The Load Transfer Method Based on Deep Convolutional Neural Networks

With the rapid development of society, there is an increasing reliance on electricity supply across various industries. Users demand higher reliability from distribution networks, leading to a significant increase in the number of network nodes and a growing complexity in structure. After a fault occurs, the uncertainty of topological changes is substantial, making it difficult for traditional load transfer methods to provide high-quality solutions in a short period. To address this challenge, we propose a load transfer control method for distribution networks based on deep convolutional neural networks (DCNNs). Firstly, we design a DCNN architecture suitable for load transfer, utilizing convolutional layers, pooling layers, and fully connected layers to capture the correlations between nodes. Then, historical load data is input into the network to learn the relationships between loads through training. Finally, predictions and optimizations are made for load transfer scenarios to provide high-quality solutions, which is crucial for minimizing power outage losses after faults.