A Multi-Node Electrical Monitoring Dataset for Appliance-Level Load Analysis, Anomaly Detection, and Power Forecasting

This repository provides a multi-node electrical monitoring dataset comprising 5,509 synchronized physical electrical observations collected across three distributed load branches under controlled laboratory conditions. Telemetry was acquired using ACS712-30A Hall-effect current transducers digitized by dedicated ATmega328P slave microcontrollers using 10-bit ADCs, while AC mains voltage was acquired using a ZMPT101B active voltage transformer digitized at the Arduino UNO Q master gateway by an STM32U585 MCU using a 12-bit SAR ADC. Telemetry frames were packetized with Cyclic Redundancy Check (CRC-16) error checking and transmitted across single-phase 230 V AC mains conductors via KQ-330F Binary Frequency-Shift Keying (BFSK) power-line communication (PLC) transceivers in round-robin time slots (effective per-node logging cadence: ~8–9 s). Dataset Contents: • Zenodo_Dataset_Upload_Package.zip: Complete unified archive containing all data tables, raw node logs, sequence tensors, trained models, and benchmark evaluation outputs. • DATASET_Enriched_5509_Samples.csv: Master tabular dataset with 41 columns capturing voltage (V), current (A), RMS envelope apparent power magnitude (W/VA), apparent circuit impedance (Ohm), power factor proxy, and 28 engineered temporal, differential, and rolling statistical features. • preprocessed_data.npz: Normalized 3D sequence tensors (shape N x 32 x 28) partitioned into 50-sample contiguous blocks across nodes, yielding 807 boundary-safe test windows (W=32). • Raw_Node_Data/: Raw per-node telemetry logs (Node 1: 1,459 observations, Node 2: 2,346 observations, Node 3: 1,704 observations). • models/: Exported PyTorch weights (best_model.pt) and edge ONNX runtime models (model.onnx, model.onnx.data) for the 1D CNN-GRU multitask baseline architecture. • results/: Multitask benchmark metrics (metrics.json) and held-out test predictions (predictions.csv). • DATASET_SCHEMA_AND_METHODOLOGY.md: Complete schema dictionary, pinout connections, calibration parameters, and experimental protocols.