Junjie Liu

Shanghai University of Sport

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

Huan Wang

Tsinghua University

Zhiliang Liu

Southwest Jiaotong University

Zilin Wang

University of California, Berkeley

Hangbiao Li

Southwest China Institute of Electronic Technology

Feng Zhou

Toyota Research Institute

Jicheng Yu

China Electric Power Research Institute

Bo Zhang

University of Electronic Science and Technology of China

Siyuan Liang

China Electric Power Research Institute

Nan Zhang

Hebei Medical University

Bin Cheng

Finnish Meteorological Institute

Cooperative Institutions

University of Electronic Science and Technology of China

National Engineering Research Center of Electromagnetic Radiation Control Materials

State Key Laboratory of Electronic Thin Films and Integrated Devices

China Electric Power Research Institute

Chongqing University

North China Electric Power University

Southwest Jiaotong University

Tsinghua University

Chinese Academy of Sciences

University of Chinese Academy of Sciences

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

Review of pattern recognition based on support vector machine

Ship Science and Technology (2006)

Nan Zhang Junjie Liu

The principle and applications of the support vector machine(SVM) are briefly reviewed.Firstly,the background and the development prospects of SVM are described and then the definition,classification and algorithms of SVM are briefly introduced.Finally,we discussed the SVM-based applications,such as in text identification,handwritten numeral recognition,face detection,recognition and verification,regression analysis and prediction,image retrieval,feature selection and so on.

Numeral system

Identification

Source

Cite

Citations (1)

Rolling Bearing Fault Diagnosis Using CNN-based Attention Modules and Gated Recurrent Unit

Zilin Wang Huan Wang Zhiliang Liu Junjie Liu

Rolling bearings are of great importance to rotating machinery. However, in real operating conditions, rolling bearings are damaged chronically by complex factors like nonuniform workload, which leads to occurrences of faults. Thus, there is necessity to recognize the bearing faults in advance. Algorithms based on deep learning (DL) have excellent feature auto-learning ability. However, many DL methods pay few attentions on channel-wise information, with less satisfactory in the results. Therefore, this paper proposed an end-to-end model named CAT-GRU. This method is composed of convolutional neural networks (CNN), channel attention mechanism (CAM) and gated recurrent units (GRU) to early recognize the bearing faults. In this structure, the CNN encodes the abstract features by multi convolution operations from the raw one-dimensional vibration data. The CAM then obtains the channel-wise fault significances from the output of CNN and weights each channel to highlight more fault-relevant features. Next, the following GRU captures time-related features before the final fault type classification. On motor bearing dataset from Case Western Reserve University, the CAT-GRU shows excellent fault diagnosis performance.

Convolution (computer science)

Feature (linguistics)

10.1109/phm-shanghai49105.2020.9280991

Cite

Citations (2)

基于CNN的工件缺陷检测方法研究 (Research of Workpiece Defect Detection Method Based on CNN).

计算机科学 (2017)

Qiao Li Erdun Zhao Junjie Liu Bin Cheng

Source

Cite

Citations (0)

A compact ADPLL based on symmetrical binary frequency searching with the same circuit

International Journal of Electronics (2014)

Hangbiao Li Bo Zhang Ping Luo Pengfei Liao Junjie Liu

A compact all-digital phase-locked loop (C-ADPLL) based on symmetrical binary frequency searching (BFS) with the same circuit is presented in this paper. The minimising relative frequency variation error Δη (MFE) rule is derived as guidance of design and is used to weigh the accuracy of the digitally controlled oscillator (DCO) clock frequency. The symmetrical BFS is used in the coarse-tuning process and the fine-tuning process of DCO clock frequency to achieve the minimum Δη of the locked DCO clock, which simplifies the circuit architecture and saves the die area. The C-ADPLL is implemented in a 0.13 μm one-poly-eight-metal (1P8M) CMOS process and the on-chip area is only 0.043 mm2, which is much smaller. The measurement results show that the peak-to-peak (Pk-Pk) jitter and the root-mean-square jitter of the DCO clock frequency are 270 ps at 72.3 MHz and 42 ps at 79.4 MHz, respectively, while the power consumption of the proposed ADPLL is only 2.7 mW (at 115.8 MHz) with a 1.2 V power supply. The measured Δη is not more than 1.14%. Compared with other ADPLLs, the proposed C-ADPLL has simpler architecture, smaller size and lower Pk-Pk jitter.

Digitally controlled oscillator

Clock rate

Process Variation

10.1080/00207217.2014.897383

Cite

Citations (3)

Multitask Learning Based on Lightweight 1DCNN for Fault Diagnosis of Wheelset Bearings

IEEE Transactions on Instrumentation and Measurement (2020)

Zhiliang Liu Huan Wang Junjie Liu Yong Qin Dandan Peng

In recent years, deep learning has been proved to be a promising bearing fault diagnosis technology. However, most of the existing methods are based on single-task learning. Fault diagnosis task (FDT) is treated as an independent task, and rich correlation information contained in different tasks is ignored. Therefore, this article explores the possibility of using speed identification task (SIT) and load identification task (LIT) as two auxiliary tasks to improve the performance of the FDT and proposes a multitask one-dimensional convolutional neural network (MT-1DCNN). Specifically, the MT-1DCNN utilizes trunk network to learn shared features required for every task and then processes different tasks through multiple task-specific branches. In this way, the MT-1DCNN can utilize features learned by related tasks to improve the performance of the FDT. The experimental results with wheelset bearing data set show that the multitask learning can make full use of the feature information captured by the SIT and the LIT to improve the fault diagnosis performance of the network, and the MT-1DCNN has a better performance than five excellent networks in accuracy.

Identification

Feature (linguistics)

10.1109/tim.2020.3017900

Cite

Citations (105)

Application of Deep Learning Methods in the Intelligent Analysis of Sprint Training Postures

Junjie Liu

Sprint

10.1109/icmiii62623.2024.00078

Cite

Citations (0)

CNN-based Digital Twin Model for Ultra-High Voltage Direct Current System Loss Measurement

IEEE Access (2024)

Xiaodong Yin Siyuan Liang Jicheng Yu Feng Zhou Junjie Liu

The ultra-high voltage direct current (UHVDC) system is widely constructed due to its suitability for large-capacity long-distance transmission, and its losses have become an important part of the power grid losses. However, the overall and subcomponent losses of the power system are difficult to measure by measuring devices or accurately calculated by existing algorithms. In this paper, a convolutional neural network (CNN)-based digital twin model for UHVDC system loss measurement is proposed. The parameters and data of the physical system are fed into the digital space system through the data interaction. In the digital twin model of the digital space system, a CNN network with added topology information extraction layers and key equipment parameter adaptive perception layers excavates the deep multidimensional correlation features of the digital space mapping data. Then a multi-task joint processing layer fuses the extracted deep features and missing information to calculate the total and sub-component losses. The training and testing results based on actual UHVDC project data show that the mean absolute error (MAE) is 0.9, while R ² is 0.9999, which proves the accuracy of the proposed digital twin model is superior to mainstream deep learning models. This model is embedded in digital twin devices and applied in actual UHVDC converter stations.

10.1109/access.2024.3402386

Cite

Citations (0)

Fault Diagnosis Based on 1D-CNN with Associated Auxiliary Tasks Boosted for Wheelset Bearings of High-Speed Trains

2017 International Conference on Sensing, Diagnostics, Prognostics, and Control (SDPC) (2020)

Junjie Liu Huan Wang Zhiliang Liu Zilin Wang

Fault diagnosis for wheelset bearings of highspeed trains (HSTs) is very important for the reliability of the train operation. In recent years, deep learning technology has been widely used for fault diagnosis of mechanical components. Most deep learning methods are optimized from the data set or network structure. However, most target tasks themselves are complicated, and the factors leading to any of its results are not limited to the two aspects mentioned above. In general, there are one or more other factors related to the target task that influence the outcome of the final results. This paper introduces the idea of multi-task learning into the one-dimensional convolutional neural network (1D-CNN) model, and explores the possibility of enhancing the learning ability of the model by using factors related to the target task as auxiliary tasks. The experiment proves that the auxiliary tasks associated with the target task can indeed enhance the learning ability of the target task. And compared with other five cutting-edge fault diagnosis methods, the proposed model also has very good performance for wheelset bearings of HSTs.

Task Analysis

10.1109/sdpc49476.2020.9353116

Cite

Citations (1)