Journal of Mechanical Science and Technology最新文献

Aiming at the prediction model of bearing performance degradation based on recurrent neural network (RNN) and its variants ignores the feature spatial correlation, and cannot effectively handle long time series data, this paper proposes a rolling bearing performance degradation prediction model based on temporal graph convolutional neural network (T-GCN). For non-stationary and nonlinear characteristics of vibration signals, this paper introduces a rolling bearing feature evaluation method based on multiscale dispersion entropy (MDE) to better characterize time series. To effectively solve the spatial correlation problem between samples and features, this paper uses the topological structure of a path graph to build a graph model and combines gated recurrent unit (GRU) and graph convolutional neural network (GCN) to build a T-GCN prediction model. Finally, this article established a rolling bearing fault prediction experimental platform and validated it using the University of Cincinnati public dataset. The experiment shows that compared with GRU, GCN, and LSTM models, the RMSE and the MAE evaluation indicators based on the T-GCN model have decreased by 6 % to 28 % and 11 % to 28 %, respectively, which suggests that the T-GCN model has a higher prediction accuracy and a better model fitting goodness.

In this study, the axial compressive stresses of hollow circular composite tubes were investigated. For this purpose, hollow circular composite tubes with various inner diameters (Ø12 and Ø13 millimeters), a height of 80 millimeters, and an outer diameter kept constant at Ø17 millimeters were fabricated using a fiber winding process. In the production of hollow circular tubes, epoxy was used as resin, and glass fiber, carbon fiber, and Kevlar fiber were used as reinforcement materials. Experimental investigations were carried out for three different reinforcement materials, two thin-wall thicknesses, and five orientation angles. Axial compression tests were performed to research the influences of reinforcement materials, thin-wall thickness, and orientation angles on the compressive stresses. The axial compressive strength of the samples was observed experimentally by applying the load in the vertical direction. The reinforcement material, orientation angle, and thin-walled thickness had an important influence on the axial compressive stress. The glass/epoxy reinforcement material was found to have the highest axial compressive strength at 204 Mpa. When the orientation angle increased from 45° to 88°, the axial compressive stress increased by 2.27 times in glass/epoxy, 2.36 times in carbon/epoxy, and 2.37 times in Kevlar/epoxy specimens, respectively. In addition, by increasing the specimen wall thickness by 0.5 millimeters, the axial compressive stress at an 88° orientation angle increased by 9.67 % glass/epoxy, 11.85 % carbon/epoxy, and 7.14 % Kevlar/epoxy specimens.

Particle distribution, wear failure, and particle passage performance in a two-stage mixed-transported pump are investigated by coupling computational fluid dynamic (CFD) and discrete element method (DEM). Results exhibit good agreement with the experimental data. The velocity and trajectory of a 10 mm particle are examined under three different concentrations. Consequently, the relationship between particle motion and the wear on the impeller and volute is revealed. The results indicate that low-speed particles lead to particle packing at the inlet of the first stage. The force variation of particles is closely related to the number of collisions with the wall. The wear on the suction side of the second-stage impeller and the second-volute wall are significantly less than that on the first-stage. Finally, the particle passage coefficient P is defined, demonstrating that the particle passage performance of the second-stage pump is better than that of the first-stage.

A prediction method based on the stepped-lap repair strength was proposed to study the parameterization of repairing the damaged wind turbine blade spar cap structure. First, the numerical analysis model of spar cap damage was established based on the continuous damage mechanics model, the puck criterion, and the cohesive zone model. Then, the influence of the repair parameters on the repair strength was studied based on the stepped-lap repair theory. Finally, the optimal repair scheme was determined by a strength recovery rate analysis. Results demonstrate that increasing lap length and the number of overply layers can enhance the load-carrying capacity of the stepped-lap repair structure within a certain range. Lap length and the number of overply layers substantially affect the bearing capacity of the spar cap. The influence of the overply layer is slightly higher than that of the lap length, while the lap width effect remains relatively minor.

High-temperature wear tests on molybdenum–tungsten–vanadium steel were conducted using a universal mechanical high-temperature friction and wear testing machine. The experimental results reveal the evolutionary mechanism of high-temperature friction in the oxide layer of the steel. The oxidation wear of the test steel occurred at 700 °C with the evolution of characteristic monolayer tribo-oxidation layers on the frictional surface. The frictional oxide underwent oxide initiation - lateral growth of massive oxide - formation of frictional oxide layer - local intrusion of oxide layer into substrate - local penetration of oxide lateral growth contact. The frictional oxide evolutionary process was accompanied by a shedding of the frictional oxide surface. Carbides in the test steel prevented the tribo-oxidation layer from invading the matrix. The unique evolutionary characteristics of the tribo-oxidation layer are essential factors affecting the mechanism of maintaining slight oxidative wear.

A human posture interaction system guaranteed by algorithmic control is proposed to improve the convenience and control accuracy of wheelchairs. Initially, a system is designed based on microelectromechanical systems (MEMS) technology that incorporates a dual-inertial measurement unit to efficiently capture data on the user’s body posture. Subsequently, a K-means clustering algorithm is implemented in the system to analyze and recognize the user’s body position in real-time. This condition allows the system to deduce the user’s intention and react accordingly. Finally, the control system is validated through identification control experiments. Experiment results demonstrate a success rate above 97 %, thereby suggesting the reliability of the posture recognition algorithm.

The remaining useful life prediction of Industrial robot RV reducer is challenging due to the strong redundancy, unstable degradation initiation point, and environmental interference that may obscure the key state information during long-term operation. To address this problem, this paper proposes a novel remaining useful life prediction method for robot RV reducer with multi-depth network and multi-kernel support vector data description. Firstly, the degradation features are constructed by the hidden layer node of deep belief network to reduce the interference and redundancy. Secondly, the first predicting time node is determined by multi-kernel support vector data description to locate the stable degradation initiation node. Then, the temporal convolutional network is applied to predict the remaining useful life in the degradation stage, which improves the prediction accuracy. Finally, the effectiveness of the proposed method is verified by the accelerated fatigue experiment of a self-made robot.

As environmental regulations on exhaust gas are enforced worldwide and marine transportation is progressively strengthened, various methods are being explored to comply with emission regulations by using alternative energy sources. This study involved the optimization of a thermoelectric power generation (TEG) system, which converts thermal energy into electrical energy. The design of the TEG system was optimized by using CFD to analyze the heat transfer phenomena resulting from the relative positions of the TEG modules, and the results of this analysis were validated with 1/100 experimental scale measurement data. The comparison of the CFD results with the experimental measurements revealed that the maximum discrepancy exists for the temperature near the cooling jacket and for the amount of power generated by the system. The significant difference between the CFD and experimental results is attributable to the adiabatic conditions assumed for CFD since the experimental facility experiences heat loss to the atmosphere. In this study, we established an analysis methodology by conducting a comparative validation between CFD calculations and experimental data. The methodology is expected to play a significant role in optimizing the design of thermoelectric power generation systems.

In the present study, the performance of a solar dryer supported by an indirect type wicked heat pipe was assessed. To transfer the heat from the solar collector to the drying chamber, a passive high thermal conductivity heat pipe was developed. At 1:30 pm, there was a 21 °C temperature difference between the drying chamber and the ambient air. The results indicated that after 12 hours of drying, the banana slices’ moisture content dropped from 150 % d.b. to 11 % d.b. It is found that the average drying rates in Trays 1, 2, and 3 are the following: 0.1452, 0.1492, and 0.1613 kg water/kg dry matter/h. The effective moisture diffusivity ranged from 1×10⁻⁷ to 9.405×10⁻⁸ m²/s, whereas open sun drying had the lowest average effective moisture diffusivity of 3.87×10⁻⁰⁷ m²/s. The estimated dryer efficiency was 26 %. Also the payback period of developed dryer was found to be 0.57 years. The developed dryer can be installed in any rural villages to dry any agricultural products.

To address the ill-conditioning of the Jacobian matrix in the geometric error calibration of parallel mechanisms, a Newton method with characteristic value correction (NMCVC) is proposed. This method integrates and enhances the principles of the characteristic value correction iteration method (CVCIM), and Newton method, offering targeted improvements for more effective calibration. First, the damping coefficient is introduced into the CVCIM, and an adaptive strategy for determining the damping coefficient is developed with rigorous proof steps according to the relationship between the condition number and the singular value, which enhances computing efficiency while avoiding the ill-conditioning of the Jacobian matrix. Second, a dynamic adjustment strategy for the CVCIM’s termination condition is designed. This strategy initially estimates the descending direction roughly to approximate the actual descending direction, enhancing computing speed, and then estimates it more accurately at the end of the training stage to obtain precise geometric error values. Finally, by taking a 3RPS parallel mechanism as the instance, three sets of simulation experiments have been designed to test and verify the effectiveness of the algorithm.