Energy Efficiency Optimizing for Port Electrical Logistics Equipment Dispatch with Scanty Data

초록

Energy efficiency optimization for port electrical logistics equipment (PELE) seeks to improve energy benefits by dispatching PELE to alter the spatiotemporal distribution of electric loads, thus affecting the voltage quality, network losses and power consumption of port power grid. However, the port power grid data are inadequate, with limited nodes equipped with measurement devices. Neither model-based nor data-driven methods are applicable for the optimal deployment of PELE. This paper proposes a finer-grained deep graph convolutional online learning algorithm to optimize PELE dispatch with scanty data. The fine-grained structure extends the receptive field of unmeasured nodes to the entire graph by residual connections, thus collecting, propagating, and aggregating information from measured neighboring nodes to compensate the unmeasured nodes, which enables accurately learn the mapping between active/reactive power and voltage at port with scanty data. The effectiveness of the proposed method is tested on IEEE 33-node system with only 39.4% measured nodes and applied to PELE energy optimization using real data from Rizhao Port. © 2015 IEEE.

키워드