Tian Xiang

Tian Xiang

Auto-generated profile, based on publicly available information

Tian Xiang

Engineer School Qinghai Institute of Technology Xining China

Research Focus & Metrics

Academic Career Length

13Years (2014-Present)

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

Representative Publications

View All Publications
1 January 2023 in Computer Modeling in Engineering & Sciences
Verified
Journal Article
Open access
PDF file
YZYan Zeng
Yan Zeng
STSiyuan Teng
Siyuan Teng
TXTian Xiang
Tian Xiang
JZJilin Zhang
Jilin Zhang
YMYuankai Mu
Yuankai Mu
JWJian Wan
Jian Wan
Federated learning is a distributed machine learning method that can solve the increasingly serious problem of data islands and user data privacy, as it allows training data to be kept locally and not shared with other users. It trains a global model by aggregating locally-computed models of clients rather than their raw data. However, the divergence of local models caused by data heterogeneity of different clients may lead to slow convergence of the global model. For this problem, we focus on the client selection with federated learning, which can affect the convergence performance of the global model with the selected local models. We propose FedChoice, a client selection method based on loss function optimization, to select appropriate local models to improve the convergence of the global model. It firstly sets selected probability for clients with the value of loss function, and the client with high loss will be set higher selected probability, which can make them more likely to participate in training. Then, it introduces a local control vector and a global control vector to predict the local gradient direction and global gradient direction, respectively, and calculates the gradient correction vector to correct the gradient direction to reduce the cumulative deviation of the local gradient caused by the Non-IID data. We make experiments to verify the validity of FedChoice on CIFAR-10, CINIC-10, MNIST, EMNITS, and FEMNIST datasets, and the results show that the convergence of FedChoice is significantly improved, compared with FedAvg, FedProx, and FedNova.

25 March 2024 in IEEE Transactions on Instrumentation and Measurement
Verified
Journal Article
Closed access
Yaochun HouYaochun Hou
Yaochun Hou
YWYuxuan Wang
Yuxuan Wang
TXTian Xiang
Tian Xiang
JXJianghui Xie
Jianghui Xie
JZJuntao Zhao
Juntao Zhao
Dazhuan WuDazhuan Wu
Dazhuan Wu
Blind deconvolution has been demonstrated to be an effective tool for rolling element bearing localized defect detection. However, the successful utilization of deconvolution methods largely depends on the selection of several tunable configuration parameters, otherwise it may lead to the incomplete extraction of malfunction information. Additionally, the effectiveness of some multiple-transient-oriented deconvolution approaches strongly relies on the accurate set of the specific prior impulse period, whose application may be severely confined in practice. Coping with the aforementioned problems, an efficient as well as automatic bearing fault detection and diagnosis method based on Multipoint Optimal Minimum Entropy Deconvolution Adjusted (MOMEDA) and Bayesian optimization (BO) is proposed in this paper, where the configuration parameters and the potential impulse period can be optimized jointly in a sequential decision-making framework. To alleviate the influence of trend terms and intricate interferences in the raw signal to some extent and promote the deconvolution result, a preprocessing method coined selective coarse enveloping (SCE) is developed based upon the kurtosis indicator. In the methodological aspect, the signal after preprocessing is analyzed by the presented BO-based MOMEDA (BO-MOMEDA) method, and the period of the purified signal can be obtained, with the fault type of bearing being identified. Furthermore, both simulation analysis and experimental verifications profoundly corroborate the efficiency and superiority of the proposed method, compared with other contradistinctive methods.

20 July 2024 in Archive of Applied Mechanics
Verified
Journal Article
Closed access
PDF file
TXTian Xiang
Tian Xiang
WCWenhua Chen
Wenhua Chen


17 April 2014 in Advanced Materials Research
Verified
Journal Article
Closed access
PDF file
SXShuo Xu
Shuo Xu
TXTian Yu Xiang
Tian Yu Xiang
Steel and concrete composite slab is a new type of bridge deck, which is consist of steel plate and concrete slab connected by shear connectors such as perfobond ribs ( usually abbreviated as PBL), studs and so on. Static experiment about four pieces of one-way steel and concrete composite slab using PBL was carried, and the whole-process structural behaviors and failure mode were investigated. Based on the spatial beam element model with the degenerated theory, the nonlinear finite element analysis on experimental specimens was executed. Such structural behaviors as concrete cracking, yield of steel plate, crushing of concrete, and so on are simulated, and are compared with the experimental results. It can be observed that excellent agreements are achieved between experimental results and numerical simulation.

9 December 2024 in Fatigue & Fracture of Engineering Materials & Structures
Verified
Journal Article
Closed access
WCWenhua Chen
Wenhua Chen
TXTian Xiang
Tian Xiang