基于最优决策树的多能系统快速鲁棒优化调度

Fast Robust Optimal Dispatch of Multi-energy System Based on Optimal Decision Tree

摘要:

新能源渗透率的持续增长造成了多能系统快速协调调度的巨大挑战，包括调度结果过于保守以及日内调度低效等问题。为此，提出了一种基于最优决策树分布式鲁棒优化的多能系统协调快速调度方法，所构建模型考虑电网日内经济调度，引入基于范数约束的概率分布置信集精准描述新能源的不确定性，防止调度结果过于保守。同时，根据新能源日内运行数据，分别通过可解释的最优分类树和最优回归树算法，优化日内机组启停状态和出力水平的初始决策量，解决日内鲁棒调度的低效问题。在四川某地区电网的验证结果表明，该模型可在兼顾调度成本和鲁棒性的同时，实现水风光多能系统的日内快速协调调度。

Abstract:

The increasing penetration of new energy has caused challenges to the fast coordinated dispatch of multi-energy systems, including overly conservative operation and low-efficient intraday dispatch. For this purpose, a fast and coordinated dispatch method for multi-energy systems based on optimal decision tree-based distributed robust optimization (ODT-DRO) is proposed. Considering the intraday economic dispatch of power grid, and in order to prevent the result from being too conservative, the proposed model introduces probability distribution confidence set based on norm constraint, which precisely discribes the uncertainties of new energy. At the same time, according to the intraday operation data of new energy, the interpretable optimal classification tree and optimal regression tree algorithms are used to optimize the initial intraday decisions of the status and output level of the units, thus solving the inefficiency problem in intraday rubust dispatch. Testing results of a power grid in Sichuan show that the proposed method can balance the operational costs and the robustness, and meanwhile achieve the fast coordinated intraday dispatch for multi-energy system.