Modelling Uncertainty in Modern Power Systems

在过去二十年中，三个主要现象对电力系统的规划和运行产生了重大影响，其中包括：(1)电力市场的重组；(2)将大型风能和太阳能可变发电机并入供应方；(3)将新资源，特别是自行发电的电力，纳入需求方。市场重组带来了经济不确定性，因为能源和备用服务的电力市场价格不再是固定的。相反，价格是在实时和远期拍卖中确定的，因此是不确定的。使用风能和太阳能发电机的能源生产是环保的。然而，由于风速和太阳辐射的自然变异性，发电出力是不确定的。在需求方面，电力消费者的消费模式在历史上是可以预测的，因为消费者一直在被动用电。然而，智能电网技术和需求侧资源使消费者能够控制自己的用电，因此，他们的消费模式不再是可预测和直接的。此外，越来越多的屋顶太阳能组件用于现场发电，抵消了消费者的消费需求。这增加了净需求的可变性，即由中央电网供应的电力。为了有效地运行电力系统，有必要对这些不确定性有一个清晰的认识。以前的研究主要集中在建模和预测不确定变量的解决方案上。然而，仍然有一些领域的解决方案是有限的。大数据解决方案和深度学习技术为解决传统建模技术的一些局限性提供了新的机会。在这项研究中，最终目标是降低社会用电成本。要实现这一目标，需要能够以实用而准确的方式表示电力系统的最优运行和规划模型。这类模型需要考虑经济因素(如短期和长期电价)、供应资源和电力净需求。申请人已经确定了这些领域的具体研究空白。将探索这些差距，并基于大数据解决方案和深度学习方法制定解决方案。这些都是探索大型、复杂数据集的有效方法。开发的解决方案将帮助电力系统政策制定者、运营商和规划者根据实际合理的模型做出短期和长期决策。这项研究将培训大数据解决方案在电力系统中的应用方面的高素质人员；这是一个新兴领域，因此，这项研究可能会在加拿大创造新的知识产权。

Over the past two decades, three major phenomena have significantly impacted how electrical power systems are planned and operated including: (i) the restructuring of electricity markets; (ii) the integration of large-scale wind and solar variable generators into the supply-side, and; (iii) the addition of new resources, particularly self-generation of electricity, into the demand-side. Market restructuring has introduced economic uncertainty because electricity market prices for energy and reserve services are no longer fixed. Rather, prices are determined in real-time and forward auctions, and thus, are uncertain. Energy production using wind and solar-powered generators is environmentally friendly. However, because of the natural variability of wind speed and solar radiation, the generation outputs are uncertain. On the demand-side, the consumption patterns of electricity consumers have historically been predictable because consumers have been using electricity passively. However, smart grid technologies and demand-side resources have enabled consumers to have control over their electricity use, and thus, their consumption patterns are no longer predictable and straightforward. In addition, the increasing adoption of roof-top solar modules for on-site power generation is offsetting the consumption needs of consumers. This adds to the variability of net demand, i.e., electricity that is supplied by the central grid. To run power systems efficiently, a clear understanding of these uncertainties is necessary. Previous research has focused on solutions for modelling and forecasting uncertain variables. Nevertheless, there are still areas where available solutions are limited. Big data solutions and deep learning techniques present new opportunities to address some of the limitations of conventional modelling techniques. In this research, the ultimate objective is to reduce the cost of electricity to society. Achieving this objective requires optimal operation and planning models that are capable of representing power systems in practical yet accurate manners. Such models need to take into account economic factors (e.g., short- and long-term electricity prices), supply resources, and the net demand for electricity. The applicant has identified particular research gaps in these areas. These gaps will be explored, and solutions will be developed based on big data solutions and deep learning methods. These are effective methods for exploring large, complex datasets. The developed solutions will help power system policymakers, operators and planners in making short- and long-term decisions based on practically sound models. The research will train highly qualified personnel on the applications of big data solutions in power systems; this is an emerging area, and thus, the research will likely create new intellectual property in Canada.