Abstract

The Brazilian Interconnected Power System is hydro dominated and characterized by large reservoirs presenting multi-year regulation capability, arranged in complex cascades over several river basins. In this way, the expansion and operation planning should take into account the uncertainties about the future inflows to hydroplants reservoirs. Currently, a stochastic model for synthetic scenarios generation of monthly streamflow, based on Periodic Auto-Regressive formulation, is used to address the uncertainty. This is the official model used in the Brazilian energy operation planning by the Ministry of Mines and Energy, the National Operator of Electrical System, the Chamber of Electric Energy Commercialization and the Energy Planning Company. Recently, a great scientific effort has been made to include relevant climatic information in stochastic streamflow models. Among several important climatic phenomena in the Brazilian hydrological cycles, El Niño–Southern Oscillation has been pointed as one of the most important. Although the stochastic models that include exogenous variables or that use wavelets present good results, they have limitations for long-term horizon projections or are not suitable for applications that use stochastic dual dynamic programming, which is the case of the Brazilian electrical system. This work proposes an improvement to the current scenario generation model, in order to consider the climate information, but still being suitable to be applied in SDDP algorithms. To achieve this goal, a Markov-Switching Periodic Auto-Regressive model is presented. It is demonstrated that the methodology is able to generate synthetic scenarios which better resembles the observed streamflow, mainly during periods when the streamflow are below-average.

中文翻译：

---

以厄尔尼诺-南方涛动为条件的月流量综合情景生成：在热液系统运行计划中的应用

摘要

巴西的互联电力系统以水力发电为主，其特点是具有多年调节能力的大型水库，以复杂的梯级分布在多个流域上。这样，扩展和运营计划应考虑到未来流入水电站水库的不确定性。当前，基于周期性自回归公式，用于每月流量的综合情景生成的随机模型用于解决不确定性。这是矿产和能源部，国家电气系统运营商，电能商业化商会和能源计划公司在巴西能源运营计划中使用的正式模型。最近，为了将相关的气候信息包括在随机流模型中，已经做出了巨大的科学努力。在巴西水文循环中的几种重要气候现象中，厄尔尼诺-南方涛动被认为是最重要的现象之一。尽管包含外生变量或使用小波的随机模型显示出良好的结果，但它们对长期的水平投影有局限性，或者不适用于使用随机双重动态规划的应用程序（巴西电气系统就是这种情况）。这项工作提出了一种对当前情景生成模型的改进，以考虑气候信息，但仍适用于SDDP算法。为了实现这一目标，提出了马尔可夫切换周期自回归模型。