"Agricultural Irrigation Non-intrusive Load Monitoring Dataset\uff08AI-NILM\uff09"

"This dataset is collected for Non-intrusive Load Monitoring (NILM) research in an agricultural irrigation scenario. The measurement site is an anonymous agricultural user in China (real name removed for privacy), with a user capacity of 100 kVA, high-voltage supply and low-voltage metering (no PT, CT ratio 150:5). The main loads include three water pumps (7.5 kW, 11 kW, 15 kW, and an additional 18 kW pump on one branch) and an office area (air conditioners, refrigerators, computers, lighting, etc.).Raw high-frequency waveform recordings were performed at both the total supply side (bus side) and the equipment side (three branches: large pump branch, small pump branch, and office area branch) from June 2 to June 4, 2021. The total recording duration is about 42.5 hours, with original waveform data size of 34 GB. Based on the raw waveforms, we calculated the three-phase RMS voltage and three-phase RMS current at an approximate sampling interval of 0.1 seconds (10 Hz). The derived RMS values are stored in CSV files.The dataset contains:Bus-side three-phase RMS voltage and current (CT ratio = 30).Equipment-side three-phase RMS voltage and current (CT ratio = 1) for three branches: large pump (formerly named \u201cLiujia\u201d), small pump, and office area.Each CSV file includes sequential index, timestamp (with possible minor inaccuracies), and six columns: UA, IA, UB, IB, UC, IC. This dataset is suitable for NILM algorithm development, load disaggregation, transient\/steady-state analysis, and agricultural load modeling. "