Optimized Power Quality Monitoring in Electrical Systems Using Adaptive Recurrent Neural Networks

The reliable and efficient delivery of electrical power is crucial in modern societies, where a wide range of industries, infrastructure, and everyday life depend on a stable and high-quality power supply. However, the increasing complexity and integration of electrical systems, coupled with the growing use of power electronics and non-linear loads, have led to the emergence of various power quality disturbances that have significant impacts on the performance and lifespan of electrical equipment. To overcome these issues this proposes a leveraging adaptive RNN (Recurrent Neural Network) for enhanced detection of power quality disturbances in electrical systems. The methodology begins with comprehensive data collection from various power system sources to capture a wide range of disturbance patterns. Preprocessing techniques, including data cleaning and remove null values, are applied to enhance data quality and ensure accurate feature extraction. Data analysis and exploration then provide insight into underlying trends and characteristics within the dataset, facilitating the selection of relevant features. These features are then split into training and test datasets to build a robust predictive model. An adaptive RNN model is utilized to detect and classify disturbances, leveraging its ability to handle sequential data efficiently. The model’s performance is assessed through python, followed by accuracy of 91% and F1 Score of 0.89 evaluation to validate its reliability. This approach ensures an effective and adaptive solution for real-time PQDS monitoring, thereby enhancing the resilience and reliability of power systems in dynamic operating environments.