Non--Intrusive Load Monitoring

非侵入式负载监控

• 批准号: RGPIN-2018-06192
• 负责人: Makonin, Stephen
• 金额: $ 2.04万
• 依托单位: Simon Fraser University
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2020
• 资助国家: 加拿大
• 起止时间: 2020-01-01 至 2021-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=713136
• 关键词: Non; Intrusive; Load; Monitoring

This proposed research program is for the invention of new algorithms, models, and systems to solve the difficult challenge of disaggregating total power readings. Disaggregating power/energy data is known as “non-intrusive load monitoring” (NILM). The goal is to use the power utility's smart meter for a house, building, microgrid, etc., to infer and track the performance of appliances and loads without the aid of sub-meters. This will help create smart, energy-efficient environments. While disaggregation is a difficult, ill-posed problem, with advanced algorithms and good statistical models, disaggregation can become feasible. Disaggregation can help reduce energy consumption to create a sustainable future by meeting our Paris Climate Agreement commitment to cut carbon emissions 30% below 2005 levels by 2030. Power consumption by residential homes and commercial buildings continues to increase each year despite various government incentives and more energy-efficient appliances on the market. The most recent Statistics Canada data show that Canadian households have increased their energy consumption by 4.5% in 2013 (97.5 GJ/household) from 2011 (93.3 GJ/household). User studies suggest that simply showing appliance data to occupants in real-time can reduce energy consumption by 14%. This program investigates the theory and limits of disaggregation by inventing and deploying highly advanced unsupervised learning algorithms and rigorous statistical models that require little or no prior data. Knowledge visualization theory is also explored to increase the occupants' understanding of how loads consume energy. Augmented reality will be explored as a way to convey consumption information. New Big Data algorithms will be needed to augment energy/power data with other available sensor data in real-time, which can potentially improve NILM accuracy. This proposed program also looks at developing standardized evaluation metrics that measure the accuracy of algorithms and models, as well as open-source tools such as emulators and simulators that can further help in the reproduction and comparison of algorithms developed by the research community at large. A program such as this requires researchers to be interdisciplinary to understand sustainability. Researchers trained in this program will gain exposure/expertise in environmental awareness, smart grid and power grid, visual analytics, software engineering, machine learning, sensors and automation, and data collection. Understanding sustainability is becoming an essential skill for the 21st-century economy, and this program will help contribute to creating the leaders and innovators of tomorrow. Additionally, disaggregation will provide solutions to meet Canada's international commitment to the Paris Climate Agreement by reducing the impacts of climate change.

这项研究计划旨在发明新的算法，模型和系统，以解决分解总功率读数的困难挑战。分解功率/能量数据被称为“非侵入式负载监控”（NILM）。目标是将电力公司的智能电表用于房屋、建筑物、微电网等，在没有分表的帮助下推断和跟踪电器和负载的性能。这将有助于创造智能、节能的环境。虽然分解是一个困难的，不适定的问题，先进的算法和良好的统计模型，分解可以成为可行的。 分解可以帮助减少能源消耗，通过实现我们的巴黎气候协定承诺，到2030年将碳排放量减少到2005年水平的30%，创造可持续的未来。尽管政府采取了各种激励措施，市场上也出现了更节能的电器，但住宅和商业建筑的耗电量每年仍在增加。加拿大统计局的最新数据显示，2013年加拿大家庭的能源消耗量（97.5 GJ/户）比2011年（93.3 GJ/户）增加了4.5%。用户研究表明，简单地向用户实时显示设备数据可以减少14%的能耗。 该计划通过发明和部署高度先进的无监督学习算法和严格的统计模型来研究分解的理论和限制，这些算法和模型需要很少或不需要先验数据。知识可视化理论也被探索，以增加乘员的负载如何消耗能量的理解。增强现实将被探索作为传达消费信息的一种方式。需要新的大数据算法来实时增加能源/电力数据与其他可用的传感器数据，这可能会提高NILM的准确性。该计划还着眼于开发标准化的评估指标，以衡量算法和模型的准确性，以及开源工具，如仿真器和模拟器，可以进一步帮助复制和比较研究界开发的算法。 像这样的项目需要研究人员跨学科来理解可持续性。在该计划中接受培训的研究人员将获得环境意识，智能电网和电网，视觉分析，软件工程，机器学习，传感器和自动化以及数据收集方面的专业知识。了解可持续发展正在成为21世纪经济的基本技能，该计划将有助于创造明天的领导者和创新者。此外，分类将提供解决方案，通过减少气候变化的影响，履行加拿大对巴黎气候协定的国际承诺。