A5016600498- Adaptive Control of Nonlinear Systems
- Underwater Vehicles and Communication Systems
- Soft Robotics and Applications
- Control and Dynamics of Mobile Robots
- Domain Adaptation and Few-Shot Learning
- Micro and Nano Robotics
- Structural Health Monitoring Techniques
- Distributed Control Multi-Agent Systems
- Brain Tumor Detection and Classification
- Neural Networks Stability and Synchronization
- Advanced Control Systems Optimization
- Adaptive Dynamic Programming Control
- Infrastructure Maintenance and Monitoring
- Iterative Learning Control Systems
- Mycobacterium research and diagnosis
- Advanced Neural Network Applications
- Modular Robots and Swarm Intelligence
- Advanced Sensor and Energy Harvesting Materials
- Concrete Corrosion and Durability
- Cancer-related molecular mechanisms research
- Fault Detection and Control Systems
Guangzhou University
2020-2024
The existing building structures are mainly composed of concrete and masonry structures. This paper proposes a Fractal Theory-based Method (FTM) designed to efficiently detect surface cracks in both structures, addressing limitations crack recognition techniques, such as weak adaptability complex surfaces requiring large amount computational human resource. A novel Soft Box-Counting (SBC) algorithm is proposed calculate texture roughness information from multi-channel multi-scale images,...
Abstract A capsule network (CapsNet) is a recently proposed neural model with new structure. The purpose of CapsNet to form activation capsules. In this paper, our team proposes dual attention mechanism (DA-CapsNet). DA-CapsNet, the first layer added after convolution and referred as Conv-Attention; second PrimaryCaps Caps-Attention. experimental results show that DA-CapsNet performs better than CapsNet. For MNIST, trained tested in testset, accuracy 100% 8 epochs, compared 25 epochs for...
Gradient-based flux observers offer promise in permanent magnet synchronous motor (PMSM) encoderless control due to their computational efficiency and adaptability but face challenges gain tuning noise sensitivity. This paper proposes a robust optimized observer combining the huber loss function (HLF) root mean square propagation (RMSProp) algorithm overcome these limitations. Firstly, nonlinear is developed with HLF enhance steady state performance of by leveraging its smoothness...
Soft robots have the characteristics of good adaptability to an environment. But at present, most soft only complete a single specific task, and as result, their ability adapt complex environments is limited. In order solve weak diverse environments, this paper presents new bionic omnidirectional bending actuator (BOBA) inspired by leeches, caterpillars, other mollusks. BOBA has less radial expansion, fast response, flexibility, bending. Using modular methods, it can be quickly assembled...
An unknown nonlinear disturbance seriously affects the trajectory tracking of autonomous underwater vehicles (AUVs). Thus, it is critical to eliminate influence such disturbances on AUVs. To address this problem, paper proposes a double-loop proportional–integral–differential (PID) neural network sliding mode control (DLNNSMC). First, PID surface proposed, which has faster convergence speed than other surfaces. Second, high-order observer and are combined observe compensate for AUV system....
An adaptive proportional integral robust (PIR) control method based on deep deterministic policy gradient (DDPGPIR) is proposed for n-link robotic manipulator systems with model uncertainty and time-varying external disturbances. In this paper, the of nonlinear dynamic model, disturbance, friction resistance are integrated into system, term used to compensate system. addition, information as input DDPG agent search optimal parameters controller in continuous action space. To ensure agent’s...
In this paper, a soft robot driven by gas–liquid phase transition actuator with new structure is designed; The the pressure generated electrically induced ethanol transition. drive was found to be able generate larger driving force using only low voltage. Compared gas of traditional robot, transition-driven does not require complex circuit system and huge external energy supply air pump, making its overall more compact. At same time, because on has good tightness less recovery time. A...
In automatic control systems, negative feedback has the advantage of maintaining a steady state, while positive can enhance some activities system. How to design controller with both modes is an interesting and challenging problem. Motivated by it, on basis idea catastrophe theories, taking as two different states system, adaptive alternating (APNF) model advantages proposed. By adaptively adjusting relevant parameters constructed symmetric function learning rule based error forward weight,...
Forming deep feature embeddings is an effective method for few-shot learning (FSL). However, in the case of insufficient samples, overcoming task complexity while improving accuracy still a major challenge. To address this problem, article considers consistency between similar data from fractal perspective, introduces priori knowledge, and proposes embedding model by combining FSL with dimension theory first time. We improve original algorithm used to describe image texture roughness suit...