A Hybrid CNN–LSTM Algorithm for Online Defect Recognition of CO<sub>2</sub> Welding
At present, realizing high-quality automatic welding through online monitoring is a research focus in engineering applications. In this paper, a CNN⁻LSTM algorithm is proposed, which combines the advantages of convolutional neural networks (CNNs) and long short-term memory networks (LSTMs)...
Main Authors: | , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
MDPI AG
2018-12-01
|
Series: | Sensors |
Subjects: | |
Online Access: | https://www.mdpi.com/1424-8220/18/12/4369 |
id |
doaj-94c5413eb1d244e8be8353be5d051b02 |
---|---|
record_format |
Article |
spelling |
doaj-94c5413eb1d244e8be8353be5d051b022020-11-24T21:28:33ZengMDPI AGSensors1424-82202018-12-011812436910.3390/s18124369s18124369A Hybrid CNN–LSTM Algorithm for Online Defect Recognition of CO<sub>2</sub> WeldingTianyuan Liu0Jinsong Bao1Junliang Wang2Yiming Zhang3College of Mechanical Engineering, Dong Hua University, Shanghai 201620, ChinaCollege of Mechanical Engineering, Dong Hua University, Shanghai 201620, ChinaCollege of Mechanical Engineering, Dong Hua University, Shanghai 201620, ChinaCollege of Literature, Science and the Arts, The University of Michigan, Ann Arbor, MI 48109, USAAt present, realizing high-quality automatic welding through online monitoring is a research focus in engineering applications. In this paper, a CNN⁻LSTM algorithm is proposed, which combines the advantages of convolutional neural networks (CNNs) and long short-term memory networks (LSTMs). The CNN⁻LSTM algorithm establishes a shallow CNN to extract the primary features of the molten pool image. Then the feature tensor extracted by the CNN is transformed into the feature matrix. Finally, the rows of the feature matrix are fed into the LSTM network for feature fusion. This process realizes the implicit mapping from molten pool images to welding defects. The test results on the self-made molten pool image dataset show that CNN contributes to the overall feasibility of the CNN⁻LSTM algorithm and LSTM network is the most superior in the feature hybrid stage. The algorithm converges at 300 epochs and the accuracy of defects detection in CO<sub>2</sub> welding molten pool is 94%. The processing time of a single image is 0.067 ms, which fully meets the real-time monitoring requirement based on molten pool image. The experimental results on the MNIST and FashionMNIST datasets show that the algorithm is universal and can be used for similar image recognition and classification tasks.https://www.mdpi.com/1424-8220/18/12/4369deep learningCNNLSTMCO<sub>2</sub> weldingmolten poolonline monitoring |
collection |
DOAJ |
language |
English |
format |
Article |
sources |
DOAJ |
author |
Tianyuan Liu Jinsong Bao Junliang Wang Yiming Zhang |
spellingShingle |
Tianyuan Liu Jinsong Bao Junliang Wang Yiming Zhang A Hybrid CNN–LSTM Algorithm for Online Defect Recognition of CO<sub>2</sub> Welding Sensors deep learning CNN LSTM CO<sub>2</sub> welding molten pool online monitoring |
author_facet |
Tianyuan Liu Jinsong Bao Junliang Wang Yiming Zhang |
author_sort |
Tianyuan Liu |
title |
A Hybrid CNN–LSTM Algorithm for Online Defect Recognition of CO<sub>2</sub> Welding |
title_short |
A Hybrid CNN–LSTM Algorithm for Online Defect Recognition of CO<sub>2</sub> Welding |
title_full |
A Hybrid CNN–LSTM Algorithm for Online Defect Recognition of CO<sub>2</sub> Welding |
title_fullStr |
A Hybrid CNN–LSTM Algorithm for Online Defect Recognition of CO<sub>2</sub> Welding |
title_full_unstemmed |
A Hybrid CNN–LSTM Algorithm for Online Defect Recognition of CO<sub>2</sub> Welding |
title_sort |
hybrid cnn–lstm algorithm for online defect recognition of co<sub>2</sub> welding |
publisher |
MDPI AG |
series |
Sensors |
issn |
1424-8220 |
publishDate |
2018-12-01 |
description |
At present, realizing high-quality automatic welding through online monitoring is a research focus in engineering applications. In this paper, a CNN⁻LSTM algorithm is proposed, which combines the advantages of convolutional neural networks (CNNs) and long short-term memory networks (LSTMs). The CNN⁻LSTM algorithm establishes a shallow CNN to extract the primary features of the molten pool image. Then the feature tensor extracted by the CNN is transformed into the feature matrix. Finally, the rows of the feature matrix are fed into the LSTM network for feature fusion. This process realizes the implicit mapping from molten pool images to welding defects. The test results on the self-made molten pool image dataset show that CNN contributes to the overall feasibility of the CNN⁻LSTM algorithm and LSTM network is the most superior in the feature hybrid stage. The algorithm converges at 300 epochs and the accuracy of defects detection in CO<sub>2</sub> welding molten pool is 94%. The processing time of a single image is 0.067 ms, which fully meets the real-time monitoring requirement based on molten pool image. The experimental results on the MNIST and FashionMNIST datasets show that the algorithm is universal and can be used for similar image recognition and classification tasks. |
topic |
deep learning CNN LSTM CO<sub>2</sub> welding molten pool online monitoring |
url |
https://www.mdpi.com/1424-8220/18/12/4369 |
work_keys_str_mv |
AT tianyuanliu ahybridcnnlstmalgorithmforonlinedefectrecognitionofcosub2subwelding AT jinsongbao ahybridcnnlstmalgorithmforonlinedefectrecognitionofcosub2subwelding AT junliangwang ahybridcnnlstmalgorithmforonlinedefectrecognitionofcosub2subwelding AT yimingzhang ahybridcnnlstmalgorithmforonlinedefectrecognitionofcosub2subwelding AT tianyuanliu hybridcnnlstmalgorithmforonlinedefectrecognitionofcosub2subwelding AT jinsongbao hybridcnnlstmalgorithmforonlinedefectrecognitionofcosub2subwelding AT junliangwang hybridcnnlstmalgorithmforonlinedefectrecognitionofcosub2subwelding AT yimingzhang hybridcnnlstmalgorithmforonlinedefectrecognitionofcosub2subwelding |
_version_ |
1725969830927925248 |