Behavioral and Predictive Analytics for Efficient Energy Consumption Management in Smart Grids

智能电网系统的目标之一是通过将终端用户作为智能电网技术的核心参与者，最大限度地提高能源消耗管理程序的效率。尽管为促进这一目标作出了巨大努力，但不幸的是，根据IndEco战略咨询公司为加拿大自然资源公司进行的一项最新研究，消费者并不是一个中心考虑因素。该研究得出的结论是，到2030年，智能电网技术“将受到消费者参与度不足的限制”。我们假设，行为和预测分析可以提高公用事业公司如何与消费者合作的知识。考虑到智能电表产生的大量消费数据为公用事业公司提供了前所未有的机会来了解电表两侧的动态，这一点尤其有希望。在消费方面，行为分析技术允许公用事业公司发现可以集成到智能电网技术中的能源使用偏好。在运营方面，预测分析技术允许公用事业公司与消费者进行近乎实时的互动，以促进基于负载需求的准确预测的基础设施规划。

One of the objectives of smart grid systems is to maximize the efficiency of energy consumption management programs, by engaging end-users as central players in smart grid technology. Although enormous effort has been undertaken to promote this objective, unfortunately, consumers are not a central consideration according to a recent study conducted by IndEco Strategic Consulting Inc. for Natural Resource Canada. The study concluded that smart grid technology by 2030 "will be limited by weakness in consumer engagement". We hypothesize that behavioral and predictive analytics can advance utilities' knowledge of how to partner with consumers. This is particularly promising given the large volume of consumption data produced by smart meters which provide unprecedented opportunities for utilities to understand the dynamics on both sides of the meter. On the consumption side, behavioral analytics techniques allow utilities to uncover energy usage preferences that can be integrated into smart grid technologies. On the operation side, predictive analytics techniques allow utilities to interact with consumers in near real-time to facilitate infrastructure planning based on an accurate forecast of load demand. This research program focuses on behavioral and predictive analytics aspects pertaining to household smart meter data. Behavioral analytics as an approach for understanding energy consumption in households is relatively new. Predictive analytics is well studied for short and long-term load forecasting. However, very-short-term [VST] (latency of few seconds or minutes) predictions that focus on the immediate use of energy are required to engage consumers in near real-time smart grid planning applications. The long-term objective of this research program is to explore innovative behavioural and predictive analytics techniques that support utilities and consumers in adopting efficient energy management programs. Specific short-term objectives are as follows. (1) The development of behavioural analytics mechanisms and methods to analyze comprehensively household energy consumption data to promote efficient energy management programs better (2) The development of new predictive analytics techniques for VST energy predictions and the development of new strategies to evaluate the performance of these techniques (3) The development of innovative platform to integrate data analytics techniques with fewer resource constraints and the development of new privacy-preserving mechanisms that balance the trade-off between privacy concerns and the use of data This research program will provide unique opportunities to train HQPs in topics considered highly in-demand by Canadian companies. Also, the developed technologies are critically important for Canadian utilities seeking to promote energy consumption management programs that benefit Canada's economy and environment.