Transforming electricity access through smart sensors and grid efficiency algorithms

发达国家和发展中国家的配电网络运营商都面临着解决能源三难困境的重大挑战——提供清洁、负担得起和安全的能源。电力需求的增加，加上供应向分布式发电的快速转变，要求dno增加监测、分析和优化，以继续提供具有成本效益的服务。然而，在印度等发展中国家，挑战可能更为极端，即：每年因电力盗窃而损失的890亿美元中，超过160亿美元来自印度，再加上基础设施老化导致的重大技术损失，导致电价上涨和频繁停电；由于可再生能源的不均匀、可变和双向性，电网无法有效地整合不断增长的可再生能源（到2027年，印度计划增长380%）；电力供应无法满足发展中市场不断增长的电力需求（预计到2040年非经合组织国家增长71%），这导致电力缺口达到峰值（2015-16年印度增长3.2%）、电力短缺和常规停电。OrxaGrid提供了一种更智能、成本更低的替代监控系统，以提高电网效率，而不是安装昂贵的传统SCADA系统或对网络进行昂贵的扩展。OrxaGrid的工作原理是“监控、分析和优化”，通过提供智能物联网传感器，在配电电网的关键节点上进行改造。实时监控的数据通过蜂窝/互联网/LoRa发送到云端进行分析。雷丁大学的研究小组将分析这些数据，以确定趋势，并开发预测模型和分类引擎，既可以预测未来变电站的能源使用情况，也可以在重要事件发生时检测到。从数据中提取价值，以可持续的商业模式实现智能电网服务，同时满足能源三难困境的需求，是一项重大挑战。该项目的研究将首先对原始变电站数据进行数据分析，以确定趋势和模式。然后，研究小组将在数据中确定对能源系统重要的关键事件，例如当能源需求接近变电站的运行极限或当电力供应中断时。通过创建此类事件的库，可以自动检测和识别未来的事件。储能等关键智能电网系统需要调度。由于储能系统的未来状态取决于其过去的状态，储能系统不能简单地调整以满足给定时刻的给定需求。例如，一个已经充满电的电池不能继续充电。因此，为了有效地使用这样的系统，对未来的需求进行一些预期是必要的。该项目将开发一个预测模型，该模型将使用历史变电站数据来预测未来的需求。然而，在实践中，在设计预测时必须考虑到特定的应用。在这项工作中，这些预测将与重要事件库一起使用，以检测何时发生了不寻常的事情并确定原因。一旦预测模型到位，关键事件可以被检测到，该平台将能够为智能电网系统（如储能、需求响应和电动汽车充电）提供推荐的时间表。将开发算法来确定这些推荐的时间表。将运行基于真实变电站数据的模拟，以展示这些算法与推荐的低碳技术组合一起运行的影响。此外，这些算法将尝试检测电能盗窃，其中预期家庭可以与实际需求进行比较，以确定是否有任何未计量的能源被使用的可能性。

Electricity distribution network operators (DNOs) in both developed and developing countries are facing significant challenges to address the energy trilemma- offering clean, affordable and secure energy. Increased demand for electricity coupled with the rapid shifting of supply to distributed generation requires DNOs to increase monitoring, analytics and optimisation in order to continue to provide a cost-effective service. However, in developing countries such as India, the challenges can be more extreme namely: over $16Bn of the $89Bn lost annually to electricity theft comes from India, which when coupled with significant technical losses due to aging infrastructure, result in increasing electricity prices and frequent power outages; the grid is unable to effectively integrate the growing renewables (380% growth planned in India by 2027) due to the uneven, variable and bidirectional nature of renewables; electricity supply is unable to meet the growing demand for electricity (71% non-OECD growth expected by 2040) in developing markets which is leading to peak deficits (3.2% in India for 2015-16), power shortages and routine blackouts.Instead of installing expensive legacy SCADA systems or making costly expansions to the network, OrxaGrid provides a smarter, lower cost alternative monitoring system for improving grid efficiency. OrxaGrid works on the principle of 'Monitor, Analyse and Optimize' by providing smart IoT sensors that are retrofitted on critical nodes of distribution electricity grids. Realtime monitored data is sent to the cloud via cellular/internet/LoRa for analytics. The research team at the University of Reading will analyse this data in order to determine trends and to develop both a forecast model and a classification engine that can both predict future substation energy use and also detect important events as they occur.Extracting value from data to enable smart grid services with a sustainable business model that also meets the needs of the energy trilemma is a significant challenge. The research in this project will first perform data analytics on the raw substation data to identify trends and patterns. The research team will then identify within the data key events that are important to the energy system such as when energy demand is approaching operational limits of the substation or when power supplies become disrupted. By creating a library of such events, future events can be automatically detected and identified.Key smart grid systems such as energy storage require scheduling. As the future state of an energy storage systems depends on its past state, energy storage systems can not simply adjust to meet a given requirement in a given moment. For example, a battery that is already charged to full capacity can not continue to charge. Therefore, to use such systems effectively, some expectation of future requirements is necessary. This project will develop a forecast model that will use historical substation data to predict future requirements. However, in practice forecasts must be designed with a specific application in mind. In this work, the forecasts will be used, alongside the library of important events, to detect when something unusual has happened and identify the cause.Once a forecast model is in place and key events can be detected, the platform will be able to provide recommended schedules for smart grid systems such as energy storage, demand response and electric vehicle charging. Algorithms will be developed to determine these recommended schedules. Simulations based on real substation data will be run to demonstrate the impact of these algorithms running in conjunction with the recommended portfolios of low-carbon technologies. In addition, these algorithms will attempt to detect electrical energy theft, where expected household can be compared to actual demand to determine the likelihood of whether any unmetered energy is being used.