Energy in the Cloud

云端能源

• 批准号: 543561-2019
• 负责人: Chen, Yuanzhu
• 金额: $ 1.64万
• 依托单位: Memorial University of Newfoundland
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2019
• 资助国家: 加拿大
• 起止时间: 2019-01-01 至 2020-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=678999
• 关键词: Energy; Cloud

Harnessing green energy such as wind and solar is key to sustainable economic growth. Energy storage plays an essential role due to the intermittent nature of these energy sources, and offers many advanced services otherwise impossible for traditional power distribution systems (e.g. energy shifting and peak shaving). The reliability and health of such an essential component of future smart grid is crucial. Thus, the system needs to gather continued operation sensory data to support autonomous decision making while minimizing early personnel intervention, a requirement of particular significance for sparsely populated areas in Canada. The challenge is the that a typical energy storage installation can have over a million battery cells generating an enormous amount of continuous sensor readings of voltage, current, and temperature. Our cloud-based solution is centered around a two-tier computation model, on-site and in-cloud. On site of energy storage systems, raw data are compressed to eliminate spatial and temporal redundancies before sending to the cloud. With sensory data collected network-wide, the in-cloud module builds models for anomaly detection and other intelligence services. Parameters of the trained model are then downloaded to the on-site computers periodically to configure their decision-support module. As such, on site of installations, decision-making is responsive using system-wide experiences and local sensory data. Meanwhile, with all sensory data compressed and uploaded to the cloud, the in-cloud module can iteratively improve its models for further improvement.

利用风能和太阳能等绿色能源是可持续经济增长的关键。 由于这些能源的间歇性，储能发挥着至关重要的作用，并提供了许多传统配电系统无法提供的先进服务（例如能量转移和调峰）。 未来智能电网的重要组成部分的可靠性和健康至关重要。 因此，该系统需要收集持续的操作传感器数据，以支持自主决策，同时尽量减少早期人员干预，这对加拿大人口稀少的地区具有特别重要的意义。 一个典型的储能装置可能有超过一百万个电池单元，产生大量的电压、电流和温度的连续传感器读数。 我们基于云的解决方案以两层计算模型为中心，即现场和云中。 在能源存储系统的现场，原始数据在发送到云端之前被压缩以消除空间和时间冗余。 通过在网络范围内收集传感数据，云模块为异常检测和其他智能服务构建模型。 然后将训练模型的参数定期下载到现场计算机，以配置其决策支持模块。 因此，在安装现场，决策是响应使用系统范围的经验和本地传感器数据。 与此同时，随着所有传感数据被压缩并上传到云端，云模块可以迭代地改进其模型以进一步改进。