Liang Sun
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Liang Sun has a total of 43 co-authors. Here are the top co-authors by number of shared publications:
- Guoli Bai
School of Mechanical Engineering, Dalian University of Technology, Dalian, China
5 shared publications - Cong Cao
School of Mechanical Engineering, Dalian University of Technology, Dalian, China
3 shared publications - Wei Sun
School of Mechanical Engineering, Dalian University of Technology, Dalian, China
3 shared publications Beiyan Jia
Dalian University of Technology
2 shared publications- Cecil G. Armstrong
Queen’s University Belfast, Belfast, UK
2 shared publications - Christopher M. Tierney
Queen’s University Belfast
2 shared publications
Latest Publications
- 12 May 2026
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part L: Journal of Materials: Design and Applications
Numerical simulation of rolling contact fatigue damage in pitch bearings using a damage-coupled constitutive model
Fatigue failure of pitch bearings critically affects wind turbine reliability. This study develops a Continuum Damage Mechanics (CDM) based elasto-plastic coupled framework to investigate rolling contact fatigue (RCF) in megawatt (MW)-level pitch bearings. Implemented via ABAQUS UMAT, the model integrates damage evolution from elastic and plastic deformations, quantifying damage accumulation and material property degradation. Results show distinct nonlinear characteristics in raceway fatigue and property degradation. Notably, elastic damage is found to dominate the RCF process in large-scale bearings. Parametric analyses indicate that higher surface hardness and compressive residual stress effectively inhibit damage and extend service life. By linking macroscopic structural response with microscopic damage evolution, this work provides a robust theoretical basis for lifespan prediction and reliability enhancement of critical wind turbine components under complex loading.
- 12 May 2026
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part L: Journal of Materials: Design and Applications
Numerical simulation of rolling contact fatigue damage in pitch bearings using a damage-coupled constitutive model
Fatigue failure of pitch bearings critically affects wind turbine reliability. This study develops a Continuum Damage Mechanics (CDM) based elasto-plastic coupled framework to investigate rolling contact fatigue (RCF) in megawatt (MW)-level pitch bearings. Implemented via ABAQUS UMAT, the model integrates damage evolution from elastic and plastic deformations, quantifying damage accumulation and material property degradation. Results show distinct nonlinear characteristics in raceway fatigue and property degradation. Notably, elastic damage is found to dominate the RCF process in large-scale bearings. Parametric analyses indicate that higher surface hardness and compressive residual stress effectively inhibit damage and extend service life. By linking macroscopic structural response with microscopic damage evolution, this work provides a robust theoretical basis for lifespan prediction and reliability enhancement of critical wind turbine components under complex loading.
- 11 May 2026
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science
Load distribution and deformation of tapered roller bearing-spindle system
To improve the modeling accuracy of the tapered roller bearing-spindle system under various operating conditions, this paper proposes a bearing modeling method based on a 5 + 3n-DOF formulation, combined with one-dimensional finite element theory to establish a coupled bearing–spindle model. The model comprehensively accounts for roller profiling, tilting, and the nonlinear contact characteristics between rollers and raceways. By introducing spatial degrees of freedom for each roller, the model enables independent solutions for contact load, contact deformation, and roller angular displacement. Under typical loading conditions, the proposed model and an existing shaft system model were compared with Romax commercial software. The results show the model outperforms existing ones in spindle analysis and matches Romax closely on key contact parameters like load and deformation, with under 8% error, confirming its accuracy and practicality.
- 11 May 2026
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science
Load distribution and deformation of tapered roller bearing-spindle system
To improve the modeling accuracy of the tapered roller bearing-spindle system under various operating conditions, this paper proposes a bearing modeling method based on a 5 + 3n-DOF formulation, combined with one-dimensional finite element theory to establish a coupled bearing–spindle model. The model comprehensively accounts for roller profiling, tilting, and the nonlinear contact characteristics between rollers and raceways. By introducing spatial degrees of freedom for each roller, the model enables independent solutions for contact load, contact deformation, and roller angular displacement. Under typical loading conditions, the proposed model and an existing shaft system model were compared with Romax commercial software. The results show the model outperforms existing ones in spindle analysis and matches Romax closely on key contact parameters like load and deformation, with under 8% error, confirming its accuracy and practicality.
- 4 February 2026
- Published by SAGE Publications in Journal of Vibration and Control
A data generation method with a novel loss for fault diagnosis without target-domain data
Data-driven bearing fault diagnosis methods have achieved significant success under single working conditions, where training and testing data follow similar distributions. However, changes in bearing working conditions often cause data distribution shifts, leading to significant performance degradation. Although transfer learning methods have been widely studied to address this issue, they typically rely on abundant and high-quality target-domain data, which are difficult to obtain in practice. To overcome this limitation, this paper proposes a data generation method based on a residual autoencoder with a novel physics-consistent loss function. The proposed loss simultaneously accounts for data distribution similarity, alignment of fault characteristic frequencies, and amplitude variations, guiding the model to learn a simulation-to-reality mapping that preserves essential physical properties. Based on data from a single working condition, the proposed method can generate fault frequency spectra under unseen conditions. Experimental results on a benchmark dataset demonstrate that the generated signals effectively reproduce the physical characteristics of vibration responses under new working conditions, and using these signals for fault diagnosis achieves excellent identification performance.
- 4 February 2026
- Published by SAGE Publications in Journal of Vibration and Control
A data generation method with a novel loss for fault diagnosis without target-domain data
Data-driven bearing fault diagnosis methods have achieved significant success under single working conditions, where training and testing data follow similar distributions. However, changes in bearing working conditions often cause data distribution shifts, leading to significant performance degradation. Although transfer learning methods have been widely studied to address this issue, they typically rely on abundant and high-quality target-domain data, which are difficult to obtain in practice. To overcome this limitation, this paper proposes a data generation method based on a residual autoencoder with a novel physics-consistent loss function. The proposed loss simultaneously accounts for data distribution similarity, alignment of fault characteristic frequencies, and amplitude variations, guiding the model to learn a simulation-to-reality mapping that preserves essential physical properties. Based on data from a single working condition, the proposed method can generate fault frequency spectra under unseen conditions. Experimental results on a benchmark dataset demonstrate that the generated signals effectively reproduce the physical characteristics of vibration responses under new working conditions, and using these signals for fault diagnosis achieves excellent identification performance.
- 2 January 2026
- Published by Emerald Publishing in Engineering Computations
- Vol. 43 (3) , 1108-1128
Simplification method for domain multi-block structures generated by cross-field method
Purpose: This paper aims to improve the quality of quadrilateral meshes by developing a method for cleaning and simplifying the topology structures generated by cross-field partitioning approaches, which significantly affect final mesh quality. Design/methodology/approach: The method extracts topology information from sub-regions obtained from streamline partitioning and establishes relationships among them. Slender quadrilateral blocks and short edges are first identified through intrinsic characteristics of the sub-regions. Virtual topology operations are applied to resolve identified problem features. Findings: The method successfully identifies and resolves problematic features in block structures, including slender quadrilateral blocks, short edges, and limit cycles. Testing on various models demonstrates improved mesh quality, with results validated through comparison with automatically generated quad meshes using Ansys. Originality/value: This research contributes a novel approach to topology cleaning in cross-field-based quadrilateral mesh generation, addressing a critical challenge in automated mesh generation.
- 2 January 2026
- Published by Emerald Publishing in Engineering Computations
- Vol. 43 (3) , 1108-1128
Simplification method for domain multi-block structures generated by cross-field method
Purpose: This paper aims to improve the quality of quadrilateral meshes by developing a method for cleaning and simplifying the topology structures generated by cross-field partitioning approaches, which significantly affect final mesh quality. Design/methodology/approach: The method extracts topology information from sub-regions obtained from streamline partitioning and establishes relationships among them. Slender quadrilateral blocks and short edges are first identified through intrinsic characteristics of the sub-regions. Virtual topology operations are applied to resolve identified problem features. Findings: The method successfully identifies and resolves problematic features in block structures, including slender quadrilateral blocks, short edges, and limit cycles. Testing on various models demonstrates improved mesh quality, with results validated through comparison with automatically generated quad meshes using Ansys. Originality/value: This research contributes a novel approach to topology cleaning in cross-field-based quadrilateral mesh generation, addressing a critical challenge in automated mesh generation.
- 1 December 2025
- Published by AIP Publishing in Physics of Fluids
- Vol. 37 (12)
Aerodynamic shape optimization using the adjoint method combined with an inverse design approach for non-parameterized geometries
Most aerodynamic shape optimization frameworks operate directly on surface meshes, yielding optimized meshes rather than true computer-aided design (CAD) geometries. This CAD-free strategy requires remeshing and CAD reconstruction, which often introduces geometric inconsistencies. Direct CAD-based optimization can avoid these issues but typically depends on manually built, precisely parameterized models, making it laborious and error-prone. To overcome this limitation, this study proposes an adjoint-based aerodynamic optimization framework combined with an inverse design approach for non-parameterized geometries. Starting from neutral-format models such as STEP files, the method reconstructs parameterized CAD geometries through automated inverse modeling, enabling the definition of design variables and velocities. These are directly linked to adjoint sensitivities from DAFoam, establishing a consistent loop between geometry and flow analysis. The framework's accuracy and efficiency are demonstrated on the NACA0012 airfoil, the ONERA M6 wing, and the Rotor 37 blade, showing reliable aerodynamic improvements across different configurations.
- 1 December 2025
- Published by AIP Publishing in Physics of Fluids
- Vol. 37 (12)
Aerodynamic shape optimization using the adjoint method combined with an inverse design approach for non-parameterized geometries
Most aerodynamic shape optimization frameworks operate directly on surface meshes, yielding optimized meshes rather than true computer-aided design (CAD) geometries. This CAD-free strategy requires remeshing and CAD reconstruction, which often introduces geometric inconsistencies. Direct CAD-based optimization can avoid these issues but typically depends on manually built, precisely parameterized models, making it laborious and error-prone. To overcome this limitation, this study proposes an adjoint-based aerodynamic optimization framework combined with an inverse design approach for non-parameterized geometries. Starting from neutral-format models such as STEP files, the method reconstructs parameterized CAD geometries through automated inverse modeling, enabling the definition of design variables and velocities. These are directly linked to adjoint sensitivities from DAFoam, establishing a consistent loop between geometry and flow analysis. The framework's accuracy and efficiency are demonstrated on the NACA0012 airfoil, the ONERA M6 wing, and the Rotor 37 blade, showing reliable aerodynamic improvements across different configurations.
- 23 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (21) , 40892-40900
A Proximity-Based Unsupervised Feature Learning Framework for Rotating Machinery Sensor Data
With the advancement of health monitoring technologies in the era of smart industry, vast amounts of sensor data are continuously collected from rotating machinery. However, labeling this data remains a major bottleneck in industrial applications. This paper proposes a novel unsupervised learning framework for fault diagnosis, based on the assumption that sensor signals within adjacent time intervals exhibit high similarity in health states. By maximizing the proximity between non-overlapping, temporally adjacent sample segments, the proposed method effectively extracts discriminative features without requiring knowledge of the number of fault types. The approach is evaluated on three public benchmark datasets through unsupervised clustering and label matching. Experimental results show that the method significantly outperforms existing unsupervised techniques and achieves accurate label alignment without expert intervention.
- 23 September 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 25 (21) , 40892-40900
A Proximity-Based Unsupervised Feature Learning Framework for Rotating Machinery Sensor Data
With the advancement of health monitoring technologies in the era of smart industry, vast amounts of sensor data are continuously collected from rotating machinery. However, labeling this data remains a major bottleneck in industrial applications. This paper proposes a novel unsupervised learning framework for fault diagnosis, based on the assumption that sensor signals within adjacent time intervals exhibit high similarity in health states. By maximizing the proximity between non-overlapping, temporally adjacent sample segments, the proposed method effectively extracts discriminative features without requiring knowledge of the number of fault types. The approach is evaluated on three public benchmark datasets through unsupervised clustering and label matching. Experimental results show that the method significantly outperforms existing unsupervised techniques and achieves accurate label alignment without expert intervention.
- 27 April 2025
- Published by Springer Nature in Engineering with Computers
- Vol. 41 (5) , 3055-3069
Deep learning-based semi-supervised approach for high-quality 2D quadrilateral mesh deformation
- 27 April 2025
- Published by Springer Nature in Engineering with Computers
- Vol. 41 (5) , 3055-3069
Deep learning-based semi-supervised approach for high-quality 2D quadrilateral mesh deformation
- 28 November 2024
- Published by Springer Nature in Engineering with Computers
- Vol. 41 (2) , 1293-1308
Enhanced block-structured quadrilateral mesh generation: integrating cross-field and distance field for optimal domain decomposition
- 28 November 2024
- Published by Springer Nature in Engineering with Computers
- Vol. 41 (2) , 1293-1308
Enhanced block-structured quadrilateral mesh generation: integrating cross-field and distance field for optimal domain decomposition
- 26 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-12
Efficient Strain Analysis of Radially Loaded Hollow Cylindrical Roller Using Symbolic Regression
The development of digital twins and smart bearings in bearing industry demands efficient computational methods for factors such as strain analysis. Symbolic regression offers a promising solution by efficiently deriving analytical expressions from limited data, offering low computational cost, rapid speed, high accuracy, and strong interpretability. In this paper, symbolic regression is applied to the strain calculation problem of radially loaded hollow cylindrical rollers. The resulting expressions are evaluated and selected based on physical insights obtained from theoretical analysis and physical experiments. Compared to other methods, the selected expression retains interpretability and generalization capabilities while achieving superior computational accuracy.
- 26 August 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Instrumentation and Measurement
- Vol. 73 (00189456) , 1-12
Efficient Strain Analysis of Radially Loaded Hollow Cylindrical Roller Using Symbolic Regression
The development of digital twins and smart bearings in bearing industry demands efficient computational methods for factors such as strain analysis. Symbolic regression offers a promising solution by efficiently deriving analytical expressions from limited data, offering low computational cost, rapid speed, high accuracy, and strong interpretability. In this paper, symbolic regression is applied to the strain calculation problem of radially loaded hollow cylindrical rollers. The resulting expressions are evaluated and selected based on physical insights obtained from theoretical analysis and physical experiments. Compared to other methods, the selected expression retains interpretability and generalization capabilities while achieving superior computational accuracy.
- 30 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 24 (12) , 20054-20063
AFCNet: A Fusion Method of Multisensor Data Based on Adaptive Feature Complementation
Sensors are essential for the prognosis and health management of equipment. The fusion of multiple sensors data could improve the accuracy of fault diagnosis. Current data fusion methods focus on high-quality sensor data. However, in practical applications, constraints such as cost or space may limit the use of multiple high-quality sensors. Therefore, it is necessary to explore how to effectively use all available sensor data, even those containing some useful information but less effective when used individually. This paper presents a novel method for sensor data fusion, utilizing the highest-performing sensor data as a back bone and complementing it with preliminary data-level fusion data through multi-stage adaptive feature complementation. Comparative results validate the effectiveness and superiority of this method. The proposed method also demonstrates strong predictive capability by using data from a single period as training samples to forecast future data states.
- 30 April 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Sensors Journal
- Vol. 24 (12) , 20054-20063
AFCNet: A Fusion Method of Multisensor Data Based on Adaptive Feature Complementation
Sensors are essential for the prognosis and health management of equipment. The fusion of multiple sensors data could improve the accuracy of fault diagnosis. Current data fusion methods focus on high-quality sensor data. However, in practical applications, constraints such as cost or space may limit the use of multiple high-quality sensors. Therefore, it is necessary to explore how to effectively use all available sensor data, even those containing some useful information but less effective when used individually. This paper presents a novel method for sensor data fusion, utilizing the highest-performing sensor data as a back bone and complementing it with preliminary data-level fusion data through multi-stage adaptive feature complementation. Comparative results validate the effectiveness and superiority of this method. The proposed method also demonstrates strong predictive capability by using data from a single period as training samples to forecast future data states.