Application of Artificial Neural Networks forMonitoring Conditions ofLiquid Friction Mechatron Bearing under Temperature-Viscosity WedgeConditions

Purpose of research.One of the basic conditions for liquid friction occurrence is a variable gap, which is usually referred to as a geometric wedge. But this condition is not necessary and the effect of a geometric wedge can be replaced by a viscous wedge. The effect of a viscous wedge for Newtonian liquids can be caused by temperature non-uniformity. And non-uniformity of strain rate tensor can be an additional causefor non-Newtonianliquids. Thus, it is possible to provide an additional bearing capacity in the fluid friction supportby controlling the tempering field in the lubricating layer. It is also possible to minimize friction powerloss. The purpose of this work is to create controlled temperature-viscosity wedge in a fluid friction bearing. Physical realization of this effect is achieved by a multi-zone supply of variable temperature lubricant.

Methods.Experimental study planning and organization were used.Results analysis was carried out by means of paths construction and AFRC oscillatory patterns. Modern methods of machine learning are used in order to solve sensory and program determination problem of bearing and lubrication conditions.Artificial neural network of direct propagation with logistical activation functions has been developed.This network allows determining the method of lubricant supply from measurements of rotor vibration displacements and fluid pressure in the bearing. Methods of linear algebra and unconditional optimization methods are used as supplementary.

Results. Experimental apparatus in the form of rotor-support system with multizone lubricant supply with information-measuring system, which can receive rotor vibration displacement and liquid pressure supply has been developed. Mathematical model in the form of artificial neural network of direct distribution with one hidden layer was developed to monitor bearing and lubrication conditions.

Conclusion.Artificial temperature-viscosity wedge and multi-zone supply of lubricant to fluid friction bearing have a significant effect on hydro mechanical system. This effect, with an accuracy of more than 95%, was defined by an artificial neural network after processing the rotor vibration displacement and bearing fluid pressure measurements.

1. Wang Q.J., Chung Y.-H. Encyclopediaoftribology. New York: Springer Science + Business Media; 2013, 4192 p.

2. Young J. Thermal wedge effect in hydrodynamic lubrication.The Engineering Journal,1962,vol. 45,pp. 46–54.

3. Meng Xi, Khonsari M.M. On the effect of viscosity wedge in microtextured parallel surfaces.Tribology International,2017,vol. 107,pp. 116 – 124. http://dx.doi.org/10.1016/j.triboint.2016.11.007.

4. Kornaev, A.V., KornaevaE.P., Savin L.A. Theoretical premises of thermal wedge effect in fluid-film bearings supplied with a non-homogeneous lubricant.International Journal of Mechanics,2017,vol. 11,pp. 197-203.

5. Kornaeva E.P., Kornaev A.V., Savin L.A. Modelirovanie neizotermicheskogo techeniya vyazkoi zhidkosti v konfuzornykh kanalakh v usloviyakh mnogozonnoi podachi smazochnogo materiala [Modelling of the Nonisothermal Current of Viscous Liquid in Konfuzorny Channels in the Conditions of Multizonal Giving Lubricant]. Izvestiya Yugo-Zapadnogo gosudarstvennogo universiteta = Proceedings of the Southwest State University, 2018, vol. 22, no. 5(80), pp. 109-118(In Russ.).

6. Xi. Li, Zhang W., Ding Q. Deep learning-based remaining useful life estimation of bearings using multi-scale feature extraction.Reliability Engineering and System Safety, 2019,vol.182,pp. 208-218.

7. Wang X., JiangB., LuN., Zhang Ch. Dynamic fault prognosis for multivariate degradation process.Neurocomputing, 2018, 275,pp. 1112-1120.

8. Sulimov V. D., Shkapov P. M., Sulimov A.V. Geometricheskie struktury i vy-chislitel'naya diagnostika dinamicheskikh sistem s ispol'zovaniem gibridnykh algoritmov [Geometric structures and numerical diagnostics of dynamic systems using hybrid algorithms]. Fundamental'nye i prikladnye problemy tekhniki i tekhnologii =Fundamental and applied problems of engineering and technology, 2018,no. 4-1 (330),pp. 3-13 (In Russ.).

9. GoodfellowI., BengioYo., CourvilleA. DeepLearning. MIT Press, 2016, 802 p.

10. Hori Y. Hydrodynamic Lubricatio. Tokyo: Yokendo Ltd., 2006, 239 p.

11. Eling R., Wierik M., Ostayen R., Rixen D. Towards Accurate Prediction of Unbalance Response, Oil Whirl and Oil Whip of Flexible Rotors Supported by Hydrodynamic Bearings. Lubricants, 2016,vol. 4, 18 p.https://doi.org/10.3390/lubricants4030033.

12. YamamotoT.Linear and nonlinear rotordynamics. A modern treatment with applications. New York, John Wiley&Sons, 2001.

13. ChildsD. Turbomachinery rotordynamics. Phenomena, modeling, and analysis. New York; John Wiley&Sons, 1993. ISBN 0-471-53840-X.