigation in Grid-Tied Photovoltaic Systems

A control technique is proposed to mitigate the unbalanced nonlinear load current harmonics in a grid-tied photovoltaic (PV) system. The approach leverages a unit vector template generator (UVTG) and the Levenberg-Marquardt neural network (LM-NN) algorithm is utilized to improve the power quality of the system after being subjected to the disturbances caused by the load and provide grid synchronization. In this work, the neural network model was created and trained using preprocessed data through the Levenberg-Marquardt algorithm. The LM-NN model is configured with multiple hidden layers and neuron topologies, which allows it to identify complex patterns in the data and generate accurate predictions. The outcomes demonstrate that the UVTG and LM-NN control strategy can reduce harmonic distortion in nonlinear loads, improving power quality and grid stability. The proposed control technique is designed and simulated using MATLAB/SIMULINK environment.