A5064050982- Advanced machining processes and optimization
- Industrial Vision Systems and Defect Detection
- Manufacturing Process and Optimization
- Machine Fault Diagnosis Techniques
- Gear and Bearing Dynamics Analysis
- Non-Destructive Testing Techniques
- Digital Transformation in Industry
- Engineering Diagnostics and Reliability
- Injection Molding Process and Properties
- Advanced Manufacturing and Logistics Optimization
- Rough Sets and Fuzzy Logic
- Fault Detection and Control Systems
- Anomaly Detection Techniques and Applications
- Structural Health Monitoring Techniques
- Oil and Gas Production Techniques
- Optimization and Packing Problems
- Vehicle Routing Optimization Methods
- Scheduling and Optimization Algorithms
- Advanced Neural Network Applications
- Integrated Circuits and Semiconductor Failure Analysis
- Vibration and Dynamic Analysis
- Image and Object Detection Techniques
- Control Systems in Engineering
- Advanced Measurement and Metrology Techniques
- Image Processing and 3D Reconstruction
Instrumentation Technology and Economy Institute
2022-2023
Chongqing University
2020-2022
State Key Laboratory of Mechanical Transmission
2021
Cytoskeleton (United States)
2020
Zhengzhou University of Light Industry
2016-2017
Aiming at the insufficient sample size and unbalanced category of metal workpiece surface defect dataset in industrial production. But traditional image augmentation methods based on generative adversarial network (GAN) do not effectively correlate dependency between local feature global feature, loss function is rational, which causes problem diversity poor quality. A relative mean (TARGAN), driven by a twin attention mechanism, proposed to generate high-quality images. This method employs...
Abstract The vibration signal of the hobbing machine is susceptible to changes in frequency domain distribution owing influence machine’s inherent and random pulses, which affects condition monitoring wear prediction machine. Variational mode decomposition (VMD) can compensate for mixing problem ensemble empirical method its equivalent filtering property. However, performance VMD depends heavily on two hyperparameters that need be set advance, i.e. number bandwidth-limited intrinsic...
Abstract Gear hobbing is currently one of the most widely used gear processing technologies. Still, due to its involvement in multiple parameters, coupling influencing factors, and limitations technical conditions, many problems need be improved process. This article uses TRIZ innovative method conduct a systematic functional analysis contradiction process, proposes series intelligent methods, verifies effectiveness proposed set methods through on-site application final ideal solution.
In the operation of bulldozer, working device bears various complex loads, which raise higher demands on rationality structural design, convenience operation, durability and reliability device. The finite element method is used to analysis strength fatigue life dangerous points, stress, strain distribution allowed cycle number were analyzed in two unfavorable conditions calculation conditions. performance was improved by changing critical dimension. Then blade analyzed. Finally, simulation...
In order to allocate cloud manufacturing resource fast and effectively, a novel method is proposed according the characteristics of allocation existing problems. The evaluation index system by deeply analyzing operation mode process allocation. Moreover, optimization model established in environment, which solved Maximum Inherit Optimization algorithm. performance demonstrated using simulation experiments.
For addressing the problem that quality indicators of gear hobbing are complicated and influencing factors unknown, a characteristic processing method combining improved multi-objective differential evolution (IMODE) clustering based on peak density (DPCA) is proposed. This can extract parameters strongly influence for multi-process multi-quality indicators, quantify their importance to comprehensive indicators. First, correlation analysis inspection by DPCA, set relatively independent...
Abstract The formulation of gear hobbing processing parameters relies on extensive professional knowledge and artificial experience, inappropriate will lead to low machining precision high manufacturing costs. In order obtain suitable in speed dry process, Gradient boosting regression (GBR) Generalized Regression Neural Network (GRNN) are adopted establish a process parameter optimization model targeting accuracy energy consumption. And then method based Differential Evolution (DE)...
It is a difficult problem to predict the machining error of helical gears, among which high-dimensional processing parameter variables are an obstacle. To solve this problem, gear quality prediction method based on parameters importance analysis (MQP-PPIA) proposed. First, Kernel Principle Component Analysis (KPCA) algorithm used preliminary reduce dimension inspection data gear. Then, improved Birch clustering with initial auto-generator The one-dimensional grade label obtained by as...
Regarding the surface structure of shaft gear parts is complex, difficulties detection are that complex background and reflection might interfere detection. Besides, different types defects in areas cannot be detected independently. Therefore, it necessary to extract each machining region. Traditional region extraction methods effectively separate object area from due unclear boundaries. A method for was offered this study, which based on Mask R-CNN. First all, ResNet50 used as feature...
Aiming at the problem that quality indicators of hobbing are complicated and influencing factors unknown, a method for analyzing importance process parameters based on density peak clustering improved multi-object differential evolution algorithm is proposed. This can obtain low-dimensional inspection indicator essence dimension in case multi-process multi-quality indicator, get characteristic affecting gear hobbing, quantify different comprehensive indicators.
As one of the most widely used equipment, rotating machinery fault diagnosis is significant. The data-driven methods aim at mining relationship between monitoring data and realizing automatic accurate equipment results. However, current generally depend on prior knowledge experienced diagnosticians, unsuitable for a variety conditions. deficiencies limit development intelligent technology machinery. In this research, an adaptive method based firefly algorithm deep sparse autoencoder...