LSTM networks based on attention ordered neurons for gear remaining life prediction.

Gear is a commonly-used rotating part in industry, it is of great significance to predict its failure in advance, which is helpful to maintain the health of the whole machine. Firstly, the isometric mapping algorithm is applied to construct the health indicator (HI) based on the statistical characteristics of gear. Then a novel variant of long-short-term memory neural network with attention-guided ordered neurons (LSTM-AON) is constructed to achieve the accurate prediction of gear remaining useful life (RUL). LSTM-AON divides the hierarchy of health characteristic information via attention ordered neurons, so that it can use the sequence information of neurons to improve the predictive performance, which improves the long-term prediction ability and robustness. The experiments show the superiority of the new gear RUL prediction methodology based on LSTM-AON compared to the current prediction methods.

Investigation of Grain Refinement Mechanism of Nickel Single Crystal during High Pressure Torsion by Crystal Plasticity Modeling.

The excellent properties of ultra-fine grained (UFG) materials are relevant to substantial grain refinement and the corresponding induced small grains delineated by high-angle grain boundaries. The present study aims to understand the grain refinement mechanism by examining the nickel single crystal processed by high pressure torsion (HPT), a severe plastic deformation method to produce UFG materials based upon crystal plasticity finite element (CPFEM) simulations. The predicted grain maps by the developed CPFEM model are capable of capturing the prominent characteristics associated with grain refinement in HPT. The evolution of the orientation of structural elements and the rotations of crystal lattices during the HPT process of the detected differently oriented grains are extensively examined. It has been found that there are mainly two intrinsic origins of lattice rotation which cause the initial single crystal to subdivide. The correlation between the crystallographic orientation changes and lattice rotations with the grain fragmentation are analyzed and discussed in detail based on the theory of crystal plasticity.

Study on contact performance of ultrasonic-assisted grinding surface.

The contact performance of ultrasonic-assisted grinding surface is studied in this paper. An improved simplified model of rough surface profile is proposed to find the microscopic feature parameters, such as the curvature radius of the asperity, which are suitable for contact analysis and calculation. Then a more accurate rough surface contact analysis model is obtained by combining the classical ZMC contact model. Based on the contact analysis model, the contact mechanism of ultrasonic-assisted grinding surface is studied. The contact stiffness and local maximum contact pressure of the surfaces under different cutting depths and ultrasonic amplitudes are calculated, and the correlation rule between the parameters of ultrasonic-assisted grinding and the contact performance of the machined surface is obtained: (1) With the increase of the cutting depth, the surface roughness of the workpiece increases; under the same load, the contact stiffness decreases and the maximum local contact pressure increases. (2) With the increase of the ultrasonic amplitude, the surface roughness of the workpiece first decreases and then increases. Under the same load, the contact stiffness increases first and then decreases, while the maximum local contact pressure resents an opposite variation trend. Under the experimental conditions, the surface contact performance of the workpiece is the best when the ultrasonic amplitude is 4 µm. Additionally, the contact performance of the ultrasonic-assisted grinding surface and the conventional grinding surface is compared: (1) When the ultrasonic amplitude is 4 µm, the surface roughness of the workpiece is at least 24% lower than that of the conventional grinding surface. (2) Under the same load, the surface contact stiffness of the ultrasonic-assisted grinding surface is increased by at least 68%, and the maximum local contact pressure is reduced by at least 17%. It is found that the interference motion of abrasive particles in the ultrasonic-assisted grinding process makes the surface height distribution more concentrated and the density of asperity increased, which results in a better contact performance compared with the conventional grinding surface.