基于深度神经网络实现电力削峰填谷的需求响应模型研究

Rearch on deep neural network-based demand response model for power peak shaving and valley filling

随着环境问题日益突出和“双碳”目标的推进,需求侧响应资源参与负荷调峰和填谷对电力行业发展具有重要意义。针对我国需求侧响应快速发展的情况,首先考虑了发电机(PG)、电网公司和服务厂商(SPs)之间的相互作用;然后利用深度神经网络(DNNs)收集数据,并通过开发精确的需求价格弹性矩阵(PEMD)来模拟客户行为;最后基于该模型预测了未来每小时短期负荷预测以及前一天每小时的电力需求。结果表明,此模型可以减小电网负荷峰谷差,使发电、用电趋于平衡。

As environmental issues become more prominent and the“double carbon”goal is promoted,the participation of demand-side response resources in load peaking and valley filling is of great importance to the development of the power industry.Based on the rapid development of demand-side response in China,this paper first considers the interactions between generators(PG),Power grid companies and service providers(SPs),then collects data using deep neural networks(DNNs)and simulates customer behavior by developing an accurate price elasticity of demand matrix(PEMD).Finally,based on this model,future hourly short-term load forecasts and hourly electricity demand for the previous day are predicted.The results show that the model can reduce the peak-to-valley load differential on the grid and bring the generation and consumption of electricity into balance.