A5101934377- Adaptive Control of Nonlinear Systems
- Control and Dynamics of Mobile Robots
- Energy Load and Power Forecasting
- Hydraulic and Pneumatic Systems
- Vehicle Dynamics and Control Systems
- Advanced Algorithms and Applications
- Power Systems and Renewable Energy
- Advanced Sensor and Control Systems
- Structural Health Monitoring Techniques
- Underwater Acoustics Research
- Smart Grid and Power Systems
- Underwater Vehicles and Communication Systems
- Inertial Sensor and Navigation
- Water Systems and Optimization
- Gas Dynamics and Kinetic Theory
- Computational Fluid Dynamics and Aerodynamics
- Non-Destructive Testing Techniques
- Neural Networks and Applications
- Robotic Path Planning Algorithms
- Geophysics and Sensor Technology
- Soil Moisture and Remote Sensing
- Robotic Locomotion and Control
- Distributed Control Multi-Agent Systems
- Low-power high-performance VLSI design
- Advanced Frequency and Time Standards
Shaoxing University
2020-2025
Hebei Normal University
2023-2024
Chongqing University
2024
Xi'an Jiaotong University
2023
Inner Mongolia Yili Industrial Group (China)
2023
East China Normal University
2023
North China Electric Power University
2018-2019
Beihang University
2015-2019
Harbin Engineering University
2007-2018
Ocean University of China
2018
Considering the strong fluctuation and nonlinearity of wind speed, atmospheric uncertainties, speed prediction based on hybrid model is presented, which composed neural network, variational mode decomposition (VMD), Lorenz disturbance. First, VMD used to process data get several intrinsic functions (IMFs). Second, network (NN model) can be established by these IMFs training set, validation set adjust parameters. Subsequently, given disturbance added determine finial model, best parameter...
Abstract Oxygen content is one of the most critical factors for high-efficiency combustion. Online measurement oxygen from flame images important but still challenging. For construction an prediction model, current feature extraction methods are not straightforward. Additionally, there always sufficient data common operating conditions in practice, while only limited other conditions. The collection process model training costly and time-consuming. To tackle problem, this work presents...
Abstract This work described in this paper aims to enhance the level of automation industrial pearl classification through deep learning methods. To better extract features different classes and improve accuracy, balanced training datasets are usually needed for machine However, obtained practice often imbalanced; particular, acquisition cost some is high. An enhanced generative adversarial network, named multiview Wasserstein network (MVWGAN), proposed imbalanced problem. For minority...
Object segmentation in 3D scenes remains an important research topic computer vision. With the emergence of Gaussian Splatting (3DGS) scene reconstruction, object methods based on 3DGS have gained increasing attention. This paper proposes a novel paradigm that utilizes initial COLMAP point cloud from reconstruction process to achieve segmentation. Specifically, we first embed semantic features into cloud, train these and classifier under guidance 2D masks, segment target object's...
3D Gaussian Splatting (3DGS), as an emerging scene representation method, has gained significant attention for its high-quality rendering effects and real-time speed. Object segmentation in 3DGS scenes plays a crucial role understanding editing. For single-object where objects are placed on plane captured 360 degrees, we propose novel object approach that directly leverages geometric features space through constraints. Our key insight is to utilize COLMAP sparse point clouds instead of...
For an electro-hydraulic servo shaking table, there are usually various nonlinearities, which cause harmonic distortion of the sinusoidal tests, lowering control performance. Based on a Kalman filter, acceleration identification scheme is developed to provide information. A linear state space model response first built. The filter employed estimate model. harmonics, including fundamental response, can be directly estimated online, and amplitude phase each also computed. As it eliminates need...
A high-performance tire X-ray defect image classification method plays a key role in enhancing the automation level of detection. In industrial practice, however, typical challenge is that collected datasets diverse defects are often imbalanced. To address this issue, Wasserstein generative adversarial network (WGAN)–assisted proposed for imbalanced expand minority classes original datasets, WGAN model established to generate high-quality images. Considering feature similarity different...
This paper focuses on an electro-hydraulic servo system, which is derived from a shaking table. It proposes control scheme based back propagation (BP) neural network, whose weights are trained by the particle swarm optimization (PSO) according to fitness, determined input and feedback signals. Each of PSO includes thresholds BP. The movement each adjusted its local best-known position global in searching space. With update, satisfactory solution can be achieved. In order show performance...
Considering the difficult maneuverability of hovercraft and driving burden pilot, an automatic tracking system including intuitive human-computer interface adaptive chattering-free full-order terminal sliding mode (ACFTSM) controller is designed for humans (non-engineers in particular). The easy to be understood used. For design controller, four degrees freedom (DOF) model established. And based on DOF model, desired velocities are proposed by a TSM manifold achieve finite-time trajectory...
Electro-hydraulic servo shaking table usually requires good control performance for acceleration replication. The poles of the electro-hydraulic are placed by three-variable method using pole placement theory. system frequency band is thus extended and stability also enhanced. phase delay amplitude attenuation phenomenon occurs in corresponding to an sinusoidal input. elimination based on LMS adaptive filtering algorithm proposed here. task accomplished adjusting weights when there exits...
The stability of a fiber optic gyroscope (FOG) in measurement while drilling (MWD) could vary with time because changing temperature, high vibration, and sudden power failure. dynamic Allan variance (DAVAR) is sliding version the variance. It practical tool that represent non-stationary behavior signal. Since normal DAVAR takes too long to deal series, fast algorithm has been developed accelerate computation speed. However, both become invalid for discontinuous series. What worse, FOG-based...
Aims: Bryophytes are commonly overlooked in field observations and collections due to their small size, making it difficult obtain an objective exhaustive checklist of bryophytes a region.As result, there great limitations the general study conservation bryophyte diversity.Taking Tianmushan National Nature Reserve (TNNR) Zhejiang as example, this paper aims investigate diversity by sampling strategy combining two methods, namely floristic habitat (FHS) method plot (PS) method.Methods: Using...
The development of wind generation has rapidly progressed over the last decade, but it must be integrated into power grids and electric utility systems. However, cannot dispatched like conventional generators because generated by changes continuous fluctuation speed direction. So is very important to predict generation. This paper discusses why can predicted in short-term, how setup construction an ANN (artificial neural network) prediction model based on chaotic time series. analysis...
Non-linearities commonly exist in an electro-hydraulic servo shaking table, causing acceleration harmonics distortion when the table is excited by a sinusoidal signal, because its response includes higher harmonics, which lower control performance for table. To cancel system response, thus to improve performance, we need know about information. An identification algorithm developed here based on Kalman filter dynamically tracking A linear state space modelled. The applied as observation...
As a kind of renewable energy, wind energy is getting more and attention with its advantage rich, clean environmentally sustainable. Stochastic volatility the inherent property essential factor hindering development power prediction research. In this paper, by considering actual movement rule described Lorenz system, system wavelet decomposition were combined to improve speed models BP, RBF, Elman neural network for first time, model based on was established. The has been compared three...
Shaking tables play a vital role in mechanical environmental simulation. Sinusoidal shaking tests are usually applied to specimens for simulating periodic motions. Due the nonlinearities electro-hydraulic servo table, its sinusoidal acceleration response contains higher harmonics, which lower system control performance. To cancel those harmonic information should be firstly known. The paper proposes an identification scheme by using extended Kalman filter. A nonlinear state space model of is...
SUMMARY To obtain better performance on unstructured environments, such as in agriculture, forestry, and high-altitude operations, more researchers engineers incline to study classes of biologically inspired robots. Since the natural inchworm can move well various types terrain, inchworm-like robots exhibit excellent mobility. This paper describes a novel inchworm-type robot with simple structure developed for application climbing trees or poles certain range diameters. Modularization is...