Expandable depth and width adaptive dynamic programming for economic smart generation control of smart grids

• Published in: Energy (Oxford) 2021-10, Vol.232, p.120964, Article 120964
• Main Authors: Yin, Linfei, Luo, Shikui, Ma, Chenxiao
• Format: Article
• Language: English
• Subjects: Adaptive control; Adaptive dynamic programming; Algorithms; Dynamic programming; Economic models; Economic smart generation control; Energy sources; Expandable depth and width adaptive dynamic programming; Redundancy; Renewable energy sources; Smart grid; Three-state energy prosumer deep neural networks
• ISSN: 0360-5442; 1873-6785
• DOI: 10.1016/j.energy.2021.120964

The uncertainty of renewable energy sources (RESs) puts pressure on the economic dispatch of smart grids. However, many traditional multiple time-scales dispatching approaches lead to problems such as inverse regulation and cost increase. To optimize the economic dispatch of smart grids, firstly, a three-state energy prosumers (TSEPs) model is proposed to facilitate flexible dispatching of RESs. Secondly, an economic smart generation control (ESGC) framework is designed to replace the traditional multiple time-scales framework. The ESGC allows TSEPs to enter and exit at any time and improves the economy of economic dispatch. Furthermore, an expandable depth and width adaptive dynamic programming (EDP) algorithm is proposed. The EDP can match the entering and exiting characteristics of TSEPs, and has high algorithm accuracy and calculation speed with reducing the network redundancy of the EDP. Finally, compared with compared algorithms, the EDP reduces 69.559% and 61.2% in |Δf| and improves 1.783% and 0.668% in profit under a two-area smart grid and a modified 1888-bus French based smart grid, respectively. The simulation results verify the feasibility and reliability of the EDP for the ESGC of smart grids. •Frequency control and economic dispatch of smart grids are simultaneously considered.•Real-time economic smart generation dispatch and control framework is built.•An expandable depth and width adaptive dynamic programming (EDP) is proposed.•EDP includes an expandable prediction network and an expandable evaluation network.•Three-state energy prosumers of smart grids are controlled by the EDP economically.