University of Ottawa Electric Motor Dataset – Vibration and Acoustic Faults under Constant and Variable Speed Conditions (UOEMD-VAFCVS)

The dataset consists of constant and variable speed data collected from 8 different D396 Marathon Electric 3-phase motors. Each motor has different types of faults artificially created by SpectraQuest. Data is collected using three accelerometers and a microphone (the first, third and fourth columns are accelerometer data, the second column is acoustic data, and the fifth column is temperature data). The labelling of data is as follows {Letter}-{Letter}-{Number}-{Number}: The following values represent the possibilities for the first letter: - H indicates healthy - R indicates rotor - S indicates stator - V indicates voltage - B indicates bowed - K indicates broken - F indicates faulty The second letter is represented by the following possible values: - H indicates healthy - U indicates unbalance - M indicates misalignment - W indicates winding - R indicates rotor - A indicates rotor bars - B indicates bearing The combination of both letters indicates the health state of the motor present in the data sample (e.g. H-H = a healthy motor, S-W = a stator winding fault). The first number represents: - 1 is for a constant speed of 15 Hz - 2 is for a constant speed of 30 Hz - 3 is for a constant speed of 45 Hz - 4 is for a constant speed of 60 Hz - 5 is for an increasing speed from 15 Hz to 45 Hz - 6 is for an increasing speed from 30 Hz to 60 Hz - 7 is for a decreasing speed from 45 Hz to 15 Hz - 8 is for a decreasing speed from 60 Hz to 30 Hz The second number represents: - 0 for a no load condition - 1 for a loaded condition The duration of data collection was 10 seconds at a sampling frequency of 42 kHz. Each file contains 420,000 samples. Raw data is provided in the folders as comma separated values files (.csv) and MATLAB files (.mat).

本数据集采集自8台不同的D396型Marathon Electric三相电机的恒速与变速运行数据。所有电机的各类故障均由SpectraQuest人工诱发。数据采集采用三台加速度传感器（accelerometers）与一台麦克风（microphone），其中第1、3、4列为加速度传感器数据，第2列为声学数据，第5列为温度数据。
数据采用"{字母}-{字母}-{数字}-{数字}"的格式进行标注，各组成部分含义如下：
1. 首字母可选取值及对应含义：
- H：健康（healthy）
- R：转子（rotor）
- S：定子（stator）
- V：电压异常（voltage）
- B：轴弯曲（bowed）
- K：断裂（broken）
- F：故障（faulty）
2. 次字母可选取值及对应含义：
- H：健康
- U：不平衡（unbalance）
- M：轴系不对中（misalignment）
- W：绕组故障（winding）
- R：转子故障（rotor）
- A：转子导条故障（rotor bars）
- B：轴承故障（bearing）
上述两个字母的组合代表对应数据样本中电机的健康状态，例如H-H代表健康电机，S-W代表定子绕组故障。
3. 第一个数字的含义：
- 1：恒速15 Hz
- 2：恒速30 Hz
- 3：恒速45 Hz
- 4：恒速60 Hz
- 5：转速从15 Hz升至45 Hz
- 6：转速从30 Hz升至60 Hz
- 7：转速从45 Hz降至15 Hz
- 8：转速从60 Hz降至30 Hz
4. 第二个数字的含义：
- 0：空载工况
- 1：负载工况
本次数据采集时长为10秒，采样频率为42 kHz，单个文件包含420000个采样点。原始数据以逗号分隔值文件（.csv）与MATLAB文件（.mat）的格式存储于对应文件夹中。

搜集汇总

背景与挑战

背景概述

该数据集由渥太华大学发布，专注于电机故障诊断，包含从8台三相电机收集的振动、声学和温度数据，覆盖健康及多种人为故障类型。数据在恒定和变速条件下采集，采样频率高（42 kHz），标签系统详细编码了电机状态、速度和负载信息，适用于机器学习模型训练和故障检测研究。