Cascade reservoirs adaptive refined simulation model based on the mechanism-AI coupling modeling paradigm

Abstract

• A cascade reservoirs adaptive refined (CRAR) simulation model is developed. • A sequential cyclic iterative solution method of CRAR model is proposed. • The CRAR model can accurately simulate the operation process of the system. Cascade reservoirs are complex engineering systems. The operation of these reservoirs is not only affected by external effects such as natural flow conditions but is also related to the state of the reservoir project, the operating rules of the reservoirs and the operating mode of the discharge facilities. The realization of refined simulations that can consider various natural and engineering factors is of great significance to improve the reservoir management level. As a method to study the dynamic behavior of complex systems, system dynamics (SD) simulation models have the potential to realize the refined simulation of cascade reservoirs. However, when the SD model is controlled by deterministic rules, the various functional objectives of reservoirs cannot be weighed and emergent scenarios cannot be addressed. In this paper, a mechanical-artificial intelligence (AI) coupling modeling paradigm is proposed, in which the AI algorithm is used to extract adaptive operating rules, and the SD method is used to refined simulate the operations of cascade reservoirs, and the mechanism model and the AI algorithm are coupled to build the cascade reservoir adaptive refined (CRAR) simulation model. The example application shows that the proposed model can be used to effectively reflect the dynamic change in system operations and realize the refined simulation of system operations and adaptive scheduling decisions under various conditions. From this study, a new idea and a reference for the simulation of cascade reservoir operations are provided.