Intelligent Smart Power Grid Intrusion Detection System Using Preprocessing and Classification Techniques

In a smart grid, intrusion detection is a key task for ensuring the network’s safety and reliability. In this research paper, we present an innovative technique for data preparation in smart grid intrusion detection that employs MDS (Multi-Dimensional Scaling), Locally Linear Embedding (LLE), and t-distributed Stochastic Neighbour Embedding (t-SNE). The preprocessed data is then utilized to train a CNN (Convolution Neural Network) for classification. The model’s performance is assessed using accuracy, precision and recalls determining its usefulness for identifying intrusions. The t-SNE-based preprocessing increases the CNN’s capacity to distinguish between normal and invasive events, resulting in greater accuracy, precision, and recall scores.