Forecasting Technologies for Operation of Modern Electric Energy Systems Under Uncertainty

在过去的二十年里，电力系统发生了三大变化。首先，电力行业放松管制导致批发电力价格不确定。其次，对环境的担忧促使人们部署风能和太阳能发电，这些发电本质上是可变和不确定的。第三，智能电网的新模式有望为消费者管理其能源使用和参与电力市场创造新的机会。智能电网受益于现有电网之上的额外智能层，以增加电力系统的灵活性和可靠性。与目前的静态负荷相比，智能电网预计将带来更多的动态和价格响应负荷。这三大变化给电价、发电和用电带来了不确定性。提出的研究建议侧重于通过预测不确定变量的未来值来减轻这种不确定性对电力系统运行和规划的影响的技术和工具。特别是，将开发对电价、电力需求、太阳能和风力发电的预测技术。价格和需求预测技术将考虑到这两者在智能电网环境中的相互影响。特别是，电力负荷的动态、电动汽车、对电力消费者的市场激励和智能电网的上网电价将被建模和整合。对于风力发电预测，重点将放在风力斜坡事件上，即预测未来某一时刻风力发电的变化率。此外，风力预报的实际应用将与训练和建立预报引擎相联系，以提高性能。考虑到辐照模式随时间的变化，还将研究太阳能发电和太阳能预测。这部分研究将建立得益于天空成像和数据分析的混合模型。

Over the past two decades, three major changes have been introduced to electric power systems. First, deregulation of power industry has led to uncertain prices of wholesale electricity. Second, environmental concerns have given raise to deployment of wind- and solar-powered electricity generation, which are by nature variable and uncertain. Third, the new paradigm of smart grid is expected to create new opportunities for consumers to manage their energy use and participate in power markets. Smart grids benefit from an extra layer of intelligence on top of the existing grid to add to power systems flexibility and reliability. Smart grids are expected to lead to more dynamic and price-responsive loads compared to today's static load. These three major changes have introduced uncertainty in electricity prices, power generation, and power consumption. The proposed research proposal focuses on techniques and tools to mitigate the impact of such uncertainties on power systems operation and planning by predicting future values of uncertain variables. In particular, forecasting technologies for electricity prices, electricity demand, solar power and wind power will be developed. Price and demand forecasting techniques will take into account the mutual impact of these two on each other in a smart grid environment. In particular, dynamics of electricity load, electric vehicles, market incentives for electricity consumers and feed-in tariffs in smart grids will be modelled and incorporated. For wind power forecasting, the focus will be on wind ramping events, i.e, to predict the rate at which wind power will change at certain times in the future. Moreover, practical applications of wind power forecasts will be tied to training and building forecasting engines for improved performance. Solar power and solar energy prediction will also be investigated considering the variations in irradiation patterns over time.Hybrid models that benefit from sky-imaging and data analysis will be built for this part of the research.