Abstract:  ABSTRACT：   To improve the control standard performance of automatic generation control （AGC） in an area power grid, the massive plug-in electric vehicles are employed for participating in AGC. The evaluation model of real-time up/down regulation capacity of electric vehicle is constructed by satisfying the charging demand of the owners. Based on this, a multi-layer framework of generation command dispatch of AGC is presented for a coordinated regulation between a cluster of electric vehicles and conventional hydro, thermal units. In order to meet the rapid economic allocation of different types of upper units, this paper proposes a novel optimization algorithm of deep interactive teaching-learning （DITL）, in which a single class of the standard teaching-learning-based optimization is extended to multiple classes, while the small world networks is adopted for constructing the interactive networks among different teachers/students, thus the global search ability and local search ability can be enhanced. In the cluster of electric vehicles, the second-layer generation command dispatch of AGC is executed based on the regulation cost coefficients of different local control center, then the bottom-layer generation command dispatch of AGC is achieved according to the charging time margin of each electric vehicle. The simulations of Hainan power grid indicate that the coordinated regulation between a cluster of electric vehicles and conventional hydro and thermal units can be effectively achieved by the proposed upper generation command dispatch, and DITL algorithm can efficiently improve the dynamic control performance of AGC and reduce the regulation cost of the system. 

Key words:   deep interactive teaching-learning, generation command dispatch, electric vehicle, automatic generation control （AGC）