Artificial intelligence analysing gearbox testing data

25. 8. 2021

No-load series testing of rail vehicle gearboxes with artificial intelligence

The new workplace for testing rail gearboxes in Wikov facility is equipped with two mutually independent test beds and is able to cover the entire range of gearbox types, from spur-gear, through bevel-spur-gear, up to special planetary-bevel ones. At the same time, to minimize the influence between the first and last tested gearbox of the given series, the test room is equipped with an automated system for oil filtration and heating. Thanks to this system, gearboxes are filled with clean, pre-heated oil of a constant quantity and temperature, which is ensured by a special filling device with measuring. With regard to the run unit, every workplace is equipped with an asynchronous motor (ASM) which is able to rev up the input shaft of the gearbox up to ±7500 rpm. In connection with a frequency converter (VSD), ASM features sufficient power for the required dynamics of no-load tests.

Proprietary online monitoring and artificial intelligence

The entire driving block is supplemented with online monitoring - WiGuard®, developed right in Wikov, used for the recording and automatic evaluation of temperature and vibrations. The system identifies the gearbox according to its type and series number. From the server, it automatically downloads the limit settings of the individual sensors, intervals for the individual records and the time line of the rotations tested, on the basis of which it then sets up the requirements for the frequency converter.

During the test, the values of the measured variables and selected characteristic quantities from the frequency converter are then saved in the server storage. Simultaneously, these values are continually verified within the given test phase limits. At preset points of time, the time recording of vibrations for the individual sensors is also made to be subjected to FFT analysis and the evaluation of the vibration state on the basis of artificial intelligence. The artificial intelligence is preset to recognize gearbox defects in relation to the gearing, bearings and/or imbalance by means of neural networks. If the system detects a serious error, the given test is suspended prematurely. If the entire test has proceeds without a serious error, it is completed; the data is saved and then evaluated. Moreover, the data collection system enables the additional evaluation and comparison of gearboxes within the tested series. This comparison is based on the statistical processing of the given gearbox series data saved on the server.

Photo gallery