Cunfeng Kang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Cunfeng Kang has a total of 29 co-authors. Here are the top co-authors by number of shared publications:
- Guanchen Zong
College of Mechanical and Energy Engineering, Beijing University of Technology, Beijing 100124, China
4 shared publications - Shujun Chen
College of Mechanical and Energy Engineering, Beijing University of Technology, Beijing 100124, China
4 shared publications Chunmin Ma
College of Mechanical & Energy Engineering, Beijing University of Technology, People’s Republic of China
2 shared publications- Xiaoqing Jiang
College of Mechanical and Energy Engineering, Beijing University of Technology, Beijing 100124, China
2 shared publications - Baoxu Sun
College of Mechanical and Energy Engineering, Beijing University of Technology, Beijing 100124, China
1 shared publication - Bendong Liu
College of Mechanical and Energy Engineering, Beijing University of Technology 1 , Beijing 100124
1 shared publication
Latest Publications
- 24 March 2026
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture
Research on error prediction and pre-compensation method of six-axis laser cutting machine tool based on digital twin
The error prediction and pre-compensation methods for multi-axis CNC machine tools have long been a research focus in the industrial manufacturing field. Despite advancements, challenges persist, including high computational costs and poor prediction accuracy. Therefore, effective error prediction and pre-compensation methods are crucial for cost reduction and efficiency improvement. This paper proposes an error prediction and pre-compensation method, demonstrated through a six-axis machining example. First, a model based on the GRU neural network technology is proposed to predict and optimize laser focal position and directional profile errors. The method was applied in automotive chassis tubing cutting experiments to assess its feasibility and precision in practical cutting processes. The maximum absolute values of the laser focal position and directional profile errors were reduced by 51.32% and 62.50%, respectively, while the average absolute errors were reduced by 59.35% and 69.64%, verifying the correctness and accuracy of the process optimization method. To address the high computational demands, we propose a digital twin-driven GRU-based approach. This methodology establishes dynamic interactions between virtual and real processes, facilitating iterative optimization of predictive accuracy and enhancing the level of manufacturing digitalization.
- 24 March 2026
- Published by SAGE Publications in Proceedings of the Institution of Mechanical Engineers, Part B: Journal of Engineering Manufacture
Research on error prediction and pre-compensation method of six-axis laser cutting machine tool based on digital twin
The error prediction and pre-compensation methods for multi-axis CNC machine tools have long been a research focus in the industrial manufacturing field. Despite advancements, challenges persist, including high computational costs and poor prediction accuracy. Therefore, effective error prediction and pre-compensation methods are crucial for cost reduction and efficiency improvement. This paper proposes an error prediction and pre-compensation method, demonstrated through a six-axis machining example. First, a model based on the GRU neural network technology is proposed to predict and optimize laser focal position and directional profile errors. The method was applied in automotive chassis tubing cutting experiments to assess its feasibility and precision in practical cutting processes. The maximum absolute values of the laser focal position and directional profile errors were reduced by 51.32% and 62.50%, respectively, while the average absolute errors were reduced by 59.35% and 69.64%, verifying the correctness and accuracy of the process optimization method. To address the high computational demands, we propose a digital twin-driven GRU-based approach. This methodology establishes dynamic interactions between virtual and real processes, facilitating iterative optimization of predictive accuracy and enhancing the level of manufacturing digitalization.
- Open Access
Pillar-Bin: A 3D Object Detection Algorithm for Communication-Denied UGVs
Addressing the challenge of acquiring high-precision leader Unmanned Ground Vehicle (UGV) pose information in real time for communication-denied leader–follower formations, this study proposed Pillar-Bin, a 3D object detection algorithm based on the PointPillars framework. Pillar-Bin introduced an Interval Discretization Strategy (Bin) within the detection head, mapping critical target parameters (dimensions, center, heading angle) to predefined intervals for joint classification-residual regression optimization. This effectively suppresses environmental noise and enhances localization accuracy. Simulation results on the KITTI dataset demonstrate that the Pillar-Bin algorithm significantly outperforms PointPillars in detection accuracy. In the 3D detection mode, the mean Average Precision (mAP) increased by 2.95%, while in the bird’s eye view (BEV) detection mode, mAP was improved by 0.94%. With a processing rate of 48 frames per second (FPS), the proposed algorithm effectively enhanced detection accuracy while maintaining the high real-time performance of the baseline method. To evaluate Pillar-Bin’s real-vehicle performance, a leader UGV pose extraction scheme was designed. Real-vehicle experiments show absolute X/Y positioning errors below 5 cm and heading angle errors under 5° in Cartesian coordinates, with the pose extraction processing speed reaching 46 FPS. The proposed Pillar-Bin algorithm and its pose extraction scheme provide efficient and accurate leader pose information for formation control, demonstrating practical engineering utility.
Open AccessPillar-Bin: A 3D Object Detection Algorithm for Communication-Denied UGVs
Addressing the challenge of acquiring high-precision leader Unmanned Ground Vehicle (UGV) pose information in real time for communication-denied leader–follower formations, this study proposed Pillar-Bin, a 3D object detection algorithm based on the PointPillars framework. Pillar-Bin introduced an Interval Discretization Strategy (Bin) within the detection head, mapping critical target parameters (dimensions, center, heading angle) to predefined intervals for joint classification-residual regression optimization. This effectively suppresses environmental noise and enhances localization accuracy. Simulation results on the KITTI dataset demonstrate that the Pillar-Bin algorithm significantly outperforms PointPillars in detection accuracy. In the 3D detection mode, the mean Average Precision (mAP) increased by 2.95%, while in the bird’s eye view (BEV) detection mode, mAP was improved by 0.94%. With a processing rate of 48 frames per second (FPS), the proposed algorithm effectively enhanced detection accuracy while maintaining the high real-time performance of the baseline method. To evaluate Pillar-Bin’s real-vehicle performance, a leader UGV pose extraction scheme was designed. Real-vehicle experiments show absolute X/Y positioning errors below 5 cm and heading angle errors under 5° in Cartesian coordinates, with the pose extraction processing speed reaching 46 FPS. The proposed Pillar-Bin algorithm and its pose extraction scheme provide efficient and accurate leader pose information for formation control, demonstrating practical engineering utility.
- Open Access
Development of an Improved Stiffness Ellipsoid Method for Precise Robot-Positioner Collaborative Control in Friction Stir Welding
This study proposes an improved stiffness ellipsoid method to enhance the stiffness and precision of robotic arms in friction stir welding (FSW) operations. The method involves establishing a joint stiffness model through static identification experiments and developing a novel stiffness index derived from the improved stiffness ellipsoid method. This index provides a refined metric for evaluating the robot’s performance under variable loads during FSW. Simulation experiments demonstrate significant improvements in welding trajectory precision and computational efficiency. The findings highlight the potential of this method to elevate FW quality and consistency.
Open AccessDevelopment of an Improved Stiffness Ellipsoid Method for Precise Robot-Positioner Collaborative Control in Friction Stir Welding
This study proposes an improved stiffness ellipsoid method to enhance the stiffness and precision of robotic arms in friction stir welding (FSW) operations. The method involves establishing a joint stiffness model through static identification experiments and developing a novel stiffness index derived from the improved stiffness ellipsoid method. This index provides a refined metric for evaluating the robot’s performance under variable loads during FSW. Simulation experiments demonstrate significant improvements in welding trajectory precision and computational efficiency. The findings highlight the potential of this method to elevate FW quality and consistency.
- 27 November 2024
- Published by World Scientific Pub Co Pte Ltd in Modern Physics Letters B
- Vol. 39 (20)
A thermal insulation microactuator for Braille display using paraffin wax/carbon nanotube composite with induction heating
A new thermal insulation microactuator for Braille display using a phase change material (PCM) composite with induction heating is proposed. The PCM composite is prepared by fully mixing paraffin with multi-walled carbon nanotubes (MWCNTs), which is mainly used as both electrically and thermally conductive enhancer. As a result of that, the PCM composite is conductive and a wireless induction heating can be realized with the paraffin composite, which avoids the connection of wires from the inner chamber with a relatively simple fabrication process. In order to prevent the penetration of the paraffin from the elastic film, a solvent-resistant TPU diaphragm is adopted to seal the PCM chamber. In addition, thermal insulation between the higher temperature of PCM and the finger contact is designed by adding the PDMS diaphragm with PDMS pins. When the input power is 1.29[ ]W at AC frequency of 1[ ]kHz, the actuation height of the Braille dot reaches 259[ ][ ]m within 7[ ]s. The microactuator for Braille display presented in this paper has the advantages of short response time, thermal isolation, simple fabrication, small size and low cost.
- 27 November 2024
- Published by World Scientific Pub Co Pte Ltd in Modern Physics Letters B
- Vol. 39 (20)
A thermal insulation microactuator for Braille display using paraffin wax/carbon nanotube composite with induction heating
A new thermal insulation microactuator for Braille display using a phase change material (PCM) composite with induction heating is proposed. The PCM composite is prepared by fully mixing paraffin with multi-walled carbon nanotubes (MWCNTs), which is mainly used as both electrically and thermally conductive enhancer. As a result of that, the PCM composite is conductive and a wireless induction heating can be realized with the paraffin composite, which avoids the connection of wires from the inner chamber with a relatively simple fabrication process. In order to prevent the penetration of the paraffin from the elastic film, a solvent-resistant TPU diaphragm is adopted to seal the PCM chamber. In addition, thermal insulation between the higher temperature of PCM and the finger contact is designed by adding the PDMS diaphragm with PDMS pins. When the input power is 1.29[ ]W at AC frequency of 1[ ]kHz, the actuation height of the Braille dot reaches 259[ ][ ]m within 7[ ]s. The microactuator for Braille display presented in this paper has the advantages of short response time, thermal isolation, simple fabrication, small size and low cost.
- Open Access
Research on the Mechanism and Processability of Roll Forming
Cold bending forming is a complex forming process, and its product quality is closely related to the forming process parameters. To mitigate issues such as bulging and waviness arising from the extension of the material at the edges during the forming process of thin-walled circular tubes, a comprehensive comparative analysis was conducted on four forming methods. This analysis determined that the combined bending method is the optimal forming technique for the equipment. For the impact of different parameters on the equivalent plastic strain distribution of the product and the force on the rollers, numerical simulations were carried out using the software COPRA (COPRA FEA RF 2023.1) after designing the pattern diagram based on the integrated bending method. The results showed that different processing speeds on the equivalent plastic strain distribution and work hardening of the plate have little effect. As the spacing between the upper and lower rollers increases, the equivalent plastic strain of the plate to a certain extent and the value of the moment of the rollers is significantly reduced. Analyzing the performance characteristics of high-strength steel materials from the aspects of the thickness strain and cross-sectional forming of the plate, this verifies the advantages of forming high-strength steel plates. The numerical simulation results of this study are in good agreement with actual production experimental results.
Open AccessResearch on the Mechanism and Processability of Roll Forming
Cold bending forming is a complex forming process, and its product quality is closely related to the forming process parameters. To mitigate issues such as bulging and waviness arising from the extension of the material at the edges during the forming process of thin-walled circular tubes, a comprehensive comparative analysis was conducted on four forming methods. This analysis determined that the combined bending method is the optimal forming technique for the equipment. For the impact of different parameters on the equivalent plastic strain distribution of the product and the force on the rollers, numerical simulations were carried out using the software COPRA (COPRA FEA RF 2023.1) after designing the pattern diagram based on the integrated bending method. The results showed that different processing speeds on the equivalent plastic strain distribution and work hardening of the plate have little effect. As the spacing between the upper and lower rollers increases, the equivalent plastic strain of the plate to a certain extent and the value of the moment of the rollers is significantly reduced. Analyzing the performance characteristics of high-strength steel materials from the aspects of the thickness strain and cross-sectional forming of the plate, this verifies the advantages of forming high-strength steel plates. The numerical simulation results of this study are in good agreement with actual production experimental results.
- Open Access
Active Vibration Avoidance Method for Variable Speed Welding in Robotic Friction Stir Welding Based on Constant Heat Input
Robotic Friction Stir Welding (RFSW) technology integrates the advantages of friction stir welding and industrial robots, finding extensive applications and research in aerospace, shipbuilding, and new energy vehicles. However, the high-speed rotational process of friction stir welding combined with the low stiffness characteristics of serial industrial robots inevitably introduces vibrations during the welding process. This paper investigates the vibration patterns and impacts during the RFSW process and proposes an active vibration avoidance control method for variable speed welding based on constant heat input. This method utilizes a vibration feedback strategy that adjusts the spindle speed actively if the end-effector’s vibration exceeds a threshold, thereby avoiding the modal frequencies of the robot at its current pose. Concurrently, it calculates and adjusts the welding speed of the robot according to the thermal equilibrium equation to maintain constant heat input. A simplified dynamic model of the RFSW robot was established, and the feasibility of this method was validated through simulation experiments. This study fills the gap in vibration analysis of RFSW and provides new insights into control strategies and process optimization for robotic friction stir welding.
Open AccessActive Vibration Avoidance Method for Variable Speed Welding in Robotic Friction Stir Welding Based on Constant Heat Input
Robotic Friction Stir Welding (RFSW) technology integrates the advantages of friction stir welding and industrial robots, finding extensive applications and research in aerospace, shipbuilding, and new energy vehicles. However, the high-speed rotational process of friction stir welding combined with the low stiffness characteristics of serial industrial robots inevitably introduces vibrations during the welding process. This paper investigates the vibration patterns and impacts during the RFSW process and proposes an active vibration avoidance control method for variable speed welding based on constant heat input. This method utilizes a vibration feedback strategy that adjusts the spindle speed actively if the end-effector’s vibration exceeds a threshold, thereby avoiding the modal frequencies of the robot at its current pose. Concurrently, it calculates and adjusts the welding speed of the robot according to the thermal equilibrium equation to maintain constant heat input. A simplified dynamic model of the RFSW robot was established, and the feasibility of this method was validated through simulation experiments. This study fills the gap in vibration analysis of RFSW and provides new insights into control strategies and process optimization for robotic friction stir welding.
- Open Access
Optimization of Installation Position for Complex Space Curve Weldments in Robotic Friction Stir Welding Based on Dynamic Dual Particle Swarm Optimization
Robotic friction stir welding (RFSW), with its wide application range, ample working space, and task flexibility, has emerged as a vital development in friction stir welding (FSW) technology. However, the low stiffness of serial industrial robots can lead to end-effector deviations and vibrations during FSW tasks, adversely affecting the weld quality. This paper proposes a dynamic dual particle swarm optimization (DDPSO) algorithm through a new comprehensive stability index that considers both the stiffness and vibration stability of the robot to optimize the installation position of complex space curve weldments, thereby enhancing the robot’s stability during the FSW process. The algorithm employs two independent particle swarms for exploration and exploitation tasks and dynamically adjusts task allocation and particle numbers based on current results to fully utilize computational resources and enhance search efficiency. Compared to the standard particle swarm optimization (PSO) algorithm, the DDPSO approach demonstrated superior search capabilities and stability of optimization results. The maximum fitness value improved by 4.2%, the average value increased by 12.74%, and the concentration level of optimization results rose by 72.91% on average. The new optimization method pioneers fresh perspectives for optimizing the stability of RFSW, providing significant grounds for the process optimization and offline programming of complex spatial curve weldments.
Open AccessOptimization of Installation Position for Complex Space Curve Weldments in Robotic Friction Stir Welding Based on Dynamic Dual Particle Swarm Optimization
Robotic friction stir welding (RFSW), with its wide application range, ample working space, and task flexibility, has emerged as a vital development in friction stir welding (FSW) technology. However, the low stiffness of serial industrial robots can lead to end-effector deviations and vibrations during FSW tasks, adversely affecting the weld quality. This paper proposes a dynamic dual particle swarm optimization (DDPSO) algorithm through a new comprehensive stability index that considers both the stiffness and vibration stability of the robot to optimize the installation position of complex space curve weldments, thereby enhancing the robot’s stability during the FSW process. The algorithm employs two independent particle swarms for exploration and exploitation tasks and dynamically adjusts task allocation and particle numbers based on current results to fully utilize computational resources and enhance search efficiency. Compared to the standard particle swarm optimization (PSO) algorithm, the DDPSO approach demonstrated superior search capabilities and stability of optimization results. The maximum fitness value improved by 4.2%, the average value increased by 12.74%, and the concentration level of optimization results rose by 72.91% on average. The new optimization method pioneers fresh perspectives for optimizing the stability of RFSW, providing significant grounds for the process optimization and offline programming of complex spatial curve weldments.
- Open Access
Dynamic Modeling, Simulation, and Optimization of Vehicle Electronic Stability Program Algorithm Based on Back Propagation Neural Network and PID Algorithm
The vehicle lateral stability control algorithm is an essential component of the electronic stability program (ESP), and its control effect directly affects the vehicle’s driving safety. However, there are still numerous shortcomings and challenges that need to be addressed, including enhancing the efficiency of processing intricate pavement condition data, improving the accuracy of parameter adjustment, and identifying subtle and elusive patterns amidst noisy and ambiguous data. The introduction of machine learning algorithms can address the aforementioned issues, making it imperative to apply machine learning to the research of lateral stability control algorithms. This paper presented a vehicle lateral electronic stability control algorithm based on the back propagation (BP) neural network and PID control algorithm. Firstly, the dynamics of the whole vehicle have been analytically modeled. Then, a 2 DOF prediction model and a 14 DOF simulation model were built in MATLAB Simulink to simulate the data of the electronic control units (ECU) in ESP and estimate the dynamic performance of the real vehicle. In addition, the self-correction of the PID algorithm was verified by a Simulink/CarSim combined simulation. The improvement of the BP neural network to the traditional PID algorithm was also analyzed in Simulink. These simulation results show the self-correction of the PID algorithm on the lateral stability control of the vehicle under different road conditions and at different vehicle speeds. The BP neural network smoothed the vehicle trajectory controlled by traditional PID and improved the self-correction ability of the control system by iterative training. Furthermore, it shows that the algorithm can automatically tune the control parameters and optimize the control process of the lateral electronic stability control algorithm, thus improving vehicle stability and adapting it to many different vehicle models and road conditions. Therefore, the algorithm has a high practical value and provides a feasible idea for developing a more intelligent and general vehicle lateral electronic stability system.
Open AccessDynamic Modeling, Simulation, and Optimization of Vehicle Electronic Stability Program Algorithm Based on Back Propagation Neural Network and PID Algorithm
The vehicle lateral stability control algorithm is an essential component of the electronic stability program (ESP), and its control effect directly affects the vehicle’s driving safety. However, there are still numerous shortcomings and challenges that need to be addressed, including enhancing the efficiency of processing intricate pavement condition data, improving the accuracy of parameter adjustment, and identifying subtle and elusive patterns amidst noisy and ambiguous data. The introduction of machine learning algorithms can address the aforementioned issues, making it imperative to apply machine learning to the research of lateral stability control algorithms. This paper presented a vehicle lateral electronic stability control algorithm based on the back propagation (BP) neural network and PID control algorithm. Firstly, the dynamics of the whole vehicle have been analytically modeled. Then, a 2 DOF prediction model and a 14 DOF simulation model were built in MATLAB Simulink to simulate the data of the electronic control units (ECU) in ESP and estimate the dynamic performance of the real vehicle. In addition, the self-correction of the PID algorithm was verified by a Simulink/CarSim combined simulation. The improvement of the BP neural network to the traditional PID algorithm was also analyzed in Simulink. These simulation results show the self-correction of the PID algorithm on the lateral stability control of the vehicle under different road conditions and at different vehicle speeds. The BP neural network smoothed the vehicle trajectory controlled by traditional PID and improved the self-correction ability of the control system by iterative training. Furthermore, it shows that the algorithm can automatically tune the control parameters and optimize the control process of the lateral electronic stability control algorithm, thus improving vehicle stability and adapting it to many different vehicle models and road conditions. Therefore, the algorithm has a high practical value and provides a feasible idea for developing a more intelligent and general vehicle lateral electronic stability system.
- Open Access
Digital Twin Virtual Welding Approach of Robotic Friction Stir Welding Based on Co-Simulation of FEA Model and Robotic Model
Robotic friction stir welding has become an important research direction in friction stir welding technology. However, the low stiffness of serial industrial robots leads to substantial, difficult-to-measure end-effector deviations under the welding forces during the friction stir welding process, impacting the welding quality. To more effectively measure the deviations in the end-effector, this study introduces a digital twin model based on the five-dimensional digital twin theory. The model obtains the current data of the robot and six-axis force sensor and calculates the real-time end deviations using the robot model. Based on this, a virtual welding model was realized by integrating the FEA model with the digital twin model using a co-simulation approach. This model achieves pre-process simulation by iteratively cycling through the simulated force from the FEA model and the end displacement from the robot model. The virtual welding model effectively predicts the welding outcomes with a mere 6.9% error in lateral deviation compared to actual welding, demonstrating its potential in optimizing welding parameters and enhancing accuracy and quality. Employing digital twin models to monitor, simulate, and optimize the welding process can reduce risks, save costs, and improve efficiency, providing new perspectives for optimizing robotic friction stir welding processes.
Open AccessDigital Twin Virtual Welding Approach of Robotic Friction Stir Welding Based on Co-Simulation of FEA Model and Robotic Model
Robotic friction stir welding has become an important research direction in friction stir welding technology. However, the low stiffness of serial industrial robots leads to substantial, difficult-to-measure end-effector deviations under the welding forces during the friction stir welding process, impacting the welding quality. To more effectively measure the deviations in the end-effector, this study introduces a digital twin model based on the five-dimensional digital twin theory. The model obtains the current data of the robot and six-axis force sensor and calculates the real-time end deviations using the robot model. Based on this, a virtual welding model was realized by integrating the FEA model with the digital twin model using a co-simulation approach. This model achieves pre-process simulation by iteratively cycling through the simulated force from the FEA model and the end displacement from the robot model. The virtual welding model effectively predicts the welding outcomes with a mere 6.9% error in lateral deviation compared to actual welding, demonstrating its potential in optimizing welding parameters and enhancing accuracy and quality. Employing digital twin models to monitor, simulate, and optimize the welding process can reduce risks, save costs, and improve efficiency, providing new perspectives for optimizing robotic friction stir welding processes.
- Open Access
Trajectory Planning and Simulation Study of Redundant Robotic Arm for Upper Limb Rehabilitation Based on Back Propagation Neural Network and Genetic Algorithm
In this study, a Back Propagation (BP) neural network algorithm based on Genetic Algorithm (GA) optimization is proposed to plan and optimize the trajectory of a redundant robotic arm for the upper limb rehabilitation of patients. The feasibility of the trajectory was verified by numerical simulations. First, the collected dataset was used to train the BP neural network optimized by the GA. Subsequently, the critical points designated by the rehabilitation physician for the upper limb rehabilitation were used as interpolation points for cubic B−spline interpolation to plan the motion trajectory. The GA optimized the planned trajectory with the goal of time minimization, and the feasibility of the optimized trajectory was analyzed with MATLAB simulations. The planned trajectory was smooth and continuous. There was no abrupt change in location or speed. Finally, simulations revealed that the optimized trajectory reduced the motion time and increased the motion speed between two adjacent critical points which improved the rehabilitation effect and can be applied to patients with different needs, which has high application value.
Open AccessTrajectory Planning and Simulation Study of Redundant Robotic Arm for Upper Limb Rehabilitation Based on Back Propagation Neural Network and Genetic Algorithm
In this study, a Back Propagation (BP) neural network algorithm based on Genetic Algorithm (GA) optimization is proposed to plan and optimize the trajectory of a redundant robotic arm for the upper limb rehabilitation of patients. The feasibility of the trajectory was verified by numerical simulations. First, the collected dataset was used to train the BP neural network optimized by the GA. Subsequently, the critical points designated by the rehabilitation physician for the upper limb rehabilitation were used as interpolation points for cubic B−spline interpolation to plan the motion trajectory. The GA optimized the planned trajectory with the goal of time minimization, and the feasibility of the optimized trajectory was analyzed with MATLAB simulations. The planned trajectory was smooth and continuous. There was no abrupt change in location or speed. Finally, simulations revealed that the optimized trajectory reduced the motion time and increased the motion speed between two adjacent critical points which improved the rehabilitation effect and can be applied to patients with different needs, which has high application value.