Research on Power Grid Stability Analysis Method Based on CNN

In the modern society with a high dependence on power systems, the stable operation of the power grid is of utmost significance. Traditional approaches for analyzing power grid stability suffer from several limitations, including high computational complexity, notable sensitivity to data quality, and inadequate representation of dynamic characteristics. To address these issues, this paper proposes a power grid stability analysis method predicated on the Convolutional Neural Network (CNN). In this method, the discrimination of power grid stability is modeled as a binary classification problem. By performing preprocessing operations on the input data, such as outlier cleansing based on the 3 criterion and min-max normalization, the CNN's convolutional layers, pooling layers, and fully connected layers are utilized to extract features and classify the states. The case study is founded on the actual monitoring data of a regional power grid. The results demonstrate that the accuracy of the proposed method on both the training set and the test set is remarkably superior to that of the Fully Connected Neural Network (FNN). This method can offer effective technical support for the safe operation of the power grid.