Vibration, acoustic, temperature, and motor current dataset of rotating machine under varying operating conditions for fault diagnosis

Rotating machines are often operated under various operating conditions. However, the characteristics of the data varies with their operating conditions. This article presents the time-series dataset, including vibration, acoustic, temperature, and driving current data of rotating machines under varying operating conditions. The dataset was acquired using four ceramic shear ICP based accelerometers, one microphone, two thermocouples, and three current transformer (CT) based on the international organization for standardization (ISO) standard. The conditions of the rotating machine consisted of normal, bearing faults (inner and outer races), shaft misalignment, and rotor unbalance with three different torque load conditions (0 Nm, 2 Nm, and 4 Nm). This article also reports the vibration and driving current dataset of a rolling element bearing under varying speed conditions (680 RPM to 2460 RPM). The established dataset can be used to verify newly developed state-of-the-art methods for fault diagnosis of rotating machines. Mendeley Data. DOI:10.17632/ztmf3m7h5x.6, DOI:10.17632/vxkj334rzv.7, DOI:10.17632/x3vhp8t6hg.7, DOI:10.17632/j8d8pfkvj2.7

a b s t r a c t Rotating machines are often operated under various operating conditions. However, the characteristics of the data varies with their operating conditions. This article presents the time-series dataset, including vibration, acoustic, temperature, and driving current data of rotating machines under varying operating conditions. The dataset was acquired using four ceramic shear ICP based accelerometers, one microphone, two thermocouples, and three current transformer (CT) based on the international organization for standardization (ISO) standard. The conditions of the rotating machine consisted of normal, bearing faults (inner and outer races), shaft misalignment, and rotor unbalance with three different torque load conditions (0 Nm, 2 Nm, and 4 Nm). This article also reports the vibration and driving current dataset of a rolling element bearing under varying speed conditions (680 RPM to 2460 RPM). The established dataset can be used to verify newly developed state-of-the-art methods for fault diagnosis of rotating machines. Mendeley Data. DOI: 10.17632/ztmf3m7h5x.6 , DOI: 10.17632/vxkj334rzv.7 , DOI: 10.17632/x3vhp8t6hg.7 , DOI: 10.17632/j8d8pfkvj2.7

Value of the Data
• This article consists of two parts: varying load conditions, and varying speed conditions. In part 1, this dataset contains data related to most of the major faults (bearing, shaft, and rotor faults) that can occur in rotating machines. Therefore, this dataset can be used to verify the performance of the newly developed rotating machine fault diagnosis methods based on rotor dynamics theories. • In particular, by securing the dataset according to various load conditions, it is possible to observe the change in the fault features according to the load variation. This provides a practical dataset to consider the load fluctuation conditions in the actual field. • Recently, many fault diagnosis researches using non-contact sensors instead of vibration sensors are being conducted due to the problem of sensor installation and cost in the actual field [ 1 , 2 ]. In this context, this dataset is expected to lead the state-of-the-art fault diagnosis research that utilizes sensor fusion, such as vibration-acoustic or vibrationcurrent. • In part 2, this dataset was acquired from rolling element bearing under varying speed conditions (680 RPM to 2460 RPM). Three different types of faults, including inner race fault, outer race fault, and ball fault, were seeded. This data consists of vibration data (in the x-and y-directions of the bearing), and current data. • ·Most of the fault diagnosis methods are proposed for extracting fault features with steady speed and cannot be directly used with varying speed conditions [3] . Practically, in wind turbines, bearing does not operate at a steady speed due to load fluctuations [4] . To solve these problems, eliminating the effect of varying speed condition such as order tracking are conducted, however, it needs to collect synchronized speed data with vibration [5] . • This dataset can be used to develop a learning-based fault diagnosis methodology despite varying speed conditions [6] . Synchronized speed data and vibration data under constant speed conditions are also provided [2] .

Objective
In part 1, this dataset was established for deep learning based rotating machine fault diagnosis research. Unlike other researches, it is very difficult to obtain data in the fault diagnosis research field because it is difficult to apply an actual failure to make training of deep learning algorithms challenging. To solve this problem, we simulated bearing faults, unbalance faults, and misalignment faults that may occurred dominantly in rotating machine. We collected vibration, acoustic, temperature and driving current data under different load conditions (0 Nm, 2 Nm, and 4 Nm). This dataset is measured based on mechanical engineering knowledge in accordance with ISO international standards. This dataset can be used for the verification of newly-developed learning-based fault diagnosis methods.
In part 2, this dataset was established for learning-based ball bearing fault diagnosis research. Unlike other researches with constant speed, it is very difficult to obtain data under the varying speed condition. In contrast, we obtained faulty vibration and driving current data under varying speed conditions (680 RPM to 2460 RPM). This dataset is measured based on mechanical engineering knowledge in accordance with ISO international standards. This dataset can be used for verification of the learning-based fault diagnosis method.

Data Description
This article presents two varying operating condition including varying load condition and varying speed condition. First dataset consists of vibration, acoustic, temperature and driving current data under varying load conditions. Vibration, temperature, motor current, and acoustic data are collected under 3 different load conditions (0 Nm, 2 Nm and 4 Nm). The load conditions are controlled by hysteresis brake with air cooling method. The main motor rotates at a rated rotating speed of 3010 RPM.
Vibration data were measured using four accelerometers (PCB352C34) at two bearing housings (A, B) in the x-direction and y-direction, simultaneously. An acoustic microphone (PCB378B02) was located nearby the bearing housing (A). Temperature and driving current data were measured using two thermocouples (K-type) and three CT sensors (Hioki CT6700). Siemens SCADAS Mobile 5PM50 was used for collecting vibration and acoustic data. NI9211, and NI9775 modules were used for collecting temperature, and driving current data, respectively. Vibration, temperature, driving current data were collected at a sampling frequency of 25.6 kHz. This dataset was collected for 120 seconds in normal state, and for 60 seconds in faulty state. Lastly, acoustic data were collected with a sampling frequency of 51.2 kHz and only acquire bearing fault data under no-load conditions to avoid the noise from air-cooled brakes.
The collected vibration and acoustic data are stored in binary MATLAB (MAT) files [ 7 , 8 ]. The vibration data file contains five columns, namely 'Time Stamp', 'x_direction_housing_A', 'y_direction_housing_A', 'x_direction_housing_B', and 'y_direction_housing_B'. The unit of the vibration data is 'gravitational constant (g)' (1g = 9.80 6 65 m/s 2 ). The acoustic data file contains two columns, namely 'Time Stamp', and 'values'. The unit of the acoustic data is 'Pascal (Pa)'. The description of the vibration and acoustic files as per operating and health conditions of the rotating machine is as follows: < vibration >   Second, the collected dataset consists of vibration and current data acquired from the ball bearing with different fault types of inner race faults, outer race faults, and ball faults, according to changes in motor speed conditions (680 RPM and 2460 RPM).
Vibration data were measured using four accelerometers (PCB352C34) at the two bearing housing A and B in the x-direction and y-direction. Current data were measured using three CT current sensors (Hioki CT6700). Vibration data were acquired by a Siemens SCADAS Mobile 5PM50 with sampling frequency of 25.6 kHz, and current data were acquired by NI9775 with sampling frequency of 100 kHz. This dataset was collected for 600 seconds at constant speed, and for 2,100 seconds at varying speed conditions (680 RPM and 2460 RPM).
The vibration data file contains five columns, namely 'Time Stamp', 'x_direction_housing_A', 'y_direction_housing_A', 'x_direction_housing_B', and 'y_direction_housing_B'. The unit of the vibration data is 'gravitational constant (g)' (1g = 9.80 6 65 m/s 2 ). The motor current data file contains five columns, namely 'Time Stamp', 'R_phase', 'S_phase' and 'T_phase'. The unit of the motor current is 'Ampere (A)'. To support more details in this dataset, synchronized speed data are also provided. Sample raw data and their time-frequency analysis of each state are shown in Figs. 1 to 4 . The description of the dataset is as follows: < vibration >

Section 1: Description of Testbed
The rotating machine testbed consists of three-phase induction motor, torque meter, gearbox, bearing housing A, bearing housing B, rotors and hysteresis brake as shown in Fig. 5 . The threephase induction motor manufactured by SIEMENS is four-pole AC motor with 3 horse-power (HP). It is driven at 380 V, 60 Hz, at a rated speed of 1770 rpm. The gearbox increases the rotating speed by 2.07 times, up to 3663 rpm. To avoid signal overlap with the driving frequency of 60 Hz, this dataset was operated at 3010 rpm. A load was applied to the rotating machine using a hysteresis brake (AHB-3A) manufactured by Valid Magnetic Ltd., and the load was measured with a torque meter (M425) manufactured by Datum Electronics. The simulated loads in this dataset are 0 Nm, 2 Nm, and 4 Nm. Acoustic data were collected under zero-load conditions, as the brake is an air-cooling method and can act as noise to the microphone.
A total of four accelerometers (PCB35234) were installed in the x-and y-directions of bearing housings A and B, based on the vibration installation guide (ISO 10816-1:1995). A microphone (PCB378B02) was located nearby bearing housing A based on the microphone installation guide (ISO 8528-10). Two thermocouples (K-type) were installed in each bearing housing to measure the bearing temperature. To measure the three-phase motor current, three CT sensors (Hioki CT6700) were used. CT sensors were installed on the U-phase, V-phase, and W-phase of the three-phase motor.

Section 2: Method of Fault Seedings
Rolling element bearings are composed of two concentric rings called races and rolling elements such as balls or rollers between the races. The inner and outer raceway are grooved. To assemble a ball bearing, the balls are inserted in between the inner race and the outer race. The inner race is snapped to a position concentric with the outer race. The balls are separated uniformly between the races, and a riveted cage is inserted to maintain the separation.   In varying load condition test, the bearing faults, including inner race fault and outer race fault were simulated according to the crack sizes (0.3 mm, 1.0 mm, and 3.0 mm) as shown in Fig. 6 . The corresponding fault-seeded bearing was installed in the bearing housing A. Depending on the rotating speed, the bearing faults frequencies can be calculated as described in Shaft fault is a parallel misalignment that moves the shaft in bearing housing A as shown in Fig. 7 . The movements consist of 0.1 mm, 0.3 mm, and 0.5 mm. Rotor faults are seeded by adding mass to the fourth rotor disk for simulating mass unbalance as shown in Fig. 8 . The unbalanced disk is the closest disk to bearing housing A. This dataset consists of the unbalanced masses: 583 milligram (mg), 1169 mg, 1751 mg, 2239 mg, and 3318 mg. The overall description of dataset is listed in Table 2 .
In varying speed condition test, type 6205 steel NSK ball bearing were used for testing. Four different state (normal, inner race faults, outer race faults and ball faults) of the ball bearing were emulated as shown in Fig. 9 . These faults were generated by spalling surface of ball bearing  using diamond tips. The varying speed conditions was simulated by adjusting the frequency of the motor as shown in Fig. 5 . The overall description of dataset is shown in Table 3 .

Ethics Statements
Human Lab., Center for Noise and Vibration Control Plus, Department of Mechanical Engineering, Korea Advanced Institute of Science and Technology, Daejeon, South Korea has given the consent that the datasets may be publicly-released as part of this publication.

Declaration of Competing Interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.