Novel statistical methods for opportunity detection in energy time series

该项目寻求开发受最终用户能源需求系统启发并反馈给最终用户能源需求系统的新的统计机会检测方法。由于当代的传感器和物联网技术，现在可以实时监控一系列设备和位置的能源使用情况。这提供了一个综合来自众多来源的数据的机会，以期确定不同的使用模式，从而确定可能的节能策略。为了实现这一目标，我们将寻求开发新的方法来表征这种能量序列，建立在变点检测的进展基础上，以及其他检测信号中异常结构的方法。在可行的情况下，我们还将设法以可复制软件的形式提供实施这种方法的通用软件。

This project seeks to develop novel statistical opportunity detection methods inspired by, and feeding back into, end user energy demand systems. Due to contemporary sensor and IoT technologies, it is now possible to monitor energy usage across a range of devices and locations in real time. This provides an opportunity to synthesise data from numerous sources with a view to identifying different usage patterns, and thereby possible energy saving strategies. To achieve this, we will seek to develop novel methods to characterise such energy series, building on advances in changepoint detection, as well as other methods for detecting anomalous structures in signals. Where feasible, we will also seek to make generic software implementing such methods available in the form of reproducible software.