Using data mining techniques to improve the efficiency of power system measurement data analysis

Efficiently processing power system measurement data is critical for guaranteeing dependable and cost-effective operation in modern electrical grids. However, the massive amounts of data created by power systems can exceed typical analysis methods, resulting in delayed responses and ineffective management. This research seeks to improve the efficiency of power system data analysis by employing data mining techniques to identify trends and discover anomalies in measurement data. This research offers a novel method, Dwarf Mongoose Optimized Multi-Layer Perceptron with Isolation Forest (DMO-MLP-IsoForest) that combines an MLP for classification tasks with an IsoForest for anomaly detection. DMO is used to optimize the hyper parameters of models, assuring great accuracy and efficiency in identifying and classifying anomalies. The dataset is made up of real-time voltage, current, and power data collected from multiple sensors across the grid, which is cleaned and normalized to ensure consistency and accuracy. K-means clustering is used to group similar data points during feature extraction, which aids in the discovery of underlying patterns in the power system's behaviour. Results, accuracy (0.985), precision (0.995), recall (0.974), f1-measure (0.985) and specificity (0.987), show that the proposed DMO-MLP-IsoForest methods significantly increase the speed and accuracy of power system data processing when compared to standard methodologies. Finally, data mining provides an effective technique to optimize the analysis of power system measurement data, hence improving the grid's operating efficiency and dependability.