Enhanced Power Quality Disturbances Classification Using the Levenberg-Marquart Neural Network Algorithm

Power quality disturbances (PQDs) such as voltage sags, swells, and transients significantly impact the stability and reliability of modern electrical systems. These disturbances can cause equipment malfunctions and economic losses, necessitating advanced detection and classification techniques. This paper presents an approach using a Multilayer Perceptron (MLP) neural network, trained with the Levenberg- Marquardt (LM) algorithm, to detect and classify PQDs effectively. A simulated power grid model in MATLAB/Simulink was used to generate a comprehensive dataset, including various PQDs scenarios. The proposed method was evaluated against other training algorithms such as Bayesian Regularization (BR) and Scaled Conjugate Gradient (SCG) using performance metrics like Mean Squared Error (MSE) and regression values. Results demonstrate the MLP's superior accuracy, convergence speed, and robustness, achieving near-perfect classification of disturbances, including simultaneous events. This work provides a scalable, real-time solution for power quality (PQ) monitoring, improving system reliability and operational efficiency. Future direction of this research should integrate adaptive feature extraction techniques, hybrid optimization schemes and real-time implementation strategies to address increasingly complex and dynamic disturbances patterns in modern power systems.