AMPERE: Detection and diagnostics of rotor and stator faults in rotating machines (2023)

Data collector : Guy Clerc [4], Hubert Razik [4]
[1] : Franche-Comté Electronique Mécanique Thermique et Optique - Sciences et Technologies (UMR 6174) (Université de Franche-Comté)
[2] : Laboratoire d'Analyse des Signaux et des Processus Industriels
[3] : Laboratoire Génie de production
[4] : Laboratoire Ampère
Description :
Dataset of speed, current, voltage and vibration measurements of an electromechanical drive system. The system is a three-phase asynchronous motor. It is composed of an aging bearing and a squirrel cage rotor with defects (broken bars). In addition, unbalanced power supply faults are investigated. All experiments are performed under different local levels.
Disciplines :
computer science, artificial intelligence (engineering science), engineering, electrical & electronic (engineering science), engineering, industrial (engineering science), engineering, mechanical (engineering science), engineering, multidisciplinary (engineering science)

General metadata

Data acquisition date : from 11 Mar 2006 to 18 Mar 2006
Data acquisition methods :
  • Experimental data :
    The AMPERE laboratory test bench is designed to monitor two parts, 1) monitor the rotor of the motor, 2) monitor the stator of the motor. Hence, there are two sets of data corresponding to two different monitoring cases. The first set concerns the monitoring of the electrical bars and the bearing at the rotor level with data from the inverter output. The second dataset corresponds to the monitoring of the motor stator windings with data collected from the power grid.

    Hence, this test bench is composed of a three-phase inverter to supply and control a 5.5 kW electric motor. The motor used is a squirrel cage motor and its rotating shaft is connected to an electromagnetic brake which operates as a load on the motor. This brake, achieved at a nominal speed, is designed to dissipate a maximum power of 5 kW with a maximum brake torque of 100 N.m.

    For the first set of data, the whole test bench is equipped with several heterogeneous sensors placed at different positions. There are three-phase current and voltage sensors corresponding to the three phases of the motor. Furthermore, three separate accelerometers are placed on the motor. The first one is in vertical direction on the opposite side of the coupling, the second one is in parallel direction to the motor axis on the opposite side of the coupling and the last one is in horizontal direction and perpendicular to the motor axis on the opposite side of the coupling. Finally, an encoder placed at the output of the brake is used to measure the rotation speed. The acquisition system, Odyssey GOuld Nicolet, used to measure these signals has eight differential inputs, known as fast channels, which can be sampled at up to 10 MHz on 14 bits. These inputs were used to measure the voltages as well as the currents. In addition, there are eight common mode inputs, called slow channels, on which the sampling frequency can reach 1 MHz on 16 bits. One of these eight inputs was used to measure the rotation speed. Then, three inputs are reserved for accelerometers to measure the vibration. The accelerometer inputs also offer the possibility of a 100 kHz sampling rate. They are powered by constant current sources providing between 1 and 10 mA at 28 V.

    All acquisitions are performed in the permanent mode over a period of 5 seconds with a sampling frequency of 20 kHz. The data are stored in .mat files with 11 columns. Column 1 represents the time steps of the acquisition, column 2 to column 4 represent the three-phase voltage signals, column 5 to column 7 are the three-phase current signals, column 8 is the peed, and the last three columns, from column 9 to 11, are for the three accelerometers.

    Regarding the monitoring dataset of the electrical bars and the motor bearing, 25 experiments are carried out with data collected at the output of the inverter. In detail, there are 5 different health states of the motor, a healthy state with a healthy rotor, a breakage of one rotor bar, a breakage of three rotor bars, breakages of four rotor bars, and a degraded bearing. In each health state, 5 load levels were applied.

    For the second set of data, the data are carried out from the power grid supply for monitoring the stator behavior. In this case, different resistances are added to one of the stator windings in order to reproduce a short-circuit between the windings of the b-phase of the motor and cause a supply unbalance. Besides, the composition of the test bench remains unchanged from what has been described previously. Also, the data collection procedure remains the same, i.e. the sampling frequency is equal to 20 kHz with an acquisition period/file of 5 seconds in .mat files. In these files, column 1 represents the acquisition time, column 2 to column 4 represent the signals of the three-phase currents, column 5 to column 7 contain the voltage measurements, column 8 is for the instantaneous motor rotation speed, and column 9 to 11 are for the accelerometers.

    In total, 25 experiments were performed for the monitoring the stator power supply with data collected from the supply grid. There are 5 different health states of the motor, a healthy state of the stator, 5%, 10%, 20\% and 40% srotor unbalance. In each health state, 5 load levels were applied 0%, 25%, 50%, and 75%.

    Two .pptx files containing the test bench figure and its scheme can be found in the AMPERE dataset repository by clicking on "ACCESS TO DATA" on the home page of the site.
Update periodicity : no update
Language : English (eng)
Formats : text/x-matlab
Audience : University: master, Research, Informal Education
Publications :
  • Ondel, O., Boutleux, E., & Clerc, G. (2006). A method to detect broken bars in induction machine using pattern recognition techniques. IEEE Transactions on industry applications, 42(4), 916-923 (doi:10.1109/TIA.2006.876071)
  • Soualhi, A., Clerc, G., & Razik, H. (2012). Detection and diagnosis of faults in induction motor using an improved artificial ant clustering technique. IEEE Transactions on Industrial Electronics, 60(9), 4053-4062. (doi:10.1109/TIE.2012.2230598)
  • Ondel, O., Blanco, E., & Clerc, G. (2007, September). Beyond the diagnosis: the forecast of state system Application in an induction machine. In 2007 IEEE International Symposium on Diagnostics for Electric Machines, Power Electronics and Drives (pp. 491-496). IEEE. (doi:10.1109/DEMPED.2007.4393143)
  • Soualhi, M., Nguyen, K. T., & Medjaher, K. (2020). Pattern recognition method of fault diagnostics based on a new health indicator for smart manufacturing. Mechanical Systems and Signal Processing, 142, 106680 (doi:10.1016/j.ymssp.2020.106680)
  • Breuneval, R., Clerc, G., Nahid-Mobarakeh, B., & Mansouri, B. (2018). Classification with automatic detection of unknown classes based on SVM and fuzzy MBF: Application to motor diagnosis. AIMS Electronics and Electrical Engineering, 2(3), 59-84. (doi:10.3934/ElectrEng.2018.3.59)
  • Casimir, R., Boutleux, E., Clerc, G., & Yahoui, A. (2006). The use of features selection and nearest neighbors rule for faults diagnostic in induction motors. Engineering Applications of Artificial Intelligence, 19(2), 169-177. (doi:10.1016/j.engappai.2005.07.00)
Collection :
Publisher : Laboratoire Ampère
Additional information :
The platform is installed at the Ampère Laboratory, in Lyon, France.
Record created 6 Mar 2023 by Moncef Soualhi.
Last modification : 25 Sep 2023.
Local identifier: FR-13002091000019-2023-03-06-03.


dat@FEMTO-ST is a sub-portal of dat@UBFC, a metadata catalogue for research data produced at UBFC.

Terms of use
Université de Bourgogne, Université de Franche-Comté, UTBM, AgroSup Dijon, ENSMM, BSB, Arts des Metiers