Distributed stochastic economic dispatch via model predictive control and data-driven scenario generation

Power systems operation has been traditionally addressed by deterministic and centralized approaches because of their low-variation behavior. However, current tendencies have introduced variability and stochasticity as a result of including renewable energy sources, active demand participation, and short-term market clearing. Thereby, operators are looking for utilizing available forecast information to enhance the system operation response to unpredictable changes from the uncertainty sources. This paper considers two distributed techniques that solve the economic dispatch problem in an hourly basis and for the ultra-short term, and a data-driven scenario generation method that reduces uncertainty impacts on operation costs. At first, the hourly and ultra-short term dispatches are presented as stochastic programming problems by relying on model predictive control (MPC), which also address the concern of variability and uncertainty. Second, since ultra-short term dispatch does not optimize the social benefit, we provide a hierarchical configuration that allows operators to efficiently coordinate it with the hourly approach to obtain enhanced operation costs. The simulation results validate the advantages of using stochastic programming instead of deterministic approaches under smart grids framework and show how a hierarchical coordination of both methods provides enhanced results. Additionally, computational time has been tested and it has been successfully shown that the proposed methods maintain a reasonable computational burden even for high complexity cases.

Automated summary

The traditional deterministic and centralized approaches to power systems operation have given way to more variable and stochastic methods due to the rise of renewable energy sources and active demand participation. This paper explores the use of stochastic programming and data-driven methods to reduce uncertainty in operation costs. By combining distributed techniques, model predictive control, and a hierarchical configuration, operators are able to enhance operation costs and manage uncertainty effectively, as validated by simulation results.