Changyang Li
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Changyang Li has a total of 47 co-authors. Here are the top co-authors by number of shared publications:
Huapeng Wu
5 shared publicationsHarri Eskelinen
Laboratory of Production Engineering, LUT University, Lappeenranta, Finland
4 shared publications- Huapeng Wu
Lappeenranta University of Technology
4 shared publications - Kun Hu
Edith Cowan University, Joondalup, WA, Australia
4 shared publications - Zhiyong Wang
The University of Sydney, Sydney, Australia
4 shared publications - Ming Li
Mechanical Engineering, Doctoral School-Doctoral Programme in Energy Systems, Lappeenranta-Lahti University of Technology (LUT), 53850 Lappeenranta, Finland
3 shared publications
Latest Publications
- Open Access
Extended Short- and Long-Range Mesh Learning for Fast and Generalized Garment Simulation
3D garment simulation is a critical component for producing cloth-based graphics. Recent advancements in graph neural networks (GNNs) offer a promising approach for efficient garment simulation. However, GNNs require extensive message-passing to propagate information such as physical forces and maintain contact awareness across the entire garment mesh, which becomes computationally inefficient at higher resolutions. To address this, we devise a novel GNN-based mesh learning framework with two key components to extend the message-passing range with minimal overhead, namely the Laplacian-Smoothed Dual Message-Passing (LSDMP) and the Geodesic Self-Attention (GSA) modules. LSDMP enhances message-passing with a Laplacian features smoothing process, which efficiently propagates the impact of each vertex to nearby vertices. Concurrently, GSA introduces geodesic distance embeddings to represent the spatial relationship between vertices and utilises attention mechanisms to capture global mesh information. The two modules operate in parallel to ensure both short- and long-range mesh modelling. Extensive experiments demonstrate the state-of-the-art performance of our method, requiring fewer layers and lower inference latency.
Open AccessExtended Short- and Long-Range Mesh Learning for Fast and Generalized Garment Simulation
3D garment simulation is a critical component for producing cloth-based graphics. Recent advancements in graph neural networks (GNNs) offer a promising approach for efficient garment simulation. However, GNNs require extensive message-passing to propagate information such as physical forces and maintain contact awareness across the entire garment mesh, which becomes computationally inefficient at higher resolutions. To address this, we devise a novel GNN-based mesh learning framework with two key components to extend the message-passing range with minimal overhead, namely the Laplacian-Smoothed Dual Message-Passing (LSDMP) and the Geodesic Self-Attention (GSA) modules. LSDMP enhances message-passing with a Laplacian features smoothing process, which efficiently propagates the impact of each vertex to nearby vertices. Concurrently, GSA introduces geodesic distance embeddings to represent the spatial relationship between vertices and utilises attention mechanisms to capture global mesh information. The two modules operate in parallel to ensure both short- and long-range mesh modelling. Extensive experiments demonstrate the state-of-the-art performance of our method, requiring fewer layers and lower inference latency.
- 31 October 2024
- Published by Springer Nature
- p. 303-319
Identity-Consistent Diffusion Network for Grading Knee Osteoarthritis Progression in Radiographic Imaging
- 31 October 2024
- Published by Springer Nature
- p. 303-319
Identity-Consistent Diffusion Network for Grading Knee Osteoarthritis Progression in Radiographic Imaging
- Open Access
Identity-Consistent Diffusion Network for Grading Knee Osteoarthritis Progression in Radiographic Imaging
Knee osteoarthritis (KOA), a common form of arthritis that causes physical disability, has become increasingly prevalent in society. Employing computer-aided techniques to automatically assess the severity and progression of KOA can greatly benefit KOA treatment and disease management. Particularly, the advancement of X-ray technology in KOA demonstrates its potential for this purpose. Yet, existing X-ray prognosis research generally yields a singular progression severity grade, overlooking the potential visual changes for understanding and explaining the progression outcome. Therefore, in this study, a novel generative model is proposed, namely Identity-Consistent Radiographic Diffusion Network (IC-RDN), for multifaceted KOA prognosis encompassing a predicted future knee X-ray scan conditioned on the baseline scan. Specifically, an identity prior module for the diffusion and a downstream generation-guided progression prediction module are introduced. Compared to conventional image-to-image generative models, identity priors regularize and guide the diffusion to focus more on the clinical nuances of the prognosis based on a contrastive learning strategy. The progression prediction module utilizes both forecasted and baseline knee scans, and a more comprehensive formulation of KOA severity progression grading is expected. Extensive experiments on a widely used public dataset, OAI, demonstrate the effectiveness of the proposed method.
Open AccessIdentity-Consistent Diffusion Network for Grading Knee Osteoarthritis Progression in Radiographic Imaging
Knee osteoarthritis (KOA), a common form of arthritis that causes physical disability, has become increasingly prevalent in society. Employing computer-aided techniques to automatically assess the severity and progression of KOA can greatly benefit KOA treatment and disease management. Particularly, the advancement of X-ray technology in KOA demonstrates its potential for this purpose. Yet, existing X-ray prognosis research generally yields a singular progression severity grade, overlooking the potential visual changes for understanding and explaining the progression outcome. Therefore, in this study, a novel generative model is proposed, namely Identity-Consistent Radiographic Diffusion Network (IC-RDN), for multifaceted KOA prognosis encompassing a predicted future knee X-ray scan conditioned on the baseline scan. Specifically, an identity prior module for the diffusion and a downstream generation-guided progression prediction module are introduced. Compared to conventional image-to-image generative models, identity priors regularize and guide the diffusion to focus more on the clinical nuances of the prognosis based on a contrastive learning strategy. The progression prediction module utilizes both forecasted and baseline knee scans, and a more comprehensive formulation of KOA severity progression grading is expected. Extensive experiments on a widely used public dataset, OAI, demonstrate the effectiveness of the proposed method.
- Open Access
Radio Frequency Signal based Human Silhouette Segmentation: A Sequential Diffusion Approach
Radio frequency (RF) signals have been proved to be flexible for human silhouette segmentation (HSS) under complex environments. Existing studies are mainly based on a one-shot approach, which lacks a coherent projection ability from the RF domain. Additionally, the spatio-temporal patterns have not been fully explored for human motion dynamics in HSS. Therefore, we propose a two-stage Sequential Diffusion Model (SDM) to progressively synthesize high-quality segmentation jointly with the considerations on motion dynamics. Cross-view transformation blocks are devised to guide the diffusion model in a multi-scale manner for comprehensively characterizing human related patterns in an individual frame such as directional projection from signal planes. Moreover, spatio-temporal blocks are devised to fine-tune the frame-level model to incorporate spatio-temporal contexts and motion dynamics, enhancing the consistency of the segmentation maps. Comprehensive experiments on a public benchmark -- HIBER demonstrate the state-of-the-art performance of our method with an IoU 0.732. Our code is available at https://github.com/ph-w2000/SDM.
Open AccessRadio Frequency Signal based Human Silhouette Segmentation: A Sequential Diffusion Approach
Radio frequency (RF) signals have been proved to be flexible for human silhouette segmentation (HSS) under complex environments. Existing studies are mainly based on a one-shot approach, which lacks a coherent projection ability from the RF domain. Additionally, the spatio-temporal patterns have not been fully explored for human motion dynamics in HSS. Therefore, we propose a two-stage Sequential Diffusion Model (SDM) to progressively synthesize high-quality segmentation jointly with the considerations on motion dynamics. Cross-view transformation blocks are devised to guide the diffusion model in a multi-scale manner for comprehensively characterizing human related patterns in an individual frame such as directional projection from signal planes. Moreover, spatio-temporal blocks are devised to fine-tune the frame-level model to incorporate spatio-temporal contexts and motion dynamics, enhancing the consistency of the segmentation maps. Comprehensive experiments on a public benchmark -- HIBER demonstrate the state-of-the-art performance of our method with an IoU 0.732. Our code is available at https://github.com/ph-w2000/SDM.
- 2 July 2024
- Published by Elsevier in Fusion Engineering and Design
- Vol. 205
Experiments of vacuum vessel in-situ milling via mobile parallel robot machine
- 2 July 2024
- Published by Elsevier in Fusion Engineering and Design
- Vol. 205
Experiments of vacuum vessel in-situ milling via mobile parallel robot machine
- Open Access
- 12 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Plasma Science
- Vol. 52 (9) , 3468-3473
Concept Design and Development of the Multipurpose Deployer (MPD) for Large Port-Based Tokamaks
The multipurpose deployer (MPD) is a remote handling equipment for performing different tasks, such as in-vessel diagnostics, inspection, vacuum cleaning, and leak detection and localization. The structure of the MPD and its tools must be designed, considering all the aspects inside the vacuum vessel (VV) and outside. There is different but comparable equipment in the International Thermonuclear Experimental Reactor (ITER), Joint European Torus (JET), and China Fusion Engineering Test Reactor (CFETR), due to different equatorial port configurations; similarly, the concept design of this MPD must be tailored for a reference port-based tokamak configuration that can be expected for demonstration power plant (DEMO) VV. In this article, a new structure of the MPD has been proposed, combining the strengths and weaknesses of various kinds of MPDs. The structure of the MPD discussed in this article is a candidate for port-based tokamak configurations, including DEMO. The workspace meets the requirements for large area maintenance operations in candidate port-based tokamak configurations, MPD operates stably with minimal deformation under heavy loads up to 1 ton, and it meets the design requirements. Moreover, the new design MPD reduces one of the joints with the highest output torque compared to the MPD for ITER and CFETR. The MPD can be folded in cask, the packaged volume is smaller, and the deformation is lesser under the full load compared to the initial concept of the MPD.
- 12 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Plasma Science
- Vol. 52 (9) , 3468-3473
Open AccessConcept Design and Development of the Multipurpose Deployer (MPD) for Large Port-Based Tokamaks
The multipurpose deployer (MPD) is a remote handling equipment for performing different tasks, such as in-vessel diagnostics, inspection, vacuum cleaning, and leak detection and localization. The structure of the MPD and its tools must be designed, considering all the aspects inside the vacuum vessel (VV) and outside. There is different but comparable equipment in the International Thermonuclear Experimental Reactor (ITER), Joint European Torus (JET), and China Fusion Engineering Test Reactor (CFETR), due to different equatorial port configurations; similarly, the concept design of this MPD must be tailored for a reference port-based tokamak configuration that can be expected for demonstration power plant (DEMO) VV. In this article, a new structure of the MPD has been proposed, combining the strengths and weaknesses of various kinds of MPDs. The structure of the MPD discussed in this article is a candidate for port-based tokamak configurations, including DEMO. The workspace meets the requirements for large area maintenance operations in candidate port-based tokamak configurations, MPD operates stably with minimal deformation under heavy loads up to 1 ton, and it meets the design requirements. Moreover, the new design MPD reduces one of the joints with the highest output torque compared to the MPD for ITER and CFETR. The MPD can be folded in cask, the packaged volume is smaller, and the deformation is lesser under the full load compared to the initial concept of the MPD.
- Open Access
- 8 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Plasma Science
- Vol. 52 (9) , 3506-3510
Design and Development on the Remote Maintenance Tools and Strategy of the Port Closure Plate in DEMO
In the international thermonuclear experimental reactor (ITER) and the demonstration power plant (DEMO), there are three kinds of ports: the upper port (UP), the equatorial port (EP), and the lower port (LP). The applications and structures of the upper, equatorial, and LPs are different between those of ITER and DEMO. In DEMO, the breeding blankets (BBs) could be removed from the UP, a multipurpose deployer is transported and installed in the EP, and the divertor is maintained through the LP. It is noted that during remote maintenance (RM), the operating environment must be safe for the operators. To achieve this goal, reliable RM tools and RM strategies must be designed and developed. One of the containment barriers of the tokamak is the port closure plates (PCPs), which weighs between 1.3 and 25 tons depends on its location, and the PCPs need to be removed during the maintenance process. Therefore, in this article, the current designs of the PCPs are first presented. Then, the RM tools in both commercial and research stages are studied for the RM of the PCPs. Next, the RM strategies are designed and developed for each port. Finally, suggestions for the development of different RM tools are proposed.
- 8 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Plasma Science
- Vol. 52 (9) , 3506-3510
Open AccessDesign and Development on the Remote Maintenance Tools and Strategy of the Port Closure Plate in DEMO
In the international thermonuclear experimental reactor (ITER) and the demonstration power plant (DEMO), there are three kinds of ports: the upper port (UP), the equatorial port (EP), and the lower port (LP). The applications and structures of the upper, equatorial, and LPs are different between those of ITER and DEMO. In DEMO, the breeding blankets (BBs) could be removed from the UP, a multipurpose deployer is transported and installed in the EP, and the divertor is maintained through the LP. It is noted that during remote maintenance (RM), the operating environment must be safe for the operators. To achieve this goal, reliable RM tools and RM strategies must be designed and developed. One of the containment barriers of the tokamak is the port closure plates (PCPs), which weighs between 1.3 and 25 tons depends on its location, and the PCPs need to be removed during the maintenance process. Therefore, in this article, the current designs of the PCPs are first presented. Then, the RM tools in both commercial and research stages are studied for the RM of the PCPs. Next, the RM strategies are designed and developed for each port. Finally, suggestions for the development of different RM tools are proposed.
- 21 October 2023
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 167, 107598
Hierarchical cumulative network for unsupervised medical image registration
- 21 October 2023
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 167, 107598
Hierarchical cumulative network for unsupervised medical image registration
- Open Access
Tackling the Incomplete Annotation Issue in Universal Lesion Detection Task By Exploratory Training
Universal lesion detection has great value for clinical practice as it aims to detect various types of lesions in multiple organs on medical images. Deep learning methods have shown promising results, but demanding large volumes of annotated data for training. However, annotating medical images is costly and requires specialized knowledge. The diverse forms and contrasts of objects in medical images make fully annotation even more challenging, resulting in incomplete annotations. Directly training ULD detectors on such datasets can yield suboptimal results. Pseudo-label-based methods examine the training data and mine unlabelled objects for retraining, which have shown to be effective to tackle this issue. Presently, top-performing methods rely on a dynamic label-mining mechanism, operating at the mini-batch level. However, the model's performance varies at different iterations, leading to inconsistencies in the quality of the mined labels and limits their performance enhancement. Inspired by the observation that deep models learn concepts with increasing complexity, we introduce an innovative exploratory training to assess the reliability of mined lesions over time. Specifically, we introduce a teacher-student detection model as basis, where the teacher's predictions are combined with incomplete annotations to train the student. Additionally, we design a prediction bank to record high-confidence predictions. Each sample is trained several times, allowing us to get a sequence of records for each sample. If a prediction consistently appears in the record sequence, it is likely to be a true object, otherwise it may just a noise. This serves as a crucial criterion for selecting reliable mined lesions for retraining. Our experimental results substantiate that the proposed framework surpasses state-of-the-art methods on two medical image datasets, demonstrating its superior performance.
Open AccessTackling the Incomplete Annotation Issue in Universal Lesion Detection Task By Exploratory Training
Universal lesion detection has great value for clinical practice as it aims to detect various types of lesions in multiple organs on medical images. Deep learning methods have shown promising results, but demanding large volumes of annotated data for training. However, annotating medical images is costly and requires specialized knowledge. The diverse forms and contrasts of objects in medical images make fully annotation even more challenging, resulting in incomplete annotations. Directly training ULD detectors on such datasets can yield suboptimal results. Pseudo-label-based methods examine the training data and mine unlabelled objects for retraining, which have shown to be effective to tackle this issue. Presently, top-performing methods rely on a dynamic label-mining mechanism, operating at the mini-batch level. However, the model's performance varies at different iterations, leading to inconsistencies in the quality of the mined labels and limits their performance enhancement. Inspired by the observation that deep models learn concepts with increasing complexity, we introduce an innovative exploratory training to assess the reliability of mined lesions over time. Specifically, we introduce a teacher-student detection model as basis, where the teacher's predictions are combined with incomplete annotations to train the student. Additionally, we design a prediction bank to record high-confidence predictions. Each sample is trained several times, allowing us to get a sequence of records for each sample. If a prediction consistently appears in the record sequence, it is likely to be a true object, otherwise it may just a noise. This serves as a crucial criterion for selecting reliable mined lesions for retraining. Our experimental results substantiate that the proposed framework surpasses state-of-the-art methods on two medical image datasets, demonstrating its superior performance.
- Open Access
- 29 January 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (3) , 1740
Performance Analysis of Hybrid Kinematic Mechanism for Fusion Reactor Maintenance
The hybrid kinematic mechanism (HKM) as an alternative remote handling subsystem of the Demonstration Fusion Power Plant (DEMO) breeding blanket (BB) is undergoing extensive theoretical analysis and feasibility verification. In this paper, the forward and inverse kinematic models of the HKM are derived by combining the Newtonian iterative method and the analytical method. Cartesian space trajectory planning is designed based on the trajectories of the HKM lifting of inboard and outboard BBs. The continuous smooth inverse kinematic solutions in the HKM joint space are obtained based on the polynomial interpolation method. For the characteristics of the HKM piston thread driving, the end-effector position error caused by the degradation of the spherical joint into a universal joint is analyzed and calculated. During the lifting of the left inboard BB, there is a maximum absolute error ∆P = 3.1 mm, and as the error continues to expand to the bottom of the BB it causes a risk of collision. Combining the overall effects of driving control, rigid–flexible coupling, etc., on position accuracy, an open-loop variable parameter error compensation plan based on the Levenberg–Marquardt (LM) nonlinear damping least-squares algorithm is proposed and validated in this paper. The simulation results show that the maximum absolute error after compensation is less than 1 mm as the mesh density increases, and the absolute position accuracy can be further improved by local mesh encryption. This study verifies the feasibility of the HKM as a BB remote handling subsystem and provides an option for high-precision control of the HKM.
- 29 January 2023
- Published by MDPI AG in Applied Sciences
- Vol. 13 (3) , 1740
Open AccessPerformance Analysis of Hybrid Kinematic Mechanism for Fusion Reactor Maintenance
The hybrid kinematic mechanism (HKM) as an alternative remote handling subsystem of the Demonstration Fusion Power Plant (DEMO) breeding blanket (BB) is undergoing extensive theoretical analysis and feasibility verification. In this paper, the forward and inverse kinematic models of the HKM are derived by combining the Newtonian iterative method and the analytical method. Cartesian space trajectory planning is designed based on the trajectories of the HKM lifting of inboard and outboard BBs. The continuous smooth inverse kinematic solutions in the HKM joint space are obtained based on the polynomial interpolation method. For the characteristics of the HKM piston thread driving, the end-effector position error caused by the degradation of the spherical joint into a universal joint is analyzed and calculated. During the lifting of the left inboard BB, there is a maximum absolute error ∆P = 3.1 mm, and as the error continues to expand to the bottom of the BB it causes a risk of collision. Combining the overall effects of driving control, rigid–flexible coupling, etc., on position accuracy, an open-loop variable parameter error compensation plan based on the Levenberg–Marquardt (LM) nonlinear damping least-squares algorithm is proposed and validated in this paper. The simulation results show that the maximum absolute error after compensation is less than 1 mm as the mesh density increases, and the absolute position accuracy can be further improved by local mesh encryption. This study verifies the feasibility of the HKM as a BB remote handling subsystem and provides an option for high-precision control of the HKM.