Difeng Zhu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Difeng Zhu has a total of 15 co-authors. Here are the top co-authors by number of shared publications:
- Mengqi Yuan
State Key Laboratory of Explosion Science and Safety Protection, Beijing Institute of Technology, Beijing 100081, China
2 shared publications - Chenhui Li
Zhejiang University
1 shared publication - Chunyang Huang
The Department of Engineering Mechanics, Kunming University of Science and Technology, Kunming 650000, China
1 shared publication - Gunnsteinn Hall
Department of Biomedical Engineering, Johns Hopkins University, USA
1 shared publication - Heng Li
School of Optoelectronics, Beijing Institute of Technology, Beijing 100081, China
1 shared publication - Hongrui Jiang
Department of Electrical and Computer Engineering, University of WisconsinMadison, Madison, WI, USA
1 shared publication
Latest Publications
- Open Access
- 27 September 2026
- Published by Elsevier in SSRN Electronic Journal
Dynamic Mechanical Response of Zero Poisson’s Ratio Honeycombs with Circular Ring Reinforcement under impact loading
Lightweight crashworthy cores are increasingly demanded by rotorcraft and urban air mobility structures, in which sacrificial honeycomb elements must dissipate impact energy within a strict mass budget. Conventional honeycomb sandwich structures suffer lateral instability, unstable plateau responses, and reduced energy absorption under high-velocity impact. Here, a circular ring-reinforced zero-Poisson's-ratio honeycomb panel (CZPR) was additively manufactured from 316L stainless steel. Drop-weight tests at 5 m/s combined with a validated nonlinear finite element model were used to examine dynamic crushing over 5–50 m/s. Experiments show gradient-controlled layer-by-layer crushing at 5 m/s, whereas simulations at 25 and 50 m/s reveal transitions to top-layer shear instability and end-localized collapse depending on the gradient configuration. At the unit-cell level, deformation proceeds through ring ovalization, inward buckling, and densification, with energy dissipation dominated by ring deformation and supported secondarily by the horizontal and diagonal ribs. Impact-end-strengthened gradients suppress localized collapse and stabilize the plateau response. Parametric analysis shows that ring radius, rib length, and wall-thickness gradient govern contact locking, crushing initiation, and instability evolution. A macroscopic critical instability stress, identified from the coupled evolution of effective Poisson's ratio and plateau-stress degradation, was combined with a semi-empirical model calibrated by the Minimax criterion and validated by grouped cross-validation to predict instability boundaries within 10.6% mean error. The critical instability stress and the calibrated model provide a quantitative design tool for additively manufactured all-metallic honeycomb cores in lightweight aerospace impact-protection structures.
- 27 September 2026
- Published by Elsevier in SSRN Electronic Journal
Open AccessDynamic Mechanical Response of Zero Poisson’s Ratio Honeycombs with Circular Ring Reinforcement under impact loading
Lightweight crashworthy cores are increasingly demanded by rotorcraft and urban air mobility structures, in which sacrificial honeycomb elements must dissipate impact energy within a strict mass budget. Conventional honeycomb sandwich structures suffer lateral instability, unstable plateau responses, and reduced energy absorption under high-velocity impact. Here, a circular ring-reinforced zero-Poisson's-ratio honeycomb panel (CZPR) was additively manufactured from 316L stainless steel. Drop-weight tests at 5 m/s combined with a validated nonlinear finite element model were used to examine dynamic crushing over 5–50 m/s. Experiments show gradient-controlled layer-by-layer crushing at 5 m/s, whereas simulations at 25 and 50 m/s reveal transitions to top-layer shear instability and end-localized collapse depending on the gradient configuration. At the unit-cell level, deformation proceeds through ring ovalization, inward buckling, and densification, with energy dissipation dominated by ring deformation and supported secondarily by the horizontal and diagonal ribs. Impact-end-strengthened gradients suppress localized collapse and stabilize the plateau response. Parametric analysis shows that ring radius, rib length, and wall-thickness gradient govern contact locking, crushing initiation, and instability evolution. A macroscopic critical instability stress, identified from the coupled evolution of effective Poisson's ratio and plateau-stress degradation, was combined with a semi-empirical model calibrated by the Minimax criterion and validated by grouped cross-validation to predict instability boundaries within 10.6% mean error. The critical instability stress and the calibrated model provide a quantitative design tool for additively manufactured all-metallic honeycomb cores in lightweight aerospace impact-protection structures.
- Open Access
- 1 January 2023
- Published by Elsevier in SSRN Electronic Journal
Mechanical Response Prediction Model of Honeycomb Structures Under Dynamic Crushing
In this study, a novel framework was presented for accelerating the prediction of the mechanical response of honeycomb structures under dynamic crushing, using 2D cells to surrogate 3D honeycomb structures by machine learning (ML). The sizes of different honeycomb structures were designed and the necessary training data obtained through finite element (FE) simulations, but without using any explicit features of the honeycomb structures in the ML model. A pixelation method of the FE model was proposed for realizing the separation of a complete cell structure from the honeycomb FE grid data, such as nodes and elements, and convert it into a matching pixel map. FE data was structuralized while reducing the large amount of computational power consumed in identifying complex array structures. Unsupervised automatic extraction of low-dimensional features of pixel maps was performed using a convolutional denoising autoencoder (CDAE). Crushing velocity was used as an explicit external parameter, together with the extracted latent features as input to the long-short term memory (LSTM) network, to complete the prediction of the crushing deformation and stress-strain curves of a honeycomb structure under dynamic loading. Results showed that the constructed ML model could describe the dynamic crushing response behavior. Compared with the traditional FE method, the prediction model was 4.45×103 and 1.05×103 times faster in predicting the stress-strain and structural deformation response, respectively. For larger array models, this order of magnitude was more improved. The mechanical response prediction model provided a method for fast analysis and evaluation of the dynamic mechanical response behavior of similar honeycomb structures, which could be helpful for the design development of bionic structure-based designs.
- 1 January 2023
- Published by Elsevier in SSRN Electronic Journal
Open AccessMechanical Response Prediction Model of Honeycomb Structures Under Dynamic Crushing
In this study, a novel framework was presented for accelerating the prediction of the mechanical response of honeycomb structures under dynamic crushing, using 2D cells to surrogate 3D honeycomb structures by machine learning (ML). The sizes of different honeycomb structures were designed and the necessary training data obtained through finite element (FE) simulations, but without using any explicit features of the honeycomb structures in the ML model. A pixelation method of the FE model was proposed for realizing the separation of a complete cell structure from the honeycomb FE grid data, such as nodes and elements, and convert it into a matching pixel map. FE data was structuralized while reducing the large amount of computational power consumed in identifying complex array structures. Unsupervised automatic extraction of low-dimensional features of pixel maps was performed using a convolutional denoising autoencoder (CDAE). Crushing velocity was used as an explicit external parameter, together with the extracted latent features as input to the long-short term memory (LSTM) network, to complete the prediction of the crushing deformation and stress-strain curves of a honeycomb structure under dynamic loading. Results showed that the constructed ML model could describe the dynamic crushing response behavior. Compared with the traditional FE method, the prediction model was 4.45×103 and 1.05×103 times faster in predicting the stress-strain and structural deformation response, respectively. For larger array models, this order of magnitude was more improved. The mechanical response prediction model provided a method for fast analysis and evaluation of the dynamic mechanical response behavior of similar honeycomb structures, which could be helpful for the design development of bionic structure-based designs.
- 17 February 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Microelectromechanical Systems
- Vol. 21 (3) , 530-540
Three-Dimensional Surface Profile Measurement of Microlenses Using the Shack–Hartmann Wavefront Sensor
We present a 3-D surface profiling method for microlenses that utilizes a Shack-Hartmann wavefront sensor. This method applies to both solid microlenses and liquid-liquid interfaces in liquid microlenses. The wavefront at the aperture stop of a microlens is measured by a Shack-Hartmann wavefront sensor and is then used to calculate the 3-D surface profile of the microlens. Three types of microlenses-a photoresist microlens, a hydrogel-driven tunable liquid lens, and an electrowetting-driven tunable liquid lens-were fabricated and measured. The variable-focus liquid lenses were tested within a wide focal length range. The obtained surface profiles were fitted to spherical and conical surface models to study their geometrical properties. The photoresist microlens was found to be approximately spherical. For the hydrogel-driven microlens, the profile was smooth and nearly spherical at the center but became steep and linear at the aperture edges. The electrowetting-driven liquid lens was also fitted better with the conical model, and its conic constant was determined. The obtained surface profiles were used to estimate the optical properties of microlenses in an optical analysis software package. The comparison between the simulation and experiment results indicated that the accuracy of the estimation is rough and the error could be due to the wavefront measurement and surface fitting approximation.
- 17 February 2012
- Published by Institute of Electrical and Electronics Engineers (IEEE) in Journal of Microelectromechanical Systems
- Vol. 21 (3) , 530-540
Three-Dimensional Surface Profile Measurement of Microlenses Using the Shack–Hartmann Wavefront Sensor
We present a 3-D surface profiling method for microlenses that utilizes a Shack-Hartmann wavefront sensor. This method applies to both solid microlenses and liquid-liquid interfaces in liquid microlenses. The wavefront at the aperture stop of a microlens is measured by a Shack-Hartmann wavefront sensor and is then used to calculate the 3-D surface profile of the microlens. Three types of microlenses-a photoresist microlens, a hydrogel-driven tunable liquid lens, and an electrowetting-driven tunable liquid lens-were fabricated and measured. The variable-focus liquid lenses were tested within a wide focal length range. The obtained surface profiles were fitted to spherical and conical surface models to study their geometrical properties. The photoresist microlens was found to be approximately spherical. For the hydrogel-driven microlens, the profile was smooth and nearly spherical at the center but became steep and linear at the aperture edges. The electrowetting-driven liquid lens was also fitted better with the conical model, and its conic constant was determined. The obtained surface profiles were used to estimate the optical properties of microlenses in an optical analysis software package. The comparison between the simulation and experiment results indicated that the accuracy of the estimation is rough and the error could be due to the wavefront measurement and surface fitting approximation.