Control and economics of power systems with renewables

能源储存和需求响应等非发电机资源是解决可再生能源间歇性问题的概念性解决方案。NGR在提高电力系统的效率和可靠性方面具有巨大的潜力，但也将因规模、不确定性和经济性而增加相当大的新的复杂性。事实上，需求响应程序可能包含超过10^5个负载，每个负载都受到建模不准确、通信有限以及人类行为和天气等随机因素的影响。一般说来，NGR具有与传统发电机截然不同的特性，例如更快的斜坡能力、硬的能量容量限制，以及由此产生的动态电荷状态。这些因素意味着，NGR不能被视为传统的发电资源。特别是，电力系统运营商必须(I)像使用单个NGR一样使用大型负荷聚合，以及(Ii)根据其独特的物理特征支付NGR，以鼓励适当参与当前市场和未来投资。事实上，2001年加州电力危机和最近的摩根大通交易丑闻等电力市场滥用的持续存在表明了基于物理模型的健全经济机制的重要性。**本研究计划将通过开发有效利用NGR的基本算法和经济框架来应对这些挑战。管理不确定负荷人群的新方法将建立在在线学习理论和基于多面体理论的负荷聚合技术的基础上。例如，将导出多臂强盗索引策略，以同时利用和改进负载估计模型。由于NGR将成为电力系统监管的重要来源，市场必须同时考虑NGR的动态和电力系统。为此，将使用最优控制和动态博弈论的工具来设计合同和定价机制，这些合同和定价机制基于NGR和电力系统监管的物理模型，并确定可能存在的博弈和市场滥用的脆弱性。**这项研究将产生一套广泛的工具，从技术和经济层面优化NGR的利用，进而促进可再生能源与现有电力基础设施的整合。由此产生的好处将包括(I)电力系统运营和经济学方面的新理论见解和研究方向，(Ii)培训高素质人才从事加拿大电力行业和学术界的职业，以及(Iii)减少对环境的影响。

Non-generator resources (NGRs) like energy storage and demand response are conceptual solutions to the intermittency of renewable energy sources. NGRs have great potential to enhance the electric power system's efficiency and reliability, but will also add considerable new complexity through scale, uncertainty, and economics. Indeed, a demand response program may contain upwards of 10^5 loads, each subject to uncertainty from modeling inaccuracies, limited communications, and random factors like human behavior and weather. NGRs in general have starkly different characteristics than conventional generators, such as faster ramping capabilities, hard energy capacity constraints, and consequently dynamic states of charge. These factors mean that NGRs cannot be treated like conventional generation resources. In particular, power system operators must (i) use large load aggregations as though they were individual NGRs, and (ii) pay NGRs with regard to their unique physical characteristics so as to encourage proper participation in current markets and future investments. Indeed, the persistence of electricity market abuses like the 2001 California electricity crisis and the more recent JP Morgan trading scandal signifies the importance of sound economic mechanisms based on physical models.**This research program will address these challenges by developing fundamental algorithmic and economic frameworks for effectively utilizing NGRs. New methodologies for managing uncertain load populations will be built on online learning theory and load aggregation techniques based on polytope theory. For instance, multi-armed bandit index policies will be derived for simultaneously utilizing and improving estimated models of loads. Since NGRs will be significant sources of power system regulation, markets must account for both the dynamics of NGRs and the power system. Toward this end, tools from optimal control and dynamic game theory will be employed to design contracts and pricing mechanisms that are based on physical models of NGRs and power system regulation, and to identify potential vulnerabilities to gaming and market abuse.**The research will produce a broad set of tools for optimizing NGR utilization in technical and economic dimensions, in turn facilitating the integration of renewable energy sources into the existing power infrastructure. The resulting benefits will include (i) new theoretical insights and research directions in power system operation and economics, (ii) training of highly qualified personnel for careers in Canadian power industry and academia, and (iii) reduced environmental impacts.