Journal of The Brazilian Society of Mechanical Sciences and Engineering最新文献

Researchers can now utilize new materials to create innovative models for lower limb prostheses and explore novel ways to use them for efficient dynamic control. To achieve user-friendliness, one area of research focuses on recovering and reusing kinetic walking energy for dynamic control. This paper proposes a new design for a magnetorheological (MR) valve, along with a rotary actuator which offers a dynamic control for a lower limb prosthesis. The design will allow the storage of the energy during heel and mid-foot contact phases and to utilize it during toe support to lift the foot off the ground and establish a balance for the lower limb prosthesis. The energy is transferred through a magnetorheological hydraulic circuit and stored using a pneumatic system. The speed of energy transfer is regulated by magnetorheological valves. A series of MR valve designs were proposed and evaluated experimentally, which allowed the identification of the most suitable variant in the targeted application context. The design of the lower limb prosthesis was simulated using SolidWorks, and its dynamic behavior was analyzed in ANSYS.

In order to alleviate the contradiction between the increasing demand of seafarers for moxibustion physiotherapy and the shortage of moxibustion doctors, a medical double-arm moxibustion robot was designed by using a six-degree-of-freedom mechanical arm and a four-degree-of-freedom mechanical arm to simulate traditional Chinese medicine moxibustion techniques. The robot coordinate system was established by D-H parameter method, and the forward and inverse kinematics of the robot model were calculated. The robot model was established and simulated by Robotics Toolbox in MATLAB. The angular velocity and angular acceleration curves of each joint and the trajectory and displacement of the robot end were obtained, and the feasibility of robot trajectory planning was verified. Through the preliminary design, the collaborative process of task assignment for double moxibustion robot was established. The simulation test bench was built to further simulate the temperature of human epidermis, and the relationship between the end distance of moxibustion robot and the heating of human epidermis was determined. The simulation and experimental results show that: a) The robot does not appear serious impact or stutter phenomenon in the simulation process, and the kinematics performance is good, which verifies the feasibility of the robot model; and b) during the simulation test, the heating temperature of human epidermis can be maintained at 43 °C, which realizes the expected moxibustion temperature of patients and verifies the effectiveness of the robot model.

The purpose of this study is to design and construct a novel interactive game. This game is a robotic learning and imitation task. It is based on visual interaction of the player. The cornerstone technique used in this game is natural learner unit pattern generator neural networks (NLUPGNN), which is able to generate required motion trajectories based on imitation learning. The systematic design of these neural networks is the main problem solved in this paper. The unit pattern generators can be divided into two subsystems, a rhythmic system and a discrete system. A special learning algorithm is designed to use these unit pattern generators. The unit pattern generators are connected and coupled to each other to form a network, and their unknown parameters are found by a natural policy gradient learning algorithm. The motion sequences train some nonlinear oscillators, then they reproduce motions for a humanoid robot. As a result, the joints of the humanoid body imitate the movements of the teacher in real time. The main contribution of this work is the development of this learning algorithm, which is able to search the weights and topology of the network simultaneously. The algorithm synchronizes the learning steps by coupling the neurons in the last step.

The present paper proposes a new formulation for the LDM framework based on the energy equivalence hypothesis for unreinforced concrete cracking. The proposed model was applied to a beam under a four-point bending test and a real-size waterway tunnel lining, where the numerical results were compared to experimental observations. Regarding the analysed examples, the proposed model presented accurate results. It was possible to observe that the LDM using the energy equivalence hypothesis applied to unreinforced concrete structures leads to results in agreement with experimental references. Therefore, it is possible to assume that the energy equivalence hypothesis can be extended to analyse several structures of different materials under different load conditions under the LDM framework.

To address the issues of poor noise resistance and insufficient generalization performance in traditional fault diagnosis methods, an end-to-end rolling bearing fault diagnosis method based on bidirectional interactive convolutional neural network (BICNN) is proposed. Firstly, the bearing vibration signal is directly input into the wide convolutional kernel for rapid feature extraction, reducing the interference of high-frequency noise. Secondly, a modified rectified linear unit (M-ReLU) activation function is designed to solve the problem of “neuron death” in the ReLU activation function. Then, a bidirectional interactive feature extraction module is constructed, and the features extracted are input into the bidirectional interactive feature extraction module to capture the channel and spatial feature information simultaneously. Next, the extracted information is imported the presented feature enhancement module to achieve more valuable information transmission and accumulation. Finally, a small convolutional kernel is applied to further extract feature information, and a global average pooling layer is used to replace the fully connected layer, reducing the number of parameters while avoiding the problem of model overfitting. The softmax is utilized to classify the types of bearing faults. Two different data sets are adopted to validate the fault diagnosis performance of BICNN model under − 4 dB and variable operating conditions. Experimental results show that the average recognition accuracy reaches 97.71% under variable load and 98.25% under variable speeds, which is higher than other comparison methods. It is verified that the proposed method has higher diagnostic accuracy and generalization ability in complex working conditions.

This paper addresses the challenges of control and human–robot interaction for lower-limb gait rehabilitation exoskeletons. To enhance the robustness of the control against unknown but bounded uncertainties, we propose a model-free adaptive sliding mode control strategy enhanced by a variable impedance approach. The adaptation law prevents the overestimation of control gain in the presence of uncertainty and ensures the sliding condition to mitigate the effects of unknown uncertainties. The variable impedance approach also allows the impedance of the entire system to adapt dynamically over the gait cycle and maintain the accuracy of the robot in tracking desired joint trajectories. We provide a detailed stability proof using Lyapunov theory and demonstrate the finite-time convergence of the defined sliding surface. The proposed strategy does not require knowledge of model parameters, resulting in reduced computational complexity. A lower extremity rehabilitation exoskeleton model was utilized as an illustrative example. To demonstrate the effectiveness of the proposed approach, we conducted several simulations using a lower-limb rehabilitation exoskeleton model. Comparative evaluations were performed against conventional control methods such as the conventional sliding mode and computed torque controllers. The results indicate the effective performance of the proposed controller in the presence of impedance, in reducing the detrimental effects of interaction forces and model uncertainty, as well as accurately tracking the desired gait trajectories.

The novel compliant Y-shaped mechanisms have mechanical advantages for the multi-axis force/moment sensor since they yield more elastic deformation and more voltage output for strain gauge measurements compared to their common alternative cross-beam structures. This study includes stress topology and natural frequency approximations of the y-shaped structure by taking into account elastic beams and the stiffness of the beam connections. Subsequently, the stiffness and strain output accuracy of the model greatly improved compared to previous studies. The accuracy of the theoretical model is examined in a sensor dimension range by statistical error and parameter analysis. Comparing with the finite element model, results show that the percent error of the theoretical model is 2% for strain, 5% for stiffness, 3% for first natural frequency, and 8% for equivalent stresses, with all correlations above 98%. Comparing the optimization results with the literature, although the y-shaped structure shows similar mechanical behavior with the cross-beam structure, the voltage output of the y-shaped structure is increased up to (100%) for force/moment axes. Moreover, the optimization process with MATLAB GlobalSearch lasts approximately 1 s for the theoretical model and 8 h for the finite element model.

In this paper, the nonlinear buckling analysis of two-dimensional functionally graded nanobeams is investigated using ensemble machine learning (ML) techniques and semi-analytical approach based on Fourier series and Stokes’ transformation. Ensemble models such as XG boosting, gradient boosting, light gradient boosting, adaptive boosting, random forest, and extra trees regressor are utilized to explore the complex relationship between different input features and the buckling loads of the nanobeams. The training data for these models are derived from the nonlinear strain gradient theory. Performance of ML models are evaluated using multiple metrics such as R², MAE, MAPE, MSE and RMSE and visual representation techniques like Taylor plots, scatter plots, and box plots. Model interpretation using SHAP analysis is also employed for studying the impact and significance of each input feature on buckling loads. Among all the established models, light gradient boosting demonstrated superior performance in predicting the buckling loads accurately. It is shown that the ensemble ML models can accurately estimate the buckling loads of a two-dimensional functionally graded nanobeam with R² value of 0.999 given the adequate amount of training data.

Time-dependent reliability-oriented sensitivity analysis (ROSA) is a potent technique utilized to assess the impact of uncertain random input variables on the time-dependent reliability of planar linkage mechanisms. However, in cases where these random variables exhibit significant correlation behavior, the existing time-dependent ROSA indices will mislead the decision-makers in distinguishing between distinct types of uncertainty effects. To tackle this problem, this study commences by examining the connotations of main and total effect indices for correlated variables. Three new time-dependent ROSA indices are proposed in this work, specifically, the individual uncorrelated effect index, the total correlated effect index, and the individual interaction effect index, which enable a comprehensive exploration of the diverse effects of correlated random variables on the time-dependent reliability of planar linkage mechanisms. Additionally, efficient computational procedure is developed by combining the envelope function method with Monte Carlo simulation to estimate the proposed time-dependent ROSA indices. Finally, the proposed approach is applied to a four-bar mechanism and a rack-and-pinion steering linkage for illustrating the necessity and superiority of exploring the correlation effects in identifying failure sources with ROSA approaches.

Designing safe and tailored strategies for robotic therapy requires the knowledge of patient joint torques, allowing control strategies to adjust the torque level provided by the robotic device according to the patient’s performance. Given the impracticability of measuring human joint torques directly, many works in the area have used estimation techniques that require complex calibration and signal processing or introduce uncertainty in their system modeling. This paper evaluates three disturbance observer techniques for estimating ankle joint torque as an alternative solution. Based on the generalized momentum and Kalman filter methodologies, the approaches were implemented on a robotic device for ankle rehabilitation. They were evaluated on six healthy voluntary users for sitting position movements. The techniques demonstrated effectiveness in estimating human joint torque across three distinct human–robot interaction modes, with performance evaluation through normalized root-mean-square error (NRMSE) metrics. Statistical analysis, including ANOVA, Kruskal–Wallis, and Dunn’s post hoc tests, was employed to assess approach performance and impact effects under different configuration settings. These analyses highlighted significant differences in performance among the techniques, enhancing the understanding of the estimation approaches and highlighting their potential integration into robotic rehabilitation settings.