Acoustic Microphone Recordings of a Mechanical Spindle Under Various Fault Conditions

This dataset contains acoustic recordings (WAV format) of a motor-driven mechanical spindle operating under various fault conditions and rotational speeds. Experimental Setup An ICS-40300 MEMS microphone was positioned near a motor-spindle assembly to capture the acoustic signature of the spindle during operation. Mechanical defects were introduced by adjusting nuts on the motor shaft to apply asymmetric bearing pressure, causing a slight imbalance that affects the spindle's behavior. The microphone specifications are: Sensor: InvenSense ICS-40300 MEMS microphone Sampling rate: 44.1 kHz Bandwidth: 20 Hz -- 20 kHz File format: WAV Note on synchronization strikes Each recording contains an audible strike (impact) on the rear end of the motor shaft (where space is available) performed by the operator. This strike serves solely as a synchronization marker to temporally align the microphone signal with other sensors used in the full experimental setup. It does not affect the spindle's operation and is not related to the fault conditions under study. Experimental Conditions The dataset comprises 32 experiments (exp-001 to exp-032) totaling 211 validated measurements. Each experiment folder contains a context.json file describing the experimental conditions, and one or more WAV audio recordings named with a YYYYMMDD-HHMMSS_audio.wav timestamp convention. The experiments systematically vary the following parameters: Motor speed: 0 rpm (stationary), 2000 rpm, 4000 rpm, and 6000 rpm Defect condition: No defect (baseline), defect on left side, defect on right side, or defect on both sides (left and right are defined when facing the spindle) Data Structure meas_audio/ ├── exp-001/ │ ├── context.json # Experiment metadata (speed, defect, protocol) │ ├── YYYYMMDD-HHMMSS_audio.wav │ └── ... ├── exp-002/ │ ├── context.json │ └── ... └── exp-032/ Context and Scope The data is intended for research in acoustic-based condition monitoring, fault detection, and signal processing for spindles. Potential applications include spectral analysis of mechanical faults, anomaly detection, and machine learning for predictive maintenance. Measurement Protocol Experimental conditions (motor speed, defect configuration) were set and recorded in context.json. The microphone was positioned near the motor-spindle assembly. The motor was set to the target speed and recording was started. Recording was stopped and the WAV file was saved with a timestamp.