With the development of NILM, it realizes the functions of device identification, energy consumption decomposition and load switching detection by DL methods. Like other machine learning problems, the research and validation of NILM require a large amount of data support, and different regions have different characteristics of the electricity environment. Therefore, it’s necessary to provide an open datasets for different regions. In this paper, we introduce a TDHA dataset: this dataset contains switch instantaneous data of 10 typical household appliances in China. Compared with other NILM datasets, the TDHA dataset is characterized by high sampling rate, accurate labeling, and close to the real appliance start-up waveforms. At the same time, the switching of appliances is realized by precise control of relay switches, avoiding the interference caused by mechanical switches. By providing such a dataset, it not only helps to improve the recognition accuracy of the existing NILM algorithm, but also helps to apply the NILM algorithm to regions with similar characteristics of electricity consumption as the Chinese region.
Dongyang Zhang;Xiaohua Zhang;Lei Hua;Ren Guan
| Aggregate.zip 4.33 GB | Displayer.zip 1.38 GB | EggBoiler.zip 1.94 GB |
| Fan_1stSpeed.zip 1.75 GB | Fan_3rdSpeed.zip 1.69 GB | Hairdryer_1stSpeedCoolWind.zip 1.82 GB |
| Hairdryer_1stSpeedLowTemperatureWind.zip 1.93 GB | LargeHumidifierHalfPower.zip 1.96 GB | LargeHumidifierMaxPower.zip 1.73 GB |
| LargeHumidifierQuarterPower.zip 1.74 GB | LargeHumidifierThreeQuartersPower.zip 1.65 GB | MicrowaveDefrosting.zip 2.36 GB |
| MicrowaveHighHeat.zip 2.39 GB | MicrowaveLowHeat.zip 2.02 GB | MicrowaveMediumHeat.zip 2.16 GB |
| MobilePhone.zip 1.70 GB | Router.zip 1.61 GB | SmallHumidifierHalfPower.zip 1.98 GB |
| SmallHumidifierMaxPower.zip 1.75 GB | TableLamp.zip 1.71 GB | WaterHeater.zip 2.10 GB |
https://github.com/TagEnd/TDHA-Acquisition-System-Submit
| Start Date | End Date |
| 2023 06 13 | 2023 09 13 |
| Data Type | Data Format |
| Current Voltage | No format supplied |
| Language | |
| English | |
| Number of Records | Umit Type |
| House | 10 appliances |
| Parameters | |
|
the file of each record by the collection and labeling system common timestamp marking (RecvTime, the timestamp can be accurate to the milliseconds), the appliance switching events (event), single appliance labeling format: appliance number - appliance switch; aggregated data labeling format: socketx:0/1), due to the difference in transmission speed data labeling file labeling time will be 20 minutes. (single appliance labeling format: appliance number - appliance switch; aggregated data labeling format: socketx:0/1), due to differences in transmission speeds data labeling file labeling time will be 20ms about a current-voltage cycle of the delay, which has no effect on the event labeling. |
|
Rights:
CC-BY
Right OverView:
This license allows reusers to distribute, remix, adapt, and build upon the material in any medium or format, so long as attribution is given to the creator. The license allows for commercial use. Credit must be given to the creator.
Further Information:
Region:
China
Sector:
Commercial
Subject:
China electricity
Related Project:
This work was supported in part by the National Natural Science Foundation of China under Grant 62371188.
Related Datasets:
No related project