An Efficient MPC-CPG Bionic Controller for Periodic Motions and Bounded Transitions

Xiaokang Li,Shuxin Wang,Yuhong Liu,Huan Bai,Zeyi Zhang 제어·로봇·시스템학회 2024 International Journal of Control, Automation, and Vol.22 No.6

The traditional controllers for performing desired periodic motions are often computationally intensive and can hardly satisfy the constraints of the control system. This article proposes a bionic controller for such motion patterns inspired by the control mechanism of central pattern generators (CPGs) found in vertebrates. The bionic controller combines the advantages of model predictive control (MPC) and CPGs to track arbitrary periodic trajectories, with smooth transitions between different gaits under constraints. The controller consists of three typical components: the CPG component as the inner loop to generate steady-state periodic motions, the MPC-CPG component as the outer loop for gait transitions, and the decision component to determine which loop should be utilized. The stability of the controller is demonstrated through Lyapunov’s method. The controller is then used to track a sinusoidal trajectory under various constraints, and its performance is compared to that of the separate CPG and MPC controllers. The results demonstrate the effectiveness of the proposed controller in tracking periodic motions with constraints, providing a promising approach for developing more efficient and robust controllers for periodic motions.

Small Sample Face Recognition Algorithm Based on Novel Siamese Network

Zhang, Jianming,Jin, Xiaokang,Liu, Yukai,Sangaiah, Arun Kumar,Wang, Jin Korea Information Processing Society 2018 Journal of information processing systems Vol.14 No.6

In face recognition, sometimes the number of available training samples for single category is insufficient. Therefore, the performances of models trained by convolutional neural network are not ideal. The small sample face recognition algorithm based on novel Siamese network is proposed in this paper, which doesn't need rich samples for training. The algorithm designs and realizes a new Siamese network model, SiameseFacel, which uses pairs of face images as inputs and maps them to target space so that the $L_2$ norm distance in target space can represent the semantic distance in input space. The mapping is represented by the neural network in supervised learning. Moreover, a more lightweight Siamese network model, SiameseFace2, is designed to reduce the network parameters without losing accuracy. We also present a new method to generate training data and expand the number of training samples for single category in AR and labeled faces in the wild (LFW) datasets, which improves the recognition accuracy of the models. Four loss functions are adopted to carry out experiments on AR and LFW datasets. The results show that the contrastive loss function combined with new Siamese network model in this paper can effectively improve the accuracy of face recognition.

Small Sample Face Recognition Algorithm based on Novel Siamese Network

( Jianming Zhang ),( Xiaokang Jin ),( Yukai Liu ),( Arun Kumar Sangaiah ),( Jin Wang ) 한국정보처리학회 2018 Journal of information processing systems Vol.14 No.6

In face recognition, sometimes the number of available training samples for single category is insufficient. Therefore, the performances of models trained by convolutional neural network are not ideal. The small sample face recognition algorithm based on novel Siamese network is proposed in this paper, which doesn’t need rich samples for training. The algorithm designs and realizes a new Siamese network model, SiameseFace1, which uses pairs of face images as inputs and maps them to target space so that the L2 norm distance in target space can represent the semantic distance in input space. The mapping is represented by the neural network in supervised learning. Moreover, a more lightweight Siamese network model, SiameseFace2, is designed to reduce the network parameters without losing accuracy. We also present a new method to generate training data and expand the number of training samples for single category in AR and labeled faces in the wild (LFW) datasets, which improves the recognition accuracy of the models. Four loss functions are adopted to carry out experiments on AR and LFW datasets. The results show that the contrastive loss function combined with new Siamese network model in this paper can effectively improve the accuracy of face recognition.

Preoperative Adjacent Facet Joint Osteoarthritis Is Associated with the Incidence of Adjacent Segment Degeneration and Low Back Pain after Lumbar Interbody Fusion

Tan Lixian,Du Xiaokang,Tang Runmin,Rong Limin,Zhang Liangming 대한척추외과학회 2024 Asian Spine Journal Vol.18 No.1

Study Design: A retrospective cohort study. Purpose: To analyze the association between preoperative adjacent facet joint osteoarthritis (FJOA) and outcomes of lumbar interbody fusion (LIF). Overview of Literature: Whether preoperative adjacent FJOA is associated with the incidence of radiological adjacent segment degeneration (RASD) and low back pain (LBP) relief after lumbar fusion remains unknown. Methods: The study included patients who underwent LIF. The demographic characteristics and radiographic and surgical data were collected and evaluated. The included patients were divided into control group and FJOA group based on the preoperative adjacent facet joint Pathria grade. Preoperative and last follow-up LBP Visual Analog Scale (VAS) score, leg pain (LP) VAS, Oswestry Disability Index (ODI) and RASD were evaluated and compared. The improvement rates in VAS and ODI were calculated and compared between the two groups. Logistic regression was used to analyze the risk factors of LBP relief and incidence of RASD. Results: In total, 197 patients (control group, 86; FJOA group, 111) were included, and the median follow-up was 46 months. The VAS and ODI in both groups significantly improved after surgery. At the last follow-up, the FJOA group had higher VAS and lower VAS improvement rates of LBP than the control group (p<0.05). However, no significant difference in the LP VAS and ODI was found between the two groups. The incidence of RASD in the FJOA group was significantly higher than that in the control group (48.6% vs. 30.2%, p=0.034). Multivariate logistic regression analysis showed that preoperative adjacent FJOA was significantly associated with LBP relief (odds ratio [OR], 0.691; 95% confidence interval [CI], 0.498–0.958) and the postoperative incidence of RASD (OR, 1.406; 95% CI, 1.020–1.939). Conclusions: The preoperative FJOA in the adjacent segments was significantly associated with LBP following LIF. Patients with preoperative FJOA were more likely to have RASD following lumbar fusion surgery.

Methodology effects on determining the energy concentration and the apparent total tract digestibility of components in diets fed to growing pigs

Huang, Chengfei,Li, Ping,Ma, Xiaokang,Jaworski, Neil William,Stein, Hans-Henrik,Lai, Changhua,Zhao, Jinbiao,Zhang, Shuai Asian Australasian Association of Animal Productio 2018 Animal Bioscience Vol.31 No.8

Objective: An experiment was conducted to investigate the effects of different diet formulations: F1 (Two complicated basal diets containing different crude protein levels plus tested feedstuff) vs F2 (A simple corn soybean meal [SBM] basal diet plus tested feedstuff) combined with total collection (TC) or chromic oxide ($Cr_2O_3$) marker or acid-insoluble ash (AIA) marker method, and freeze-dry or oven-dry (OD) technique on estimation of nutrient digestibility in diets fed to growing pigs. Methods: In F1, twelve barrows were allocated to two $6{\times}4$ Youden Squares. The treatment diets included a high protein basal (HPB) diet, a low protein basal (LPB) diet, a corn diet and a wheat bran (WB) diet formulated based on the HPB diet, and a SBM diet and a rapeseed meal (RSM) diet formulated based on the LPB diet. In F2, eight barrows were allocated to two $4{\times}4$ Latin Squares. The treatment diets included a corn basal diet, a SBM basal diet formulated based on the corn diet, and a WB diet and a RSM diet formulated based on the SBM diet. Results: Concentration of digestible (DE) and metabolizable energy (ME), and the apparent total tract digestibility of gross energy, ash, neutral detergent fibre, and acid detergent fibre determined by $Cr_2O_3$ marker method were greater than those determined by TC and AIA marker methods in HPB, LPB, and RSM diets formulated by F1 and in corn diet formulated by F2 (p<0.05). The DE values in WB and both DE and ME values in SBM and RSM estimated using F1 were greater than those estimated using F2 (p<0.05). Conclusion: From the accuracy aspect, the AIA marker or TC method combined with OD technique is recommended for determining the energy concentration and nutrient digestibility of components in diets fed to growing pigs.

Effect of Nanoparticle Size on the Interface Bond Energy in KTa0.5Nb0.5O3 / Polyimide Composites

Chuntian Chen,Juan Wang,Wenlong Yang,Xiaokang Li,Jingyun Zhang 보안공학연구지원센터 2014 International Journal of Hybrid Information Techno Vol.7 No.6

The effects of particle size on the interface bond energy in KTa0.5Nb0.5O3/ polyimide composites were investigated using the molecular dynamics method. The composites of polyimide and the KTa0.5Nb0.5O3 nanoparticles with different size (0.4 nm, 0.5 nm, 0.6 nm, 0.8 nm, 1.2 nm, and 1.4 nm) have been established and simulated using the Forcite program package of Materials Studio software. The calculated results show that the interface bond energy is 38~150 Kcal/mol, which indicate the Van der Waals force exist between the nanoparticle and polyimide matrix. The interface bond energy of the smaller nanoparticle is greater than the lager nanaparticle due to the size effect. The introduction of the smaller nanoparticle could generate the stronger intermolecular bonding interaction between the nanoparticle and polyimide matrix.

A Sensorless Rotor Position Estimation Scheme for IPMSM Using HF Signal Injection with Frequency and Amplitude Optimization

Lu, Jiadong,Liu, Jinglin,Hu, Yihua,Zhang, Xiaokang,Ni, Kai,Si, Jikai The Korean Institute of Electrical Engineers 2018 Journal of Electrical Engineering & Technology Vol.13 No.5

High frequency signal injection (HFI) is an alternative method for estimating rotor position of interior permanent magnet synchronous motor (IPMSM). The general method of frequency and amplitude selection is based on error tolerance and experiments, and is usually set with only one group of HF parameters, which is not efficient for different working modes. This paper proposes a novel rotor position estimation scheme by HFI with optimized frequency and amplitude, based on the mathematic model of IPMSM. The requirements for standstill and low-speed operational modes are met by applying this novel scheme. Additionally, the effects of the frequency and amplitude of the injected HF signal on the position estimation results under different operating conditions are analyzed. Furthermore, an optimization method for HF parameter selection is proposed to make the estimation process more efficient under different working conditions according to error tolerance. The effectiveness of the propose scheme is verified by the experiments on an IPMSM motor prototype.

A Sensorless Rotor Position Estimation Scheme for IPMSM Using HF Signal Injection with Frequency and Amplitude Optimization

Jiadong Lu,Jinglin Liu,Yihua Hu,Xiaokang Zhang,Kai Ni,Jikai Si 대한전기학회 2018 Journal of Electrical Engineering & Technology Vol.13 No.5

High frequency signal injection (HFI) is an alternative method for estimating rotor position of interior permanent magnet synchronous motor (IPMSM). The general method of frequency and amplitude selection is based on error tolerance and experiments, and is usually set with only one group of HF parameters, which is not efficient for different working modes. This paper proposes a novel rotor position estimation scheme by HFI with optimized frequency and amplitude, based on the mathematic model of IPMSM. The requirements for standstill and low-speed operational modes are met by applying this novel scheme. Additionally, the effects of the frequency and amplitude of the injected HF signal on the position estimation results under different operating conditions are analyzed. Furthermore, an optimization method for HF parameter selection is proposed to make the estimation process more efficient under different working conditions according to error tolerance. The effectiveness of the propose scheme is verified by the experiments on an IPMSM motor prototype.