"Three-Level NPC Inverter Dataset Under Normal and Open-Circuit Fault Conditions"

"This dataset provides experimental electrical signals acquired from a laboratory prototype of a three-level Neutral Point Clamped (NPC) inverter under distinct normal and open-circuit fault operating conditions. The dataset was developed to support research on artificial intelligence-based fault detection, fault diagnosis, anomaly detection, signal preprocessing, and condition monitoring in power electronics converters. The experimental setup includes a three-phase NPC inverter operating in open-loop mode, with electrical measurements collected from oscilloscope channels during different operating scenarios. The available conditions include normal operation with a resistive-inductive load, normal operation feeding an induction motor, and resistive-inductive load operation under an open-circuit fault in switch S1a.The recorded signals include voltage and current waveforms associated with the inverter operation, such as phase voltage, line voltage, and output phase currents. The dataset is organized by operating condition, and each folder contains CSV files exported from oscilloscope channels, acquisition screenshots, and Jupyter Notebooks to assist users in loading, visualizing, and inspecting the data in Python. The file naming convention indicates the oscilloscope channel, measured variables, and operating condition, supporting traceability between the raw measurements and the experimental scenario.This dataset is intended for benchmarking and developing machine learning, deep learning, signal processing, and recurrence-based approaches for fault diagnosis and reliability assessment of three-level NPC inverters. It can be used for binary normal-versus-fault classification, multiclass operating-condition classification, anomaly detection using normal data, and open-set diagnostic studies. The dataset contributes to the availability of experimental data for intelligent monitoring of power electronics systems, especially in applications where converter reliability, early fault detection, and data-driven maintenance strategies are critical."