A Smartphone-Based Acoustic Dataset for Permanent Magnet Debonding Diagnosis in Brushed DC Motor
Description
This dataset contains acoustic data acquired from a brushed permanent magnet DC motor with a detached permanent magnet and from the same motor after permanent magnet re-bonding. Acoustic recordings were acquired with an Android smartphone positioned approximately 1 m from the motor during steady-state no-load operation. Measurements cover two motor conditions, five rotational speeds (100, 200, 500, 750, and 1000 rpm), and two rotation directions, giving 20 operating conditions. Supplementary low-rate AV-160B vibration tables and vibrometer plot images are included as descriptive supporting files. ## Contents - `MobilePhoneRecordings/`: original smartphone acoustic acquisition files retained as M4A source recordings, organized by motor condition. - `TrimmedWav/`: decoded and trimmed smartphone acoustic recordings in WAV format, organized by motor condition. - `VibrationMeasurementTables/`: supplementary AV-160B vibrometer table exports in XLS format, organized by motor condition. - `VibrationPlotImages/`: supplementary AV-160B vibrometer chart images in JPG format, organized by motor condition. - `DocumentationImages/`: JPG photographs documenting the investigated motor, stator inner surface, and motor nameplate. - `MetadataFiles/metadata.csv`: one row per motor condition, rotational speed, and rotation direction, with direct links to the corresponding M4A, WAV, XLS, and JPG files. - `MetadataFiles/file_manifest.csv`: minimal secondary file list for provenance and file-format checking. ## File Counts - Acoustic M4A source files: 20 - Acoustic WAV working-copy files: 20 - Vibration XLS files: 20 - Vibration plot JPG files: 20 - Documentation JPG files: 3 - Operating-condition rows in `MetadataFiles/metadata.csv`: 20 ## Notes All filenames are in English and use condition, rotational speed, and rotation-direction tokens where applicable. Most users should start with `MetadataFiles/metadata.csv`, which links each operating condition to its corresponding acoustic M4A source file, acoustic WAV working copy, vibration XLS file, and vibration plot image. The M4A files in `MobilePhoneRecordings/` are the smartphone acoustic source recordings. The WAV files in `TrimmedWav/` are decoded working copies with matching basenames. The first 2.0 s of each M4A recording were removed before writing the WAV file to exclude leading silence and possible microphone-handling artifacts introduced when the smartphone recording was started. No additional duration clipping was applied. The vibration XLS files contain low-rate readings, typically 10-12 readings per measured quantity for each operating condition. They are included as descriptive supplementary files rather than as a primary vibration machine-learning dataset.