基于双评判网络的强化学习的无功电压优控制训练数据集

现有配电网优化运行方法以模型驱动为主，对模型准确性高度依赖。随着分布式灵活资源的大量接入，快速建立并实时维护准确、时变的配电网模型变得愈发困难。强化可以从与电网交互的数据中学到最优决策。为进一步提升强化学习的性能，考虑到无功电压优化的两个目标，最小画网损和消除电压越界，设计了双评判网络的强化学习，分别学习网损的奖励函数和电压越界的奖励函数。该数据集为双评判神经网络强化学习的仿真数据集Existing optimal operation methods for distribution networks are primarily model-driven, which heavily rely on model accuracy. With the large-scale integration of distributed flexible resources, it has become increasingly difficult to rapidly establish and maintain accurate, time-varying distribution network models in real time. Reinforcement learning (RL) enables learning optimal decision-making strategies from data generated through interactions with the power grid. To further enhance the performance of reinforcement learning, considering the two objectives of reactive power and voltage optimization—minimizing network loss and eliminating voltage violations—a reinforcement learning framework with dual critic networks is designed, which separately learns the reward functions for network loss minimization and voltage violation elimination respectively. This dataset is a simulation dataset for the reinforcement learning method based on dual critic neural networks.

搜集汇总

背景与挑战

背景概述

该数据集用于训练基于双评判网络强化学习的无功电压优化控制模型，旨在解决配电网中模型驱动方法对准确模型依赖性强的问题。它通过分别学习网损和电压越界的奖励函数，以最小化网损并消除电压越界为目标。