ETT

These datasets originate from two separate power transformers, identified as Transformer 1 and Transformer 2. Each dataset is further categorized into two distinct time intervals for data collection. The first time interval, labeled as "m," represents a short-term sampling period with data recorded every 15 minutes, capturing detailed temporal fluctuations over shorter durations. The second time interval, labeled as "h," signifies a longer-term sampling period with data recorded at hourly intervals, providing an overview of broader trends and patterns. By combining data from these two transformers and their corresponding time intervals, the datasets offer a comprehensive view of power transformer behavior, accommodating both fine-grained and coarse-grained temporal analysis. This structure is particularly beneficial for studying operational characteristics, identifying patterns, and addressing challenges associated with multi-scale time series modeling.

本数据集源自两台独立供电变压器，分别标识为变压器1和变压器2。每个数据集进一步细分为两个不同的数据收集时间段。第一个时间段，标记为“m”，代表短期采样周期，每15分钟记录一次数据，捕捉较短期内的详细时间波动。第二个时间段，标记为“h”，表示长期采样周期，每小时记录一次数据，概述更广泛的趋势和模式。通过结合这两台变压器及其对应时间段的数据，该数据集提供了对电力变压器行为的全面视角，既适用于细粒度时间分析，也适用于粗粒度时间分析。此结构对于研究运行特性、识别模式以及解决与多尺度时间序列建模相关的问题尤为有益。