Dynamic carbon emission factor based interactive control of distribution network by a generalized regression neural network assisted optimization

To reduce the peak-valley difference of power consumption, the distribution system operator (DSO) usually guides the electricity consumers to change their load profiles based on the time of use electricity prices. However, electricity consumers will also pay attention to their carbon emissions except the electricity cost in a carbon emission trading market. It easily causes an adverse influence on the peak-valley difference of power consumption. Therefore, this work constructs a new interactive control between DSO and electricity consumers based on the dynamic carbon emission factor (CEF). The interactive control is a hierarchical optimization, including an upper-layer optimization and a lower-layer optimization. The upper-layer optimization is served for DSO, which aims to minimize the peak-valley difference of power consumption. The low-layer optimization is served for an electric vehicle (EV) aggregator with multiple EV groups, which attempts to minimize the electricity and carbon emission costs. To avoid the frequent and time-consuming interactions between DSO and EV aggregator, a generalized regression neural network is used to generate an accurate surrogate model for the upper-layer optimization. Finally, the proposed technique is verified on an extended IEEE 33-bus system and an extended IEEE 69-bus system with multiple distributed generators and EV groups.