Fulltext search in archive

Today, machine learning regression methods are quietly but fundamentally transforming cost and time estimation in manufacturing: from early pricing to labor planning to operational order management. This survey offers a comprehensive map of approaches - from linear models, to tree ensembles (RF, GBM, XGBoost) and shallow neural networks, to multi-target and tensor regressions that can exploit data structure across BOM items and sequences of operations. With an emphasis on SME conditions, we show how to reconcile three often conflicting requirements of practice: accuracy, explainability, and integration into existing data flows (MES/ERP). The paper presents a comparative taxonomy of methods, recommended validation practices (MAE, RMSE, MAPE, R² including confidence intervals) and a pragmatic adoption trajectory: from regularized multiple regressions to tree models to multi-output formulations sharing re-presentations across operations. Consolidated findings show that modern learners consistently outperform traditional baselines when supported by careful flag engineering, drift management, and data standardization. As a major research-application contribution, we propose a unified multi-objective framework for simultaneous cost and time prediction that combines domain (queueing/simulation) features with data-driven regression to enable transparent decision making in pricing and capacity planning. The study thus creates a bridge between theory and manufacturing practice and invites the reader to systematically but achievably deploy ML in everyday decision making.

The article described a new methodology for testing the torsional resistance of a single-stage gear transmission used in agricultural machinery. The analysis encompassed the entire mechanical system rather than focusing solely on its individual components. The research identified three key ranges of structural resistance. The first range, with twist angles from 0° to 1.85° and torques up to 1050 Nm, was associated with the elimination of structural play and the alignment of contact surfaces. The second range, from 1.85° to 4.76° and torques up to 3450 Nm, confirmed the resilient behavior of the gearbox according to Hooke's law. In this range, the system worked stably and maintained repeatability of parameters. The third range, above 4.76° and 3050 Nm, showed the presence of permanent but local deformations. However, these displacements did not affect the functionality of the system in less demanding applications. The maximum torque of 5500 Nm did not cause macroscopic damage or oil leaks, which proves the high quality of the design and the effectiveness of material optimization. The developed method allows for an accurate determination of the safety factor and a detailed assessment of the strength properties. It can be used to optimize transmissions in various sectors such as agriculture, automotive and aerospace. The results also form the basis for further experiments, including fatigue tests and contact stress analyses. The proposed methodology enhances the predictive accuracy of gearbox durability under various load conditions. These advancements support the development of sustainable and efficient mechanical systems across multiple industries.

The study focuses on the evaluation of the degradation behaviour of structural materials used in a heat exchanger boiler exposed to elevated thermal, pressure, and cyclic loading conditions. The research is aimed at non alloy pressure steels P235GH, P265GH, and P355GH employed in critical boiler components. The objective of the study was to analyse the effect of temperature in the range of 250–600 °C on the residual mechanical properties, plastic deformation, and fatigue behaviour of these materials. The results indicate that at temperatures above 300–400 °C, a significant degradation of mechanical properties occurs. The residual strength of P235GH steel decreases by more than 60 % at 400 °C, while the plastic deformation of P235GH and P265GH steels is reduced to 5–8 %, representing a critical threshold from the perspective of fatigue damage. Steel grade P355GH exhibits higher thermal stability; however, at 400 °C its yield strength decreases to approximately 195 MPa. Based on the obtained results, an optimised material concept was proposed utilising heat resistant Cr–Mo steels 16Mo3, 13CrMo4 5, and 10CrMo9 10, which retain 50–100 % higher plastic deformation and significantly greater creep resistance at temperatures of 500–600 °C compared to the original materials.

Vehicle Routing Problem is a common problem in logistics, which can simulate in-plant and out-plant material handling. In the article, we demonstrate a Vehicle Routing Problem, which contains period, time window and multiple depots. In this case, customers must be served from several depots. The position of the nodes (depots and customers), the demand and time window of the customers are known in advance. The number and capacity constraint of vehicles are predefined. The vehicles leave from one depot, visit some customers and then return to the depot. The above-described vehicle routing is solved with construction algorithms and Ant Colony algorithms. The Ant Colony algorithms are used to improve random solutions and solutions generated with construction algorithms. According to the test results the Elitist Strategy Ant System and the Rank-Based Version of Ant System algorithms gave the best solutions.

The article deals with the evaluation of the amount of wear of the base material and selected hardfac-ing materials intended for tools for wood processing in forestry after a test of resistance to abrasive wear in laboratory conditions. The values of average weight loss Wh[g] and relative resistance to abrasive wear Ψh[-] were determined by calculation. The topography of the surface after the track of the rubber disc and the abrasive of the testing device was evaluated with a confocal microscope. The depth of the disc track Pt[μm;mm] was also evaluated with a confocal microscope. The state of the samples surface after the test, as well as the overall structure and mixing of the hardwearing material with the base material was evaluated by light microscopy. A touch roughness meter was used to de-termine the profile of the track surface after the test. Based on the results, we can recommend certain hardfacing materials for practice. Their abrasive resistance and thus also the loss of material during the work load could ensure a longer service life of the tool.

This paper aims to study the efficiency of using prefabricated layers made from plain cementitious composite materials for enhancing the flexural behavior of reinforced concrete (RC) continuous beams. The strengthening system was applied at 20 mm thickness, 150 mm width, and adequate development length. The prefabricated layers were placed in the tension cover in the positive and negative zones. All beams have the same geometric dimensions and positive and negative steel reinforcement ratios. The results showed that the prefabricated layer was deformed with the RC specimen without debonding, which enhanced the cracking patterns and distributed the crack width. A slight improvement in the strengthened beam capacity was 7% for the yielding load and 6% for the ultimate load. The energy absorption capacity of the strengthened beam decreased by 30.67%, whereas both beams achieved the same ductility index.

Dynamic mechanical analysis (DMA) is an important method for evaluating the viscoelastic properties of polymeric materials, especially when investigating their mechanical response to various manufacturing parameters and surface treatments. In recent years, DMA analysis has been intensively used, among others, for the analysis of polylactide (PLA) produced by the fused filament fabrication (FFF) additive technology. The present study focuses on the effect of DCSBD plasma treatment on the dynamic-mechanical properties of PLA samples with different infill geometries (Line, Rectilinear and Concentric). In the study, experimental PLA samples were subjected to DMA analysis in the temperature range of 40 °C to 90 °C in order to analyze the changes in their viscoelastic properties after plasma discharge surface treatment. The results showed a decrease in the glass transition temperature (Tg) for all tested samples, while the extent of the decrease depended on the infill geometry used. The most significant changes were observed in samples with Rectilinear infill, which showed the best mechanical stability after plasma treatment. The study shows that plasma treatment can influence the mechanical properties of PLA products, opening new possibilities for optimizing their processing, reuse and application in technical areas requiring controlled mechanical response.

The development and characterizing of thin layers of AgCu, Cu, Ti, and Zr on nylon filters using PVD magnetron sputtering technology was conducted. The evaluation of these thin layers was mainly focused on characterizing specific parameters that may influence the expected functionality of the modified filter materials. The surface treatment of nylon filters with thin layers does not significantly affect the mechanical properties of the original nylon material. Thin layers deposited at a power of 0.9 kW exhibited greater thickness and lower static friction coefficient values than the layers deposited at 0.4 kW, except for a thin layer of the element titanium. The surface modification of the filters did not significantly change resistance to deformation and had no significant reduction in pore size. However, a significant effect on surface wettability (increased hydrophobicity) was demonstrated.

The article deals with sinterhardening process of lean Cr-Mo prealloyed steel for moderately loaded applications. New material Astaloy CrS with low alloying volume of chromium and molybdenum was analyzed as possible basis for sinterhardening process. Standard mechanical properties of frequently used and more expensive materials such as DistaloyDH and Astaloy CrM are chosen as a compara-tive criterion. Astaloy CrS+0.85%C samples with different compaction densities and Ni content were studied, mechanical properties and hardness after sinterhardening process were compared. The influ-ence of additional high-temperature sintering on mechanical properties was assessed. The micro-structure of the sinterhardening (SH) and high-temperature sintering + sinterhardening (HTS+SH) samples was studied quantitative analysis of the phase was given. As result, tensile strength greater than 900 MPa and hardness greater than 33 HRC can be obtained for investigated material.

Today, considerable attention is paid to the production of solid castings (approx. 2000 kg) from cast iron with spheroidal graphite. The metallurgical preparation of large quantities of melt is very difficult. This difficulty is related not only to the melting and preparation of large quantities of melt, but above all to its metallurgical treatment - inoculation and modification. Melt modification ensures the production of cast iron with spheroidal graphite. Material castings, such as machine tool components, cannot be destroyed to determine the quality of the cast iron produced. Therefore, this paper outlines a methodology to proceed in determining the quality of manufactured castings. It is possible to observe the chemical composition of cast iron, thermal analysis of cast iron using liquidus temperature value, subcooling temperature, eutectic recalescence, primary solidification recalescence, eutectic solidification time. Furthermore, to observe the mechanical values of cast iron (yield strength, ultimate strength and ductility) on fabricated bars of overmolded Y blocks or to observe the micro-structure of cast iron on microscope.

Core-shell powders have been extensively studied due to their complex structure and wide range of applications. W@Ag core-shell powders are particularly interesting due to the synergy between the tungsten and silver, which can be beneficial in the electronics industry. However, knowledge of their thermal stability is limited, particularly concerning the impact of annealing temperatures on structural integrity and oxidation resistance. In this work, W@Ag core-shell powder was heat-treated in the temperature range 100–700 °C for 1 h in air. Investigation of the microstructural changes using scanning electron microscopy equipped with energy-dispersive X-ray spectroscopy showed that the limiting temperature is 500 °C, when the shell began to decompose and the core began to oxidize. Moreover, X-ray diffraction analysis determined that the phase composition of the thus heat-treated material consisted of approxi-mately 50 % Ag and 50 % Ag2WO4.

The present paper discusses important aspects of residual stress measurements in forged shafts with defects using the X-ray method. A random population of shafts was selected for the study, for which, depending on the type of rolling process, turning was performed, measuring stress changes after successive machining passes. In the forged shafts studied, the existence and location of internal defects were identified using computed tomography. The impact of internal defects on the stress distribution on the surface of the machined workpiece was observed. It was observed that the use of the X-ray method to measure residual stresses makes it possible to determine the state of stresses and their distribution, which is crucial for the safety and durability of shaft-type parts, and allows the impact of a defect on the distribution of residual stresses to be identified. On the basis of the results obtained, it was observed that there is a correlation between the occurrence of internal defects in forged shafts and the distribution of residual stresses in characteristic sections along the length of the shaft after machining

The article describes the creation of a mathematical 3D model of the original hovercraft structure, which will be further used for research into modifying the shapes and materials of the structure to ensure better driving conditions. Proposals for new materials for individual parts of the hovercraft structure will be addressed in order to reduce the weight of the hovercraft and thereby ensure a higher possible speed of movement, reduce fuel consumption and ensure the necessary mechanical properties of individual segments. The mathematical model of the simplified hovercraft model was created in the Cradle and Adams simulation programs. The paper is presented by analyzing the hovercraft properties in order to obtain sets of advantages and disadvantages of the hovercraft. The following is a description of the creation of a geometric 3D model of the hovercraft, which is built using Autodesk Inventor. The article further describes the transformation of the 3D model into a simulation model that can be used for co-simulation of movement in the Adams and Cradle computer simulation systems. The simulations will be the first step towards modifying the structure of a real rescue UAV prototype with improved maneuverability, stability and the ability to traverse terrain with surfaces unsuitable for hovering.

This study centers on 7B50 aluminium alloy. The intention is to reduce the pre-treatment and post-treatment times of the shot peening model. By comparing and analyzing different process parameters, the best combination of peening solutions can be obtained. The pre-processing is implemented through a GUI interactive interface. Post-processing is carried out by using Python for secondary development in ABAQUS. Orthogonal test method is employed for post-processing analysis of shot peening simulations under various process conditions. The results are evaluated by using a weighted composite scoring method to determine the depth of the residual compressive stress layer on the workpiece surface, the surface residual compressive stress, and the extreme deviation of the maximum residual compressive stress value after shot peening. The combined influence degree of shot peening process parameters such as impact speed, projectile diameter and impact angle is determined. The optimal combination of shot peening process parameters is analyzed and verified through simulation.

The paper focuses on the application of machine learning techniques and optimization algorithms in predictions and controls of grinding temperature variations. The major thrust of investigation has been on how the different input conditions such as feed, depth of cut, and cooling conditions influence grinding temperatures and the effectiveness of these conditions on the control of their thermal effects. Three machine learning models: Random Forest (RF), Gradient Boosting (GB), and Artificial Neural Networks (ANN) were then used to develop prediction models for the grinding temperature on both face and shoulder of the workpiece. Out of all the models, RF achieved a much higher R² score of 0.96 as compared to both GB and ANN, indicating its greater predictive performance. Furthermore, Bayesian optimization and genetic algorithms were employed in model optimization and grind parameters and cooling condition optimization to avoid damages caused due to temperature. MQL has been found to be highly superior to the inefficient dry cooling methods in terms of achieving lower grinding temperatures and, therefore, seems to be most suited as an eco-friendly yet practical cooling solution as based on this comparison. Altogether, these research findings indicate that AI-based techniques and traditional optimization methods can lead to much better grinding in terms of efficiency and energy consumption, as well as surface quality, and assist towards greener manufacturing altogether.

Robotics plays a key role in industry and its use continues to grow. Robots are used in many industries to increase efficiency, productivity, and safety of work processes. This manuscript focuses on the spatial calibration of collaborative robot arms using appropriate statistical tools. Nowadays, there are many special programming languages, simulations or virtual realities (VR), which in most cases perform calibration using matrix relations. The mathematical-statistical solution is not solved very often, and the use of linear relationships is valid only in certain parts of the workspace of the collaborative robot. The purpose of this article is to demonstrate how to find a suitable statistical method that would respect the wear of the arm mechanism in predefined positions based on the requirements of ISO 230-2:2015. Based on these measurements, it is possible to assume that optimal solutions can be obtained using a polynomial regression function. This optimization method will be searched using the Newton and Markwartel methods.

This research focused on the production of brake pads reinforced with bronze fibers to see the anticipated performance principles for braking systems. Three unique amalgamated formulations, labeled BRZ-I, BRZ-II, and BRZ-III, were set by varying the bronze fiber content to 5%, 10%, and 15% by weight. The tribological characteristics of these composites were systematically evaluated to determine their effectiveness. Traditional manufacturing processes were used in developing the brake pad. Various properties such as physical, chemical, mechanical and tribological possessions were assessed by means of chase test rig. Worn-superficial examination stayed carried out by using chase test rig. Base results it was evident that the 10 weight percentages of the bronze fibers showed better physical, chemical, mechanical and tribological properties. Chase test results confirmed that the composite brake pad developed with 10 weight percentages of bronze showed better results at higher pressure-speed conditions than others due to better plateau formation and less wear rate. The results obtained after performing various performances such as physical, chemical, mechanical and tribological properties concluded that the bronze fiber possessed lesser wear and stable coefficient of friction.

An innovative way of using DCPD (Direct Current Potential Drop) method for off-line and online monitoring of critical parts of energy equipment in operation is presented. There are only a few NDT methods that allow detection and monitoring of defect growth in components at high temperatures and pressures. Monitoring of steam pipes and critical pipeline components in operation has been carried out for several years with different results. a relatively new way of using the DCPD method outside the laboratory is described. The carried-out tests were intended to resemble operational loads as much as possible. Therefore, the tests were performed at a temperature of 20 °C and at an increased temperature of 550 °C. By gradually deepening the groove (slot) simulating the crack type defect in predefined steps, the growth of the defect was simulated up to the full wall thickness of the test sample. The primary evaluation was carried out from the absolute and relative values of measured resistance. The disadvantage of these values is their dependence on the temperature of the monitored area of the test sample and on possibly interfering DC voltages.

The paper presents work aimed at building practical applications of virtual reality (VR) in manufacturing environments. It contains studies of the optical properties of cameras and lenses aimed at the selection of an optimal set (camera, adapter, lens) for the realization of recordings and video transmissions in stereoscopic format for VR. In response to the increasing trend in the number of applications of VR systems in the industry, works have been initiated with the purpose of building a system levelling image noise identified thus far as an obstacle to the effective utilization of VR in production systems. It was considered that picture error correction can significantly increase an already big data stream from the recordings. Based on it, a set of parameter values was defined which determined the selection of study equipment. Three research areas were set: the verification of the optical correctness, the study of image defects and their correction and the determination of the maximum optical resolution and the achievable image parameters in various lighting and environmental conditions. An example was presented for the application of a projected system for the monitoring of undesirable events/movement at work stands and key areas of production halls as well as training in the high-risk production zones.

The study focuses on an evaluation of mechanical properties of the H13 tool steel manufactured by the material extrusion and further comparison with conventionally produced material. Notably, for achieving sufficient surface quality of functional parts further post-processing is required. Thus, a comprehensive investigation, encompassing hardness, ultimate tensile strength (UTS) and yield strength (YS) measurement, microstructure, and machinability was performed. The material extrusion, an increasingly utilized additive manufacturing (AM) technique, offers a viable alternative to the prevalent laser powder bed fusion (LPBF) methods. This method enables a creation of complex geometries using various materials. The investigation revealed that the horizontal orientation of parts yielded the highest mechanical properties, reaching the ultimate tensile strength of approximately 1200 MPa. Additionally, the material exhibited the hardness of 47 HRC in the as-built state. The conventionally produced steel resulted in the higher UTS and YS in comparison to the AM material. The machinability of the as-built material in regard to cutting forces and surface roughness was also evaluated Lower surface roughness was achieved by decreasing feed per tooth. Optically measure material porosity was 6.13 % with maximum pore size 7.43 µm. The primary objective of this research is to optimize the mechanical properties of H13 tool steel post-printing, with a broader aim to apply the gained insights to improve other materials produced by the material extrusion.

Biodegradable zinc-based alloys have recently attracted attention as promising candidates for temporary implant applications due to their favourable corrosion behaviour and biocompatibility. In this study, three materials — pure Zn, Zn–1Mg alloy, and a Zn + Zn–1Mg composite — were fabricated via powder metallurgy and extrusion to evaluate their microstructural characteristics and tensile performance. The composite material was designed to combine ductile Zn regions with a reinforcing Zn–1Mg network, aiming to achieve a balance of strength and ductility. Microstructural analysis revealed coarse-grained Zn regions surrounded by ultrafine-grained Zn–1Mg areas containing Mg₂Zn₁₁ particles, with oxide shells present at the Zn/Zn–1Mg interfaces. Tensile testing showed improvement in mechanical performance compared to the individual constituents. However, the oxide shells prevented effective load transfer between the fine-grained and coarse-grained areas of the microstructure.

The precise setting of the zero point represents a critical factor in non-contact measurement of mechani-cal components, particularly in areas such as the engineering and automotive industries, where high accuracy is key to quality control. This study analyzes the impact of various alignment methods—specifically the best-fit method and the datum method (3–2–1)—on the measurement results of complex geometric shapes. Experimental measurements were conducted using a laser scanner and Polyworks 2015 software. The results indicate that the best-fit method achieves higher accuracy when measuring complex and freely oriented shapes, while the 3–2–1 method provides more consistent results for simply defined geometries. These findings confirm the importance of proper alignment method selection in op-timizing non-contact measurement processes and offer new insights for improving efficiency in industrial quality control.

The thesis deals with the surface treatments of bearing steel processed for wind turbines, on which the quality parameters of the surface treatments performed were compared. This is blackening, which is a method of surface treatment that allows the protection of the base material from the negative effects of external influences, in particular from moisture and associated corrosion. The application of surface treatment by blackening contributes to a better and more efficient start-up of the bearing in service. In the experimental part, the individual results of the structural analysis carried out for all types of materials investigated are evaluated, with the analysis focusing on the structural properties, the quality of the adhesion properties and the influence on the service life of the machine components. Electron microscopy was used to investigate the structural properties of the layer as well as the base material, which allowed to obtain the necessary data to meet the objectives of this work.

To design an engine intake system that complies with FSC racing regulations while achieving enhanced operational stability, this study conducts a comprehensive review of domestic and international research advancements in racing engine intake systems. Through computational fluid dynamics simulations performed in Workbench Fluent, critical structural parameters of the restrictor valve were optimized, resulting in a 12.06% improvement in outlet mass flow rate compared to the baseline design. A three-dimensional parametric model of the racing intake system was developed using Siemens NX platform. Taking the intake plenum chamber as a representative component, this research systematically analyzes the CNC machining process for the mold of the pressure stabilization chamber. The investigation encompasses toolpath generation, cutting simulation verification, and ultimately implements the optimized NC program on machining centers for physical manufacturing. The fabricated mold exhibits high dimensional accuracy and superior surface finish, providing both theoretical guidance and practical manufacturing references for intake system development. This integrated approach combining numerical optimization with advanced manufacturing techniques demonstrates significant potential for performance enhancement in motorsport engineering applications.

The article presents the possibilities of using wavelet transform and fast Fourier analysis (FFT) to evaluate the signal collected during roughness measurement. During the tests, high-density polyeth-ylene was turned using variable cutting parameters. During cutting, the tool feed was changed to ob-tain roughness structures of different types and with varying degrees of anisotropy. The measured roughness profiles were filtered with Daubechies 6 (db6), Morlet and "Mexican Hat" wavelets and examined using Fourier analysis. The research carried out shows how the machining conditions affect the surface condition and the stability of the cutting process under variable machining conditions for high molecular weight polymers. The effectiveness of the continuous wavelet transform (CWT), sup-plemented with data obtained from Fourier analysis, in identifying places and detecting the nature of disturbances in the generated roughness signal is also shown.

Geometric deviations play a crucial role in the quality of additive manufacturing, particularly in parts made with biodegradable resins. Accurately controlling dimensional and geometric variations in manufactured components is critical for achieving defect-free production and meeting functional standards. However, defining a final quality score can be challenging due to numerous dimensional and geometric deviations associated with a part. An innovative metric for evaluating geometric performance was created to measure dimensional precision in components produced through VAT photopolymerization. The index measures the dimensional and geometrical deviations, revealing that external surfaces exhibit greater precision than internal ones. This difference is likely due to internal surfaces overcoming heat dissipation challenges during the cooling process, resulting in less shrinkage for external surfaces. This index is essential in various stages of the manufacturing process, including part design, design for manufacturing and assembly, quality assurance, and process planning, helping to select the appropriate additive manufacturing technology and optimal process parameters.

This study used JMatPro software to comprehensively analyze the new low-iron-loss cold-rolled non-oriented high-grade electrical steel 35W210X, calculating phase composition, Gibbs free energy, stress-strain relationships, and yield strength changes. Results showed its ferritic structure and consistent calculated room-temperature yield strength with experiments. To study production cracks, JMatPro data was used in Deform-3D to simulate the five-pass reciprocating cold rolling on a Sendzimir 20-roll mill, successfully replicating the cracks. Aiming at the problems of frequent cracking and low yield rate (<50%), the study found the original single normalizing annealing process inadequate. Thus, an optimized double annealing process was adopted, controlling cracks and raising the yield rate to over 85%. This research offers theoretical and technological support for rolling high-silicon electrical steels like 35W210X.

Recently, engineering polymers like PMMA have increasingly replaced traditional materials in industry where feasible, with CO2 laser cutting gaining attention for its high quality and speed in processing these materials. Achieving precise cuts is crucial for product accuracy, with kerf width serving as a key quality attribute to ensure quality and functionality of the final product. This study focuses on the im-pact of three critical process variables: stand-off distance, laser power, and cutting speed, on the kerf width in CO2 laser cutting of PMMA. Through a full-factorial experiment, the process parameters are systematically varied to understand their individual and interaction effects on the cutting process. The kerf width is measured as an indicator of precision using an optical microscope to evaluate the quality of the laser cuts. To address the non-linear relationships between these process parameters and kerf width, several machine learning models were utilized. Performance comparisons indicated that the Artificial Neural Network (ANN) model provided the highest accuracy, with R² values of 0.98 for the validation dataset and 0.95 for the testing dataset. The optimized ANN model offers a robust tool for parameter optimization, facilitating the determination of optimal settings to achieve the desired kerf width while ensuring productivity.

In this study, the microstructure of surface layers with varying roughness (Ra parameters) was analyzed using scanning electron microscopy (SEM) to optimize tribological properties in engineering design. SEM revealed key microstructural features – sharp and mild protrusions, pitting, microcracks and contaminants – that were not available in traditional profilometry. Reducing the Ra value improved surface uniformity by reducing irregularities and defect lengths, which had a positive effect on tribological properties and surface durability. However, defects were still present even at Ra < 1.25 μm, indicating the "Law of Microstructural Roughness," which emphasizes the inevitability of surface irregularities despite minimizing roughness. The integration of SEM results with profilometric methods enabled comprehensive identification and assessment of defects, combining microstructure with tribological properties. Results suggest that controlled roughness is key in combining materials and optimizing functional surfaces, particularly in the aerospace, biomedical and automotive industries, where reliability under demanding operating conditions is a priority.

To enhance the accuracy of fault diagnosis (FD) in motor rotor systems, this study introduces a novel method that leverages feature extraction (FE) combined with a CNN-BiGRU-Attention deep learning model. Initially, the time-domain features of the vibration signals are extracted using Variational Mode Decomposition (VMD), which also effectively denoises the data. Subsequently, the frequency-domain features of the vibration signals are extracted via Fast Fourier Transform (FFT). The aggregated features are then fed into the CNN-BiGRU-Attention model to perform fault classification. In this model, the Convolutional Neural Network (CNN) module extracts local spatial features, the Bidirectional Gated Recurrent Unit (BiGRU) module models the temporal dependencies, and the Attention mechanism enhances the focus on critical fault information, thereby improving the model's classification performance. Experimental results demonstrate that the proposed FD method achieves an accuracy of 99.58%. Compared to other commonly used models, the performance metrics of our model show significant advantages and superior performance.