Predictive accommodation of spatiotemporal variability in distributed photovoltaic generation

为了实现对全人类至关重要的可持续能源目标，我们必须提高光伏发电在配电网中的渗透率。然而，光伏发电随着时间和天气的变化而波动。我将发现如何通过分布式控制本地光伏发电、存储和负载，在产消框架内适应光伏发电（PVG）的可变性。智能多代理系统（MAS）将被表示为交互能源（TE）代理，其操作由响应动态输入和约束而影响资产的服务支持。这个分布式MAS将通过概率PVG预测得到。这项工作的本质是通过研究受不确定变量（如当地能源和电网条件、定价信号和产消偏好）影响的这些TE代理的社区协调，了解全天运营范围内的预测调节潜力。这些不确定变量之间存在的依赖关系将被处理为使用随机规划的代理之间的预测控制问题。所有智能电网领域都有预期的结果。资产成果是通过模块级电力电子（MLPE）适应光伏板在城市环境中的动态遮蔽，并实现智能家庭能源管理系统（HEMS）作为交互代理。服务贡献是使用阴天预报器生成PVG小时内预报的时间顺序概率模型。在操作上，本研究提供了一个分布式智能电网控制系统，具有完整的协议支持，将电力系统的不同级别联系在一起，形成一个可互操作的整体。对研究领域的好处在于智能电网技术的融合。MLPE方法产生的PVG模型为建筑集成光伏的适应性研究提供了最大的广度，而具有前瞻性逆变器、存储和负载控制的开源HEMS则有利于能源管理领域。PVG的概率预测是通过对能量预测问题的融合而成熟起来的。在操作上，本研究加速了智能电网向强大的交互能源互联网的发展。对加拿大的好处超越了能源部门。这项研究将给生产消费者更多的选择和控制权，同时创建一个能源服务行业，在他们和分销服务运营商之间进行调解。资产技术的进步将使商品解决方案能够帮助消除阻碍光伏渗透的瓶颈。能源预测的进步将在关键的日内时间范围内提供更大的操作确定性，以更低的成本增强能源安全。该项目解决了智能电网发展的新兴方法，并将为政策、市场设计和商业模式提供信息，同时培训急需的工程师。全球智能电网市场预计将以每年20%的速度增长，到2023年将达到520亿美元。

In order to achieve sustainable energy goals, which are of importance to all humanity, we must increase photovoltaic (PV) penetration into the distribution grid. However, PV power fluctuates with time & weather. I will Discover how variability in PV generation (PVG) can be accommodated within a prosumer framework through distributed control of local PVG, storage, and loads. An intelligent multi-agent system (MAS) will be expressed as transactive energy (TE) agents, with operations supported by services that impact assets in response to dynamic inputs and constraints. This distributed MAS will be informed by probabilistic PVG forecasting. The nature of the work is to understand the predictive accommodation potential across intraday operating horizons by studying the community coordination of these TE agents subject to uncertain variables, like local energy & grid conditions, pricing signals, and prosumer preferences. The dependencies that exist between these uncertain variables will be addressed as a predictive control problem between agents using stochastic programming. Anticipated outcomes are in all smart grid domains. Asset outcomes are the accommodation of dynamic overshadowing of PV panels in the urban environment by module level power electronics (MLPE), and the realization of an intelligent home energy management system (HEMS) as a transactive agent. The service contribution is of overcast sky predictors used to generate chronological probability models for PVG intrahour forecasting. Operationally, this research provides a distributed smart grid control system, with full protocol support, that ties the different levels of the electrical system together into one interoperable whole. Benefits to the research field are in smart grid technology convergence. The MLPE approach yields a PVG model that gives the greatest breadth to accommodation studies for building-integrated PV, while an open source HEMS with forward-looking inverter, storage and load control benefits the energy management field. Probability prediction in PVG is matured through the fusion of energy forecasting problems. Operationally, this research accelerates smart grid evolution toward a robust transactive energy internet. Benefits to Canada transcend the energy sector. The research will give prosumers greater choice and control, while creating an energy services industry that mediates between them and distribution service operators. Advances in asset technologies will enable commodity solutions that aid in removing bottlenecks to increased PV penetration. Advances in energy forecasting will provide greater operational certainty over critical intraday time horizons, enhancing energy security at lower cost. This program addresses an emerging approach to smart grid evolution, and will inform policy, market design, and business models while training critically needed engineers. The global smart grid market is expected to grow annually by 20%, reaching USD 52 billion by 2023.