Search

Purpose: As it is already known, modern industry is based on CNC machines, robots, actuators etc. Most of this equipment has in its structure threaded spindles. Installation of the threaded spindles is the most demanding task in mechanical engineering. Methods: The bearings which are designed for this purpose have to meet many requirements regarding bigger number of revolutions, significant axial load, work accuracy, thermal stability etc. Results: In order to ensure thermal stability of this kind of bearings, it is necessary to predict values of the main operating parameters of the bearing: total friction torque as well as temperature. This paper presents prediction of the total friction torque and temperature by axial angular contact ball bearings designed for threaded spindles using ANN (Artificial Neural Network). Conclusion: As the main validating parameter was selected Mean Absolute Percentage Error (MAPE) which was in range 0,18- 1,32% for all ANNs. [ABSTRACT FROM AUTHOR]

For a given workpiece material, in addition to the cutting tool material used and its geometry, a set of adopted cutting parameters significantly affect the actual turning process. With the ultimate aim of improving machining efficiency while ensuring the specified part quality and lowest possible energy consumption and costs, the optimization of the turning regimes is of great importance. Therefore, a multi-objective optimization model for the multi-pass turning of grey cast iron was developed in the present research. Part production costs and consumed energy were both considered as conflicting objective functions, while several practical constraints related to the part quality, machine tool limitations, and product specifications in finishing and roughing were also included in the optimization model. The formulation of the objective functions was based on the analytical model for the unit production time whose validation was performed using toolpath simulation in SinuTrain and FeatureCAM software. In order to obtain Pareto compromise solutions in a reasonable computational time as well as to avoid getting stuck in local optima in the optimization hyper-space, a deterministic approach based on the application of the brute force optimization algorithm is proposed to solve the developed multi-objective optimization problem with several constraints of the equality and inequality type. The obtained optimization results were validated experimentally and through a comparison with related analytical and empirical turning optimization studies. [ABSTRACT FROM AUTHOR]

Surface roughness is one of the main indicators of quality assessment of finished parts processed by laser cutting. This paper presents the experimental results regarding surface roughness of laser cuts in high-power CO2 laser cutting of AA5754 aluminum alloy by using nitrogen as the assist gas. Based on carrying out the full factorial design experiment, the collected data were used for the development of an artificial neural network prediction model of surface roughness in terms of laser cutting parameters such as cutting speed, laser power and assist gas pressure. In addition to the modeling and analysis of the interdependencies between the considered process inputs and surface roughness, this paper presents results regarding single- and multi-objective optimization, determined by the use of a genetic algorithm, by considering the surface roughness as the main criterion, as well as other criteria such as kerf width, assist gas consumption and material removal rate. [ABSTRACT FROM AUTHOR]

Solving problems of stress-strain analysis in plastic deformation processes can be carried out by many methods. Almost all solutions can be reduced to an analysis of theoretical and experimental results, obtained by some mathematical models as well as by experimental testing of the plastic deforming process. The ultimate goal is to determine the stress-strain state, the field of strain rate, and deformation velocity within the volume of the material that is being plastically deformed. The aim of this investigation is to determine the parameters given earlier by tracing the grain structure microdeformations of a low carbon steel in a meridian cross section at ambient temperature by means of an artificial neural network (ANN) during bulk forming. The three key parameters selected for the description of the deformed microstructure are the angle of rotation and the major and the minor axes of the ideal grain, which are used to represent the plastic deformation in a selected point. The ideal grain model, for the selected point of the meridian cross section, represents the plastic deformation in line with the selected parameters for the defined number of ferrite grains. The ANN models were developed using steel microstructure data, which were obtained experimentally by using three experimental tool dies for forward extrusion. Their verification was carried out on three different angles of extrusion. This method yielded the size of the plastic deformation of the grain structure in the meridian cross section during forward extrusion, which can serve as a basis for further stress-strain analysis. [ABSTRACT FROM AUTHOR]