Shengzhi Yuan

Shengzhi Yuan

Auto-generated profile, based on publicly available information

Shengzhi Yuan

College of Weaponry Engineering, Naval University of Engineering, Wuhan 430033, China

Research Focus & Metrics

Academic Career Length

16Years (2011-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
15 August 2023 in Optical Fiber Technology
Verified
Journal Article
Closed access
SWSen Wang
Sen Wang
Renlai ZhouRenlai Zhou
Renlai Zhou
ZDZhenyu Duan
Zhenyu Duan
GHGuanqu Hu
Guanqu Hu
SYShenzhi Yuan
Shenzhi Yuan
Nakkeeran KaliyaperumalNakkeeran Kaliyaperumal
Nakkeeran Kaliyaperumal

11 January 2024 in International Journal of Aerospace Engineering
Verified
Journal Article
Open access
PDF file
JLJing Li
Jing Li
XLXinru Liang
Xinru Liang
SYShengzhi Yuan
Shengzhi Yuan
HLHaiyan Li
Haiyan Li
CGChangsheng Gao
Changsheng Gao
In this paper, a variable-structure multimodel (VSMM) filtering algorithm based on the long short-term memory (LSTM) regression-deep network (L-DQN) is proposed to accurately track strong maneuvering targets. The algorithm can map the selection of the model set to the selection of the action label and realize the purpose of a deep reinforcement-learning agent to replace the model switching in the traditional VSMM algorithm by reasonably designing a reward function, state space, and network structure. At the same time, the algorithm introduces a LSTM algorithm, which can compensate the error of tracking results based on model history information. The simulation results show that compared with the traditional VSMM algorithm, the proposed algorithm can quickly capture the maneuvering of the target, the response time is short, the calculation accuracy is significantly improved, and the range of adaptation is wider. Precise tracking of maneuvering targets was achieved.

16 October 2025 in Electronics
Verified
Journal Article
Open access
PDF file
YGYingbin Gao
Yingbin Gao
HDHaidi Dong
Haidi Dong
ZXZhongying Xu
Zhongying Xu
HLHaiyan Li
Haiyan Li
JLJing Li
Jing Li
SYShenzhi Yuan
Shenzhi Yuan
An MC (minor component) is usually referred to as the noise part of a time-varying signal. Extracting multiple MCs from an input signal is very useful in many practical applications. Compared with single MC estimating algorithms and subspace tracking algorithms, multiple MC extraction algorithms have a wider range of applications and greater research significance. To address the existing issues with multiple MC extraction algorithms, such as the need to estimate parameters in advance and with numerous constraints, this paper proposes a novel multiple MC extraction algorithm by adding a diagonal weighted matrix to the Möller algorithm. The proposed algorithm’s fixed points are analyzed using ODE (ordinary differential equation) methods, and it is demonstrated that the algorithm achieves stable convergence only when the weight matrix converges to the desired MCs of the signal. The simulation results illustrate the effectiveness of the proposed algorithm.

18 February 2011 in Journal of Software
Verified
Journal Article
Closed access
PDF file
GHGuang Hu
Guang Hu
SYShengzhi Yuan
Shengzhi Yuan
Infrared imaging guidance is a research hotspot in accurate terminal guidance field now. An idea on automatic extraction of region (ROI) of interest in infrared serial images is proposed in order to treat intelligently infrared serial images which are captured. The target detection algorithm of target is researched and the applying scene of the target detection algorithm is confirmed based on analysis for the features of infrared serial images. According to differences of detection algorithm for infrared serial images between static and dynamic scenes, the corresponding detection and extraction algorithm of ROI is discussed respectively. The algorithms are all simulated in real scene. A new stepwise approaching and recurring threshold search algorithm based on two-dimensional maximum entropy principle was proposed by studying recurring formulation optimized of two-dimensional maximum entropy in order to realize to detect target and extract ROI of serial images under complex background. The algorithm above realizes automatic extraction of ROI in Infrared serial images. The results improve the efficiency of accurate terminal guidance and they have good application value by practicing.

1 January 2016 in International Journal on Smart Sensing and Intelligent Systems
Verified
Journal Article
Open access
GHGuang Hu
Guang Hu
SYShengzhi Yuan
Shengzhi Yuan
Local Segmentation is the fundamental task for image processing. Consider to the problem of low segmentation precision and contour control instability for image local segmentation, a local segmentation theory is researched that based on SSCACM (spectral clustering with spatial coherence property jointing active contour model). First, by applying spatial coherence property constraint of image pixels to spectral clustering, an adaptive similarity function is constructed and the corresponding spectral clustering algorithm is used to extract initial contour of the local region of an image. Then, the NBACM (narrow band active contour model) is combined with the priori information of initial contour to evolve contour curve to get the segmentation result. At last, the local segmentation experiment is realized on synthetic images and medical images. The experimental results show that the method proposed can extract contour accurately and can improve the effectiveness and robust for image local segmentation.

17 November 2023
Verified
Conference Paper
Open access
SSSen Shen
Sen Shen
RXRenkai Xiao
Renkai Xiao
MLMin Li
Min Li
SYShengzhi Yuan
Shengzhi Yuan
In order to effectively solve the problem that the network structure of the existing target recognition algorithm is too complicated and the calculation amount is too large, it is difficult to deploy and carry out target detection and other tasks in real time on the edge computing platform with limited resources. In order to effectively improve the real-time online target detection capability of UAV remote sensing images, this paper carried out lightweight research based on YOLOv5s model by replacing the backbone network with MobileNet V3 method. Through a series of training and testing, the training weight is reduced by 67.7%, the detection effect parameters mAP and FPS are slightly decreased, meeting the expected requirements, and the front-end deployment of the target detection task can be better realized. The test on the edge computing platform in the picture has also achieved good results, verifying the feasibility of the optimization scheme.