Optimum control for UPQC-embedded source-network-load-storage system using extreme learning machine scheme

• A novel control scheme for UPQC-embedded SNLS system is proposed to improve the penetration of RES and PQ simultaneously. • The multi-objective optimization problem is split as a few single-objective ones based on hierarchical optimization idea. • The ELM is used to train the surrogate model of the multi-objective optimum decision. This study presents an optimum control scheme for the unified power quality conditioner (UPQC)-embedded source-network-load-storage system (SNLSS) to improve the penetration of renewable energy, power quality (PQ) and system utilization rate simultaneously. In the proposed optimum control scheme, four optimization objectives are considered and they include: 1) maximizing the output power of PV arrays to improve the related penetration rate; 2) minimizing the load voltage deviation to improve the voltage quality; 3) maximizing the power factor of the power grid the improve the current quality; and 4) maximizing the utilization rate of proposed system to make full use of the UPQC. By adopting the hierarchical optimization idea, the multi-objective optimization problem is converted into a few single-objective ones. Then, all optimum solutions are solved and trained as an ELM surrogate model, which realizes the quick and precise implementation of optimum control. Time-domain simulation results demonstrate the usefulness of the proposed optimum control scheme.