HVAC

该数据集涉及学习型规划实例，具体是针对具有连续动作和状态空间的暖通空调控制任务。该数据集的暖通空调领域包含6个房间，并在规划任务中展示了其有效性。我们对多个实例进行了评估，以衡量其在学习型神经网络模型下的最优规划性能。This dataset centers on learning-based planning instances, specifically designed for heating, ventilation, and air conditioning (HVAC) control tasks with continuous action and state spaces. The HVAC setup within this dataset consists of 6 rooms, and its effectiveness has been demonstrated in planning tasks. We evaluated multiple instances to quantify the optimal planning performance of learning-based neural network models for these tasks.

搜集汇总

背景与挑战

背景概述

HVAC数据集是一个混合深度MILP规划器，通过学习状态转移函数并转化为MILP问题进行求解，支持离散/连续的动作和状态空间，适用于复杂的规划问题。