Topology identification of distribution network based on feature selection and convolutional neural network

In the context of smart grid and new power system, distributed generations, energy storage systems, electric vehicles and other new power equipment are connected to the distribution network on a large scale, which makes the topology of the distribution network change more frequently. The traditional topology identification methods of the distribution network are limited by measuring equipment, and difficult to apply. Aiming at the difficulty of obtaining real-time topology of distribution network, this paper proposes a topology identification based on feature selection and Convolutional Neural Network. Firstly, XGBoost is used to calculate the feature importance of node voltage amplitude, and then modified according to the correlation analysis results to select topology identification features. This step can be used to reduce the time and space complexity of the model. Secondly, a distribution network topology identification model based on Convolutional Neural Network is constructed. The correlation between the feature data and the topology is learned by using the operation characteristics of convolutional and pooling and the characteristics of the model structure. Then the mapping rules are established. Finally, IEEE33-node distribution network is used to verify the effectiveness of the proposed method in feature selection and topology identification.