Network Structure Analysis of Ship Charging and Replacing Power Station Based on Transfer Learning

As the shipping industry progresses towards electrification to meet global carbon neutrality goals, the development of efficient, reliable, and sustainable ship charging and swapping stations becomes paramount. This paper aims to address the critical gap in the structural design of these stations, which is essential for ensuring high reliability and operational efficiency. We present a comprehensive structural design methodology that includes various network configurations—single-ring, dual-source single-ring, 1.5-ring, dual-ring, “double-petal,” and two-main-two-backup modes. Each configuration is analyzed for its potential to enhance the resilience and reliability of power supply. Utilizing deep learning methods based on transfer learning (TL-DL), we evaluate the reliability of these network structures, focusing on key indicators such as average customer interruption time, frequency, and duration. The findings reveal that the two-main-two-backup and “double-petal” wiring modes exhibit superior performance in terms of redundancy and reliability, providing crucial insights for the design and optimization of ship charging and swapping stations.