Multi-energy Park Scheduling Method and System Based on Double-layer Reinforcement Learning

Abstract

The multi-energy park dispatching method and system based on double-layer reinforcement learning provided by the present invention includes obtaining dispatchable controllable objects in the integrated energy system, namely source-side units, load-side units, energy conversion units and storage units; constructing a double-layer optimization The decision-making model includes an upper-level reinforcement learning sub-model and a lower-level mixed integer linear programming sub-model; the upper-level reinforcement learning sub-model obtains the action variable information of the storage unit under the current state variable information, and transmits it to the lower-level mixed integer linear programming sub-model; The lower mixed integer linear programming sub-model obtains the corresponding reward variable and the state variable information of the storage unit at the next moment, and feeds back to the upper reinforcement learning sub-model; iteratively executes the above steps until the scheduling ends. The embodiment of the present invention uses a data-driven reinforcement learning method, only needs to make decisions based on the current state, and does not need to predict future information. The timeliness of decision-making is high, the decision-making effect is excellent, and real-time optimal decision-making can be realized.

Description

technical field [0001] The embodiment of the present invention relates to the field of computer technology, in particular to a multi-energy park scheduling method and system based on double-layer reinforcement learning. Background technique [0002] In recent years, with the crisis of fossil energy and the increasingly prominent environmental problems, countries all over the world are looking for new energy utilization methods. The development trend of future energy has the following characteristics: the demand for energy continues to grow, and fossil energy will still be the main primary energy for a long time in the future; the urgency of environmental issues requires constant adjustment of the energy structure with environmental issues as the core; The proportion of renewable energy continues to increase. [0003] Under various energy pressures, the construction of an integrated energy system can achieve multi-energy coupling complementarity and cascaded utilization, so ...

AI Technical Summary

Problems solved by technology

[0008] Finally, the introduction of renewable energy will bring great uncertainty to the comprehensive energy system, and when the scale of the multi-energy park is small, the effect of load due to the randomness of users is more obvious

Therefore, when the future information of the system load, such as: renewable energy output and system load, is difficult to obtain or accurately predict, traditional optimization and dynamic programming methods are difficult to solve the difficult scheduling problem of the multi-energy park energy system

[0009] To sum up, on the one hand, the existing technology is difficult to deal with strong uncertainty scenarios, and it is difficult to deal with scenarios where future information is difficult to predict, and when performing optimization solutions, the solution speed is relatively slow

Images