Huawei Zhu
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Research Interests
Huawei Zhu has a total of 10 co-authors. Here are the top co-authors by number of shared publications:
- Yanli Shao
School of Computer Science and Technology, Hangzhou Dianzi University, China
4 shared publications - Yiming Zhao
Key Laboratory of Complex Systems Modeling and Simulation, School of Computer Science and Technology, Hangzhou Dianzi University, Hangzhou, 310018, China
2 shared publications - Yusheng Liu
State Key Laboratory of CAD&CG, Zhejiang University, Hangzhou 310027, China
2 shared publications - Chunguang Li
College of Information Science and Electronic Engineering, Zhejiang University, Hangzhou, China
1 shared publication - Feng Yu
Key Laboratory of Complex Systems Modeling and Simulation, School of Computer Science and Technology, Hangzhou Dianzi University, Hangzhou 310018, China
1 shared publication - Jinglong Fang
Hangzhou Dianzi University
1 shared publication
Latest Publications
- Open Access
- 10 February 2022
- Published by IOS Press
- p. 598-608
Short-Term Traffic Flow Prediction Method Based on Spatio-Temporal Characteristics of Complex Road Network
- 10 February 2022
- Published by IOS Press
- p. 598-608
Open AccessShort-Term Traffic Flow Prediction Method Based on Spatio-Temporal Characteristics of Complex Road Network
- Open Access
- 31 May 2021
- Published by Wiley in Mobile Information Systems
- Vol. 2021, 1-16
The Traffic Flow Prediction Method Using the Incremental Learning-Based CNN-LTSM Model: The Solution of Mobile Application
With the acceleration of urbanization and the increase in the number of motor vehicles, more and more social problems such as traffic congestion have emerged. Accordingly, efficient and accurate traffic flow prediction has become a research hot spot in the field of intelligent transportation. However, traditional machine learning algorithms cannot further optimize the model with the increase of the data scale, and the deep learning algorithms perform poorly in mobile application or real-time application; how to train and update deep learning models efficiently and accurately is still an urgent problem since they require huge computation resources and time costs. Therefore, an incremental learning-based CNN-LTSM model, IL-TFNet, is proposed for traffic flow prediction in this study. The lightweight convolution neural network-based model architecture is designed to process spatiotemporal and external environment features simultaneously to improve the prediction performance and prediction efficiency of the model. Especially, the K-means clustering algorithm is applied as an uncertainty feature to extract unknown traffic accident information. During the model training, instead of the traditional batch learning algorithm, the incremental learning algorithm is applied to reduce the cost of updating the model and satisfy the requirements of high real-time performance and low computational overhead in short-term traffic prediction. Furthermore, the idea of combining incremental learning with active learning is proposed to fine-tune the prediction model to improve prediction accuracy in special situations. Experiments have proved that compared with other traffic flow prediction models, the IL-TFNet model performs well in short-term traffic flow prediction.
- 31 May 2021
- Published by Wiley in Mobile Information Systems
- Vol. 2021, 1-16
Open AccessThe Traffic Flow Prediction Method Using the Incremental Learning-Based CNN-LTSM Model: The Solution of Mobile Application
With the acceleration of urbanization and the increase in the number of motor vehicles, more and more social problems such as traffic congestion have emerged. Accordingly, efficient and accurate traffic flow prediction has become a research hot spot in the field of intelligent transportation. However, traditional machine learning algorithms cannot further optimize the model with the increase of the data scale, and the deep learning algorithms perform poorly in mobile application or real-time application; how to train and update deep learning models efficiently and accurately is still an urgent problem since they require huge computation resources and time costs. Therefore, an incremental learning-based CNN-LTSM model, IL-TFNet, is proposed for traffic flow prediction in this study. The lightweight convolution neural network-based model architecture is designed to process spatiotemporal and external environment features simultaneously to improve the prediction performance and prediction efficiency of the model. Especially, the K-means clustering algorithm is applied as an uncertainty feature to extract unknown traffic accident information. During the model training, instead of the traditional batch learning algorithm, the incremental learning algorithm is applied to reduce the cost of updating the model and satisfy the requirements of high real-time performance and low computational overhead in short-term traffic prediction. Furthermore, the idea of combining incremental learning with active learning is proposed to fine-tune the prediction model to improve prediction accuracy in special situations. Experiments have proved that compared with other traffic flow prediction models, the IL-TFNet model performs well in short-term traffic flow prediction.
- 3 January 2020
- Published by Wiley in Journal of Computing and Information Science in Engineering
- Vol. 20 (2) , 1-26
A Simulation Data-Driven Design Approach for Rapid Product Optimization
Traditional design optimization is an iterative process of design, simulation, and redesign, which requires extensive calculations and analysis. The designer needs to adjust and evaluate the design parameters manually and continually based on the simulation results until a satisfactory design is obtained. However, the expensive computational costs and large resource consumption of complex products hinder the wide application of simulation in industry. It is not an easy task to search the optimal design solution intelligently and efficiently. Therefore, a simulation data-driven design approach which combines dynamic simulation data mining and design optimization is proposed to achieve this purpose in this study. The dynamic simulation data mining algorithm—on-line sequential extreme learning machine with adaptive weights (WadaptiveOS-ELM)—is adopted to train the dynamic prediction model to effectively evaluate the merits of new design solutions in the optimization process. Meanwhile, the prediction model is updated incrementally by combining new “good” data set to reduce the modeling cost and improve the prediction accuracy. Furthermore, the improved heuristic optimization algorithm—adaptive and weighted center particle swarm optimization (AWCPSO)—is introduced to guide the design change direction intelligently to improve the search efficiency. In this way, the optimal design solution can be searched automatically with less actual simulation iterations and higher optimization efficiency, and thus supporting the rapid product optimization effectively. The experimental results demonstrate the feasibility and effectiveness of the proposed approach.
- 3 January 2020
- Published by Wiley in Journal of Computing and Information Science in Engineering
- Vol. 20 (2) , 1-26
A Simulation Data-Driven Design Approach for Rapid Product Optimization
Traditional design optimization is an iterative process of design, simulation, and redesign, which requires extensive calculations and analysis. The designer needs to adjust and evaluate the design parameters manually and continually based on the simulation results until a satisfactory design is obtained. However, the expensive computational costs and large resource consumption of complex products hinder the wide application of simulation in industry. It is not an easy task to search the optimal design solution intelligently and efficiently. Therefore, a simulation data-driven design approach which combines dynamic simulation data mining and design optimization is proposed to achieve this purpose in this study. The dynamic simulation data mining algorithm—on-line sequential extreme learning machine with adaptive weights (WadaptiveOS-ELM)—is adopted to train the dynamic prediction model to effectively evaluate the merits of new design solutions in the optimization process. Meanwhile, the prediction model is updated incrementally by combining new “good” data set to reduce the modeling cost and improve the prediction accuracy. Furthermore, the improved heuristic optimization algorithm—adaptive and weighted center particle swarm optimization (AWCPSO)—is introduced to guide the design change direction intelligently to improve the search efficiency. In this way, the optimal design solution can be searched automatically with less actual simulation iterations and higher optimization efficiency, and thus supporting the rapid product optimization effectively. The experimental results demonstrate the feasibility and effectiveness of the proposed approach.
- 20 July 2015
- Published by Taylor & Francis in International Journal of Computer Integrated Manufacturing
- Vol. 29 (8) , 821-838
Mid-surface abstraction for complex thin-wall models based on virtual decomposition
Mid-surface abstraction is an effective simplification method for thin-wall models. A mid-surface model after abstraction can reduce the complexity of finite element analysis greatly. There are some methods for mid-surface abstraction. However, they are not satisfactory to handle complex models. To address this problem, this study proposes a mid-surface abstraction approach based on virtual model decomposition. The concept of face group pair is proposed to help virtually decompose thin-wall models into several virtual simple parts, and the connections between the parts are classified into three types. The topological boundary edges of the mid-surface model are then generated correctly with the help of the connection types. Meanwhile, an improved mesh-based discrete method is proposed to calculate the geometry of the mid-surface with high accuracy. Finally, some typical complex models are given to illustrate the effectiveness of the proposed method.
- 20 July 2015
- Published by Taylor & Francis in International Journal of Computer Integrated Manufacturing
- Vol. 29 (8) , 821-838
Mid-surface abstraction for complex thin-wall models based on virtual decomposition
Mid-surface abstraction is an effective simplification method for thin-wall models. A mid-surface model after abstraction can reduce the complexity of finite element analysis greatly. There are some methods for mid-surface abstraction. However, they are not satisfactory to handle complex models. To address this problem, this study proposes a mid-surface abstraction approach based on virtual model decomposition. The concept of face group pair is proposed to help virtually decompose thin-wall models into several virtual simple parts, and the connections between the parts are classified into three types. The topological boundary edges of the mid-surface model are then generated correctly with the help of the connection types. Meanwhile, an improved mesh-based discrete method is proposed to calculate the geometry of the mid-surface with high accuracy. Finally, some typical complex models are given to illustrate the effectiveness of the proposed method.