Research on High-Power Rapid Charge Approach for EV Based on Clustered Multi-node Learning Gaussian Process

In order to satisfy the urgent demand for charging a higher power EV (electric vehicle) within a short time, the Clustering Multi Node Learning Gaussian Process (CMNL-GP) has been used to develop a high-power rapid charge technology in this paper. Core work is to obtain a higher charge current to match the power demand of EVs by clustering the charging piles based on CMNL-GP. To overcome the lack of the circulation effect caused by the inconsistency between the parallel charging piles, the CMNL-GP method has been used to fuse data from multiple charging piles and to learn them simultaneously. It can improve the estimating accuracy of each charging pile by transferring meaningful information among multiple nodes. For minimizing charging energy use and reducing waste, an objective function to obtain within-cluster similarity has been given and used to overcome the stronger correlation and inconsistency between the charging piles. This work is validated by real EV charging stations data in North East China, to demonstrate the effectiveness of the proposed approach. It can give a reference for charging a high-power EV based on optimizing existing piles of a charging station.