Datasets
Standard Dataset
Non-intrusive load monitoring datasets for two industrial scenarios
- Citation Author(s):
- Submitted by:
- Zhenyu Zhang
- Last updated:
- Wed, 03/01/2023 - 07:20
- DOI:
- 10.21227/7b5a-g160
- Data Format:
- License:
1421 Views- Categories:
- Keywords:
Abstract
The dataset contains two industrial cases with measurement devices installed at the medium voltage bus entrance and at the target load to be identified. It can be used for the research of non-invasive load monitoring algorithms. The types of data measured include three-phase voltage, current, active power, reactive power, as well as the amplitude and phase angle of each harmonic.
The dataset contains two cases.
Case 1 was collected by Hioki PQ3198, with a data duration of about one week. Includes bus measurements of the plant, and load measurements of the compressor. Data types include three-phase voltage, current, active power, reactive power, apparent power, frequency, and data for each harmonic and phase angle. Data times have been calibrated and aligned. Data resolution is 1s per line.
Case 2 is collected by a self-developed measurement terminal with a data duration of about 2 months. It includes bus measurements of a combined heat, cold and power energy station, as well as load measurements of one of the hot water circulation boilers. The data types include three-phase voltage, current, power, frequency and harmonics and phase angle data. Due to data transmission stress, only the maximum 5 data per phase are retained for each second of harmonic data. The data times have been calibrated and aligned with exact Unix timestamps. Data resolution is 1s per line.
More detailed description of the dataset is available in the paper.
Z. Zhang et al., “A Multi-State Load State Identification Model Based on Time Convolutional Networks and Conditional Random Fields,” IEEE Trans. Artif. Intell., pp. 1–9, 2022, doi: 10.1109/TAI.2022.3203685.
