Optimal configuration of concentrating solar power in multienergy power systems with an improved variational autoencoder

Abstract

The proportion of renewable energy sources and the flexibility demands of the power grid have increased simultaneously, which has caused difficulties in renewable energy consumption and the secure operation of power systems. Concentrating solar power (CSP) can provide additional flexibility for power systems and change the uncontrollable characteristics of variable renewable energy generation. Due to the high construction cost, it is necessary to fully evaluate the configurational scheme of CSP. In this paper, we propose an optimal configuration method for CSP in multienergy power systems to fully utilize the CSP benefits. We first improve the variational autoencoder (VAE) to describe the uncertainty in power systems and generate scenarios for the configuration model. The model consists of two stages: planning and operation. The planning model determines the configuration capacity of each component of the CSP station. The operation model considers the day-ahead and real-time periods to analyze the operational cost of the power systems with CSP. By means of linearized methods, the configuration model is treated as a mixed integer linear programming (MILP) formulation to obtain a rapid solution. A case study is conducted based on the power source data of a province in northwestern China. The simulation results show that flexibility values makes CSP more competitive in the configuration process, and the configuration results are significantly affected by the coordination with other power sources. We use sensitivity analysis with different situations to explore the development prospects of CSP.