Sheng Xu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Sheng Xu has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Xiaojia Wang
School of Management Hefei University of Technology Hefei China
3 shared publications - Changyan Shao
Department of Information Management, School of Management, Hefei University of Technology, Hefei, China
1 shared publication - Chong Di
School of Electrical Engineering and Automation, Hefei University of Technology, Hefei, China
1 shared publication Jingjing Niu
School of Management, Hefei University of Technology, Hefei, Anhui, 230009, China
1 shared publication- Jinlong Fang
Hefei University of Technology, 12513 Hefei, Anhui China
1 shared publication - Liang Zhizhen
Department of Information Management, School of Management, Hefei University of Technology, Hefei, China
1 shared publication
Latest Publications
- Open Access
- 6 March 2025
- Published by Springer Nature in International Journal of Computational Intelligence Systems
- Vol. 18 (1) , 1-24
Application of Distributionally Robust Optimization Markov Decision-Making Under Uncertainty in Scheduling of Multi-category Emergency Medical Materials
- 6 March 2025
- Published by Springer Nature in International Journal of Computational Intelligence Systems
- Vol. 18 (1) , 1-24
Open AccessApplication of Distributionally Robust Optimization Markov Decision-Making Under Uncertainty in Scheduling of Multi-category Emergency Medical Materials
- Open Access
A Novel Rolling Bearing Fault Diagnosis Method Based on BLS and CNN with Attention Mechanism
In actual industrial application scenarios, noise pollution makes it difficult to extract fault features accurately via conventional methods. A novel method for rolling bearing fault diagnosis combining a convolutional neural network (CNN), an attention mechanism for a squeeze-and-excitation network (SENet) module and a broad learning system (BLS) is proposed (SECNN–BLS). The one-dimensional bearing vibration signal is processed by using multiple short-time Fourier transforms (STFT); the two-dimensional image in the time-frequency domain is used as the model input. In the CNN fault feature extraction process, SENet is introduced to replace the CNN convolution layer, and global information is obtained through the squeeze operation. Excitation operation captures the importance of image channels, assigns weights adaptively to improve the attention on important information and eliminates the interference of irrelevant features without increasing the spatial and temporal complexity. The weighted feature representation is then transferred to the BLS input. The BLS has the characteristics of a flat structure and ridge regression to quickly solve the weights; as a fault classifier, it can save more computing resources and improve the accuracy of the fault classification. The SECNN–BLS fault diagnosis has achieved more than 98% accuracy on the Society for Machinery Failure Prevention Technology (MFPT) dataset. We also demonstrate the excellent performance of SECNN–BLS in a noisy environment.
Open AccessA Novel Rolling Bearing Fault Diagnosis Method Based on BLS and CNN with Attention Mechanism
In actual industrial application scenarios, noise pollution makes it difficult to extract fault features accurately via conventional methods. A novel method for rolling bearing fault diagnosis combining a convolutional neural network (CNN), an attention mechanism for a squeeze-and-excitation network (SENet) module and a broad learning system (BLS) is proposed (SECNN–BLS). The one-dimensional bearing vibration signal is processed by using multiple short-time Fourier transforms (STFT); the two-dimensional image in the time-frequency domain is used as the model input. In the CNN fault feature extraction process, SENet is introduced to replace the CNN convolution layer, and global information is obtained through the squeeze operation. Excitation operation captures the importance of image channels, assigns weights adaptively to improve the attention on important information and eliminates the interference of irrelevant features without increasing the spatial and temporal complexity. The weighted feature representation is then transferred to the BLS input. The BLS has the characteristics of a flat structure and ridge regression to quickly solve the weights; as a fault classifier, it can save more computing resources and improve the accuracy of the fault classification. The SECNN–BLS fault diagnosis has achieved more than 98% accuracy on the Society for Machinery Failure Prevention Technology (MFPT) dataset. We also demonstrate the excellent performance of SECNN–BLS in a noisy environment.
- Open Access
- 30 August 2022
- Published by Springer Nature in International Journal of Computational Intelligence Systems
- Vol. 15 (1) , 1-20
Analysis and Prediction of Gestational Diabetes Mellitus by the Ensemble Learning Method
Gestational diabetes mellitus (GDM) is the most common disease in pregnancy and can cause a series of maternal and infant complications. A new study shows that GDM affects one in six deliveries. Identifying and screening for risk factors for GDM can effectively help intervene and improve the condition of women and their children. Therefore, the aim of this paper is to determine the risk factors for GDM and to use the ensemble learning method to judge whether pregnant women suffer from GDM more accurately. First, this study involves six commonly used machine learning algorithms to analyze the GDM data from the Tianchi competition, selects the risk factors according to the ranking of each model, and uses the Shapley additive interpreter method to determine the importance of the selected risk factors. Second, the combined weighting method was used to analyze and evaluate the risk factors for gestational diabetes and to determine a group of important factors. Lastly, a new integrated light gradient-boosting machine-extreme gradient boosting-gradient boosting tree (LightGBM-Xgboost-GB) learning method is proposed to determine whether pregnant women have gestational diabetes mellitus. We used the gray correlation degree to calculate the weight and used a genetic algorithm for optimization. In terms of prediction accuracy and comprehensive effects, the final model is better than the commonly used machine learning model. The ensemble learning model is comprehensive and flexible and can be used to determine whether pregnant women suffer from GDM. In addition to disease prediction, the model can also be extended for use to many other areas of research.
- 30 August 2022
- Published by Springer Nature in International Journal of Computational Intelligence Systems
- Vol. 15 (1) , 1-20
Open AccessAnalysis and Prediction of Gestational Diabetes Mellitus by the Ensemble Learning Method
Gestational diabetes mellitus (GDM) is the most common disease in pregnancy and can cause a series of maternal and infant complications. A new study shows that GDM affects one in six deliveries. Identifying and screening for risk factors for GDM can effectively help intervene and improve the condition of women and their children. Therefore, the aim of this paper is to determine the risk factors for GDM and to use the ensemble learning method to judge whether pregnant women suffer from GDM more accurately. First, this study involves six commonly used machine learning algorithms to analyze the GDM data from the Tianchi competition, selects the risk factors according to the ranking of each model, and uses the Shapley additive interpreter method to determine the importance of the selected risk factors. Second, the combined weighting method was used to analyze and evaluate the risk factors for gestational diabetes and to determine a group of important factors. Lastly, a new integrated light gradient-boosting machine-extreme gradient boosting-gradient boosting tree (LightGBM-Xgboost-GB) learning method is proposed to determine whether pregnant women have gestational diabetes mellitus. We used the gray correlation degree to calculate the weight and used a genetic algorithm for optimization. In terms of prediction accuracy and comprehensive effects, the final model is better than the commonly used machine learning model. The ensemble learning model is comprehensive and flexible and can be used to determine whether pregnant women suffer from GDM. In addition to disease prediction, the model can also be extended for use to many other areas of research.
- Open Access
- 20 January 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 23069-23081
Study on the Location of Private Clinics Based on K-Means Clustering Method and an Integrated Evaluation Model
Appropriate location is an important prerequisite for the long-term survival and development of private medical institutions. However, in both theory and practice, the issue of location decision-making for private clinics has not been fully studied. We therefore aimed to provide a feasible scheme for the location of new private clinics. This paper combines the k-means clustering method and an integrated 2DULVs (two-dimensional uncertain language variables)-TOPSIS (technique for order preference by similarity to ideal solution)-DSCCR (Dempster-Shafer conjunctive combination rule) model to screen and evaluate all of the areas in the target region for an Internet medical company to set up offline clinics. We first created geographic grids using GIS and collected point of interest (POI) data. We then used the k-means clustering method to obtain 10 suitable grids as alternatives. Last, we established an evaluation index system and used the proposed model to rank them. The results show that grids 178, 179 and 202 are more suitable for the company to establish offline clinics in the expansion of business. The results of this study are also consistent with those of the other three fusion methods. This paper provides a beneficial attempt for private clinics to make location decisions and can be extended to the strategic decision-making of other industries or other issues.
- 20 January 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 23069-23081
Open AccessStudy on the Location of Private Clinics Based on K-Means Clustering Method and an Integrated Evaluation Model
Appropriate location is an important prerequisite for the long-term survival and development of private medical institutions. However, in both theory and practice, the issue of location decision-making for private clinics has not been fully studied. We therefore aimed to provide a feasible scheme for the location of new private clinics. This paper combines the k-means clustering method and an integrated 2DULVs (two-dimensional uncertain language variables)-TOPSIS (technique for order preference by similarity to ideal solution)-DSCCR (Dempster-Shafer conjunctive combination rule) model to screen and evaluate all of the areas in the target region for an Internet medical company to set up offline clinics. We first created geographic grids using GIS and collected point of interest (POI) data. We then used the k-means clustering method to obtain 10 suitable grids as alternatives. Last, we established an evaluation index system and used the proposed model to rank them. The results show that grids 178, 179 and 202 are more suitable for the company to establish offline clinics in the expansion of business. The results of this study are also consistent with those of the other three fusion methods. This paper provides a beneficial attempt for private clinics to make location decisions and can be extended to the strategic decision-making of other industries or other issues.
- Open Access
- 1 January 2020
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 1437 (1) , 012115
Optimize Logistics cost model for shared logistics platform based on time-driven activity-based costing
Transportation cost management plays an important role in both traditional logistics management and supply chain logistics management. Time-driven activity-based costing bypasses the expensive, time-consuming and subjective activity investigation and is more suitable for the intermediate link of enterprise cost accounting. Linear programming is a mathematical theory and method to study the extreme value problem of objective function under linear constraints, and the optimal decision is made by using limited human, material and financial resources. Based on the concept of Shared logistics, this paper studies transport enterprises to solve problems such as information asymmetry in the process of transport. By establishing a linear programming time-driven activity-based costing accounting model, the allocation of scarce resources in transport enterprises can be reasonably solved, and the efficiency of cost accounting can be improved to achieve considerable cost savings.
- 1 January 2020
- Published by IOP Publishing in Journal of Physics: Conference Series
- Vol. 1437 (1) , 012115
Open AccessOptimize Logistics cost model for shared logistics platform based on time-driven activity-based costing
Transportation cost management plays an important role in both traditional logistics management and supply chain logistics management. Time-driven activity-based costing bypasses the expensive, time-consuming and subjective activity investigation and is more suitable for the intermediate link of enterprise cost accounting. Linear programming is a mathematical theory and method to study the extreme value problem of objective function under linear constraints, and the optimal decision is made by using limited human, material and financial resources. Based on the concept of Shared logistics, this paper studies transport enterprises to solve problems such as information asymmetry in the process of transport. By establishing a linear programming time-driven activity-based costing accounting model, the allocation of scarce resources in transport enterprises can be reasonably solved, and the efficiency of cost accounting can be improved to achieve considerable cost savings.
- 18 June 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Energy Conversion
- Vol. 33 (4) , 2174-2182
A Novel Computational Method of Skewing Leakage Reactance for a Doubly Skewed Rotor Induction Motor
In this paper, a novel computational method which can accurately calculate the skewing leakage reactance of doubly skewed rotor induction motor is proposed. It is based on the three-dimensional (3D) finite-element (FE) models to calculate the magnetic energy stored in every section of the machine and the rotor bar currents. In addition, the magnetic saturation and magnetic variation in the axial direction are taken into account in the models. Also, the proposed method is applied to calculate the stator leakage reactances and the rotor leakage reactances. The comparisons between the testing results and the computational results show that the proposed method can obtain the more accurate results than the conventional computational algorithm.
- 18 June 2018
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Energy Conversion
- Vol. 33 (4) , 2174-2182
A Novel Computational Method of Skewing Leakage Reactance for a Doubly Skewed Rotor Induction Motor
In this paper, a novel computational method which can accurately calculate the skewing leakage reactance of doubly skewed rotor induction motor is proposed. It is based on the three-dimensional (3D) finite-element (FE) models to calculate the magnetic energy stored in every section of the machine and the rotor bar currents. In addition, the magnetic saturation and magnetic variation in the axial direction are taken into account in the models. Also, the proposed method is applied to calculate the stator leakage reactances and the rotor leakage reactances. The comparisons between the testing results and the computational results show that the proposed method can obtain the more accurate results than the conventional computational algorithm.