Maintenance, Reliability and Condition Monitoring
Not a member yet
1200 research outputs found
Sort by
A multi-scale convolutional Siamese network for few-shot fault diagnosis of unmanned aerial vehicle rotor
Get PDFUnmanned Aerial Vehicles (UAVs) often face infrequent fault occurrences and high manual annotation costs, resulting in a critical shortage of valid fault samples for diagnostic research. Traditional fault diagnosis methods struggle with small sample sizes. This paper proposes a novel deep metric learning method, the Multi-Scale Convolutional Siamese Network (MSCSN), to address the few-shot learning problem in UAV rotor fault diagnosis. First, discrete wavelet transform (DWT) is used to compress and normalize the vibration signals, enhancing the prominence of signal features. Then, based on the multi-scale convolutional neural network (MS-CNN) model, the network automatically extracts multi-level features from rotor fault vibration signals, improving its adaptability to complex data. Finally, the Siamese network structure, with shared parameters and identical architecture, processes sample pairs and incorporates a small number of support samples for few-shot learning. Experimental results show that the proposed model achieves a highest accuracy of 94.42 % in few-shot tasks. In cross-domain transfer learning tests, the model achieves an average accuracy of 90.28 %, demonstrating its superior generalization ability and robustness across different environments. We also validated the model's stability using the publicly available MVS-UAV-BF dataset
Active fuzzy control of a suspension vehicle on wet and dry roads
Get PDFThis paper presents a co-simulation of MATLAB and CarSim to control and model a vehicle suspension system under different road surface conditions, either wet or dry, using an active fuzzy controller in MATLAB. CarSim is a professional vehicle simulation software capable of modeling nonlinear car dynamics with various uncertainties. These uncertainties are addressed by the fuzzy set approach due to its qualitative and robust control capabilities, effectively handling noise, disturbances (such as road conditions), and unknown parameters in CarSim’s vehicle model. The design of an active steering controller and rotational torque system using a fuzzy controller is crucial for enhancing road safety, especially given the increasing number of vehicle crashes. The research methodology varies based on the study's purpose, nature, and implementation capabilities. Accordingly, this research focuses on designing an integrated controller for an active four-wheel-drive system and direct rotary torque control using a fuzzy control method in the MATLAB Simulink environment. This study is analytical and functional, utilizing CarSim for simulation. A fuzzy logic-based integrated control system was designed for steady-state control to improve vehicle stability and steering. The controller adjusts the steering angle and torque to regulate the vehicle’s angular velocity and slip angle under various conditions. As tire performance changes during different maneuvers, the controller dynamically adapts its output to maintain optimal operation within the effective performance range. The significance of using fuzzy logic lies in its ability to handle non-linearity without requiring approximation, ensuring high accuracy. Additionally, it delivers excellent results in enhancing vehicle stability. The findings indicate that the controller significantly improves the vehicle’s dynamic behavior across different driving maneuvers compared to an uncontrolled vehicle
PSO-PPO-based reinforcement learning control strategy for active suspension systems under multiple operating conditions
Get PDFTo address the poor generalization capability and extended training duration of reinforcement learning (RL)-based active suspension control systems, this study proposes a PSO-PPO algorithm for multiple operating condition suspension control. The methodology initiates with establishing a 4-DOF suspension dynamic model under three characteristic driving conditions: constant-speed operation, vehicle launch, and emergency braking, which is subsequently converted into state-space representation. The novel PSO-PPO framework synergizes particle swarm optimization with proximal policy optimization to train condition-specific agents. Based on the trained optimal agents, the entropy weight method is applied to adjust the reward function weight coefficients to develop a generalized multi-condition controller. Finally, the control effectiveness of the PSO-PPO algorithm is validated through constant-speed, launch, emergency braking, and multi-condition concatenated scenarios. Simulation results demonstrate that the PSO-PPO algorithm achieves shorter training times while maintaining balanced performance in ride comfort, handling stability, and safety across all conditions
A vision-based deep learning approach for non-contact vibration measurement using (2+1)D CNN and optical flow
Get PDFThis paper introduces a proof-of-concept vision-based deep learning approach for vibration measurement, proposing a factorized (2+1)D Convolutional Neural Network (CNN) model to predict four vibration metrics: acceleration, velocity, displacement, and frequency, with a focus on rigid body motion. Unlike conventional neural network models that primarily focus on frequency prediction alone, this approach uniquely enables the simultaneous estimation of four critical vibration metrics, offering a comprehensive and cost-effective alternative to traditional contact-based sensors such as accelerometers. The framework relies on the visibility of a training fiducial marker, eliminates the need for calibration in controlled settings, enhancing scalability across specific environments. A curated dataset was generated using a controlled experimental setup comprising a single object in a lab-scale environment, augmented synthetically to enhance frequency diversity. An optical flow-based preprocessing algorithm synchronized motion features in recorded video inputs with measured vibration labels, improving measurement accuracy. The proposed model achieved an average Mean Absolute Percentage Error (MAPE) of 7.51 %, with acceleration predictions exhibiting the lowest error at 4.84 % and displacement the highest at 8.80 % across varying brightness levels and object-camera distances. Techniques such as Region of Interest (ROI) cropping and multi-section frame extraction were implemented to reduce computational complexity while further enhancing accuracy. These results highlight the framework’s potential for non-invasive vibration analysis, though its generalizability is limited by the single-object dataset. Future work will expand the dataset, integrate multi-sensor inputs, explore marker-less tracking methods, and enable real-time deployment for predictive maintenance and structural health monitoring
A k-kNN miscalibrated current transformer identification method based on line topology for distribution networks
Get PDFThe operational duration and environmental factors associated with current transformers (CTs) in distribution networks makes them prone to measurement miscalibration during their operation. To address this, a kernel k-nearest neighbor (k-kNN) miscalibrated CT identification method based on line topology is proposed. This method relies on the composite characteristics of load currents specific to certain line topologies. High-precision secondary-side CT current data provided by the current acquisition devices in the feeder area are used to construct a multiple linear regression model. The multiple linear regression model is established in the complex domain, and indirectly assesses the measurement status of the current transformers by analyzing the complex coefficients. Building upon the kNN identification algorithm, a kernel function is introduced to map low-dimensional distance feature vectors into a higher-dimensional feature space where linear separability is significantly enhanced, thus improving the accuracy with which abnormal coefficients can be detected in the multiple linear regression model. Experimental simulations and field application scenarios demonstrate that the proposed method significantly outperforms traditional kNN algorithms in terms of classification performance. Specifically, there is an increase of 12.0 % in the F1 score, a rise of 13.3 % in accuracy, and an improvement of 12.0 % in recall. Moreover, in practical engineering applications, the recognition metrics consistently exceed 93 %, which substantiates the effectiveness of the proposed miscalibrated CT identification method
Analysis of the structural performance of reinforced concrete under fire loading
Get PDFThis study examined the behavior of reinforced concrete structures when exposed to high temperatures resulting from fire. Deterioration in material strength due to fire exposure alters a reinforced concrete structure’s load-bearing capacity and overall behavior. Elevated temperatures negatively affect key material properties of reinforced concrete, including density, coefficient of thermal expansion, thermal conductivity, and elastic modulus. As a result, if a structure experiences fire either concurrently with or prior to an earthquake, these changes in material properties will significantly influence its dynamic performance. For the numerical simulation, the selected structure was designed with a formwork plan and load-bearing system in accordance with earthquake-resistant design principles. Based on this design, fixed and variable loads acting on the beams were assigned. By promoting resilient infrastructure capable of withstanding severe environmental conditions such as earthquakes and fires, this study contributes to the achievement of sustainable development goals. It underscores the necessity of integrating fire resistance into earthquake-resistant design to foster disaster-resilient urban development. The findings may encourage more flexible and sustainable construction practices aligned with SDGs 9 (Industry, Innovation and Infrastructure), 11 (Sustainable Cities and Communities), and 13 (Climate Action)
CFD analysis of model rocket using the VDI 2206 approach
Get PDFA model rocket system serves as an excellent example of a mechatronic system, integrating mechanical, electrical, and control components. Computational Fluid Dynamics (CFD) plays a critical role in mechatronic system design by enabling the analysis and optimization of fluid interactions within these integrated systems. In rocket design, the accurate assessment of aerodynamic forces – thrust, weight, drag, and lift – is essential for optimizing performance. CFD analysis is employed to determine the drag coefficient (Cd) and lift coefficient (Cl), both of which contribute to improving the rocket's aerodynamic efficiency. CFD is a powerful tool for evaluating key aerodynamic parameters such as velocity, pressure, and temperature while also identifying and mitigating design flaws to enhance overall performance. This study examines the model rocket system from a mechatronic system design perspective, evaluating three different mesh structures in two- and three-dimensional CFD simulations to determine the most suitable configuration. The accuracy of the mesh depends on factors such as element size, quality metrics (skewness, orthogonal quality), and first-layer thickness. A well-refined mesh that adheres to these criteria significantly enhances the reliability of the simulation results, ensuring more precise aerodynamic analysis and performance optimization. The analysis results obtained in this study indicate that the rocket’s nose cone and the area around the wings are subjected to the highest forces, and that mechanical and structural improvements are needed in these areas
Improved APF-based path planning for aircraft towbarless towing vehicle system
Get PDFTo enhance the maneuvering efficiency and safety of the aircraft towbarless towing vehicle (TTV) system, this study presents an optimized path planning method based on an improved artificial potential field (APF) algorithm. First, comprehensive kinematic and dynamic models are established, incorporating both lateral and yaw motions of the TTV system. Second, to mitigate obstacle interference challenges in complex airport environments, the proposed method introduces an innovative relative-distance safety factor and implements a dual-repulsive-force cooperative planning strategy, effectively overcoming the traditional APF algorithm’s limitations regarding goal unreachability and local minima. Furthermore, the integration of Bézier curves ensures curvature continuity in the planned path, thereby maintaining compliance with kinematic constraints. Finally, a constrained-motion TTV simulation model is developed to validate the algorithm’s performance. Simulation results demonstrate that, in static obstacle scenarios, the proposed method successfully enables autonomous path planning, generating smooth and collision-free trajectories. This approach offers a robust solution for ensuring stable and reliable operation of the TTV system in real-world airport environments
Simulation analysis and safety performance assessment of a novel am opening barrier for highway
Get PDFTo address the need for quick opening, easy mobility, and convenient maintenance of barrier in highway central medians, a novel Am rotatable open barrier has been designed. Based on guardrail safety performance evaluation standards, a finite element model of the vehicle-guardrail interaction is established for collision simulations to validate the adequacy of the new guardrail structure. In the meantime, full-scale vehicle crash tests are conducted to assess the safety performance of the proposed open guardrail. The results demonstrate that safety performance metrics, including vehicle post-collision acceleration, maximum dynamic inclination, maximum lateral dynamic deformation, and displacement extension values, meet standard requirements in both simulations and real-world validations. Additionally, the vehicle doesn’t penetrate, overturn, or ride over the barrier, and no rollover occurred. This indicates that the newly designed barrier not only fulfills the functions of quick opening and easy mobility but also provides excellent blocking and guiding capabilities, contributing to the enhanced safety and operational efficiency of highway service
Multi source heterogeneous data diagnosis method of rotating machinery based on parameter collaborative optimization of multi-scale convolutional autoencoder
Get PDFIn order to fully utilize the features of multi-source heterogeneous data and effectively improve the accuracy and efficiency of fault diagnosis of rotating machinery, a multi-source heterogeneous data diagnosis method based on parameter collaborative optimization multi-scale convolutional autoencoder (MSCAE) is proposed. Firstly, multi-scale information learning is integrated into the convolutional autoencoder (CAE) to consider the temporal and spatial feature information of the diagnostic object simultaneously. To improve the training and diagnostic efficiency of MSCAE, a quantum particle swarm optimization (QPSO) module is used to perform hyperparameter optimization on it using chaos initialization and dynamic weight strategy (DWS). Besides, the sparse attention mechanism is introduced into the MSCAE model to improve the recognition rate of key fault features hidden in the original heterogeneous signals. Finally, the confusion matrix and visualization techniques are used to achieve fault classification. The experimental results demonstrate that after 100 experiments, the proposed method has an average diagnostic accuracy of 98.5 % and strong robustness to noise, providing a new method for rotating machinery fault diagnosis based on multi-source heterogeneous data