Rejection depth differential dynamic programming real-time power generation scheduling and control algorithm

Abstract

The invention provides a rejection deep differential dynamic programming real-time power generation scheduling and control algorithm. The algorithm can effectively solve the problem that a traditional combined algorithm (economic scheduling + automatic power generation control + power instruction distribution) is difficult to coordinate and control. The algorithm provided by the invention is composed of a depth model network module, a depth evaluation network 1 module, a depth evaluation network 2 module and a depth execution network module, wherein the core of each module is a depth neural network. According to the invention, the rejection operation provided by the invention is the limiting operation of a power generation power instruction outputted by a microgrid system: when the power generation power instruction is greater than a rejection threshold, a result is outputted; when the power generation power instruction is less than or equal to the rejection threshold value, a result of a traditional proportional integral differential algorithm is outputted. According to the algorithm provided by the invention, the deep differential dynamic programming algorithm is used for learning a system, so that the system has higher generalization capability, and the problems of power generation scheduling and control in the microgrid can be well solved.

Description

technical field [0001] The invention belongs to the field of power generation scheduling and control in electric power systems, and relates to an algorithm that replaces the traditional combination of "economic scheduling + automatic power generation control + power command distribution", which is suitable for power generation scheduling and control in power systems. Background technique [0002] Nowadays, problems such as energy shortage, climate change and environmental pollution are becoming more and more serious. In order to effectively solve these problems, the development and utilization of renewable energy has been further accelerated. A microgrid is a system that combines various micro sources, loads, and energy storage devices to form a controllable system. On the one hand, with the continuous access of renewable energy (such as biomass power generation, wind turbines, photovoltaics, energy storage and electric vehicles, etc.) to the distribution network, the power ...

AI Technical Summary

Problems solved by technology

The difference between this algorithm and the traditional combined algorithm of "economic dispatch + automatic generation control + power command allocation": (1) The real-time generation dispatch and control algorithm of recognition depth differential dynamic programming can solve the problem of "economic dispatch + automatic generation control + power command (2) The real-time power generation scheduling and control algorithm of deep differential dynamic programming can not only learn the power generation scheduling and control operation of the microgrid, but also improve the system’s robustness. stickiness and feasibility

Images