Class-Imbalance Adversarial Transfer Learning Network for Cross-Domain Fault Diagnosis With Imbalanced Data

Kuang, Jiachen; Xu, Guizhi; Tao, Tangfei; Wu, Qingqiang

doi:10.1109/tim.2021.3136175

Cited by 66 publications

(27 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As it is very difficult to obtain image datasets with large sample sizes in the mechanical field [21,22] , for datasets with small sample sizes, directly using CNNs without initial weights to train small datasets will result in overfitting. However, transfer learning can first train a network on a large dataset, such as ImageNet [23] .…”

Section: Transfer Learning Modelmentioning

confidence: 99%

Tool wear assessment and life prediction model based on image processing and deep learning

Wang

2023

Int J Adv Manuf Technol

View full text Add to dashboard Cite

In the machinery field, drilling is one of the most important machining methods. Realtime monitoring of drill wear can effectively prevent the part quality from not meeting the specifications due to drill failure. This paper proposes a tool wear assessment and life prediction model based on image processing and deep learning methods, which shows great performance for small sample datasets and for low quality images. We construct an image database of drill bits and extract the normal areas and worn areas of the drill bits using the U-Net network and traditional image processing methods, respectively. Moreover, the original dataset is classified using the migration learning technique. The wear level of a drill bit can be accurately evaluated through experimental tests. Testing results show that the proposed method is more convenient and efficient than the previous methods that used manual measurements; the results can be applied to real-time drill wear monitoring, thereby reducing part damage caused by tool wear.

show abstract

Section: Transfer Learning Modelmentioning

confidence: 99%

Tool wear assessment and life prediction model based on image processing and deep learning

Wang

2023

Int J Adv Manuf Technol

View full text Add to dashboard Cite

show abstract

“…As the representative of the adversarial model, the domain adaptation model expects that the domain invariant and class separable features can be extracted after training. This method has been widely studied in cross-domain fault diagnosis [ 29 , 30 , 31 ].…”

Section: Related Workmentioning

confidence: 99%

Cross-Domain Open Set Fault Diagnosis Based on Weighted Domain Adaptation with Double Classifiers

Wang

Tong

et al. 2023

Sensors

View full text Add to dashboard Cite

The application of transfer learning in fault diagnosis has been developed in recent years. It can use existing data to solve the problem of fault recognition under different working conditions. Due to the complexity of the equipment and the openness of the working environment in industrial production, the status of the equipment is changeable, and the collected signals can have new fault classes. Therefore, the open set recognition ability of the transfer learning method is an urgent research direction. The existing transfer learning model can have a severe negative transfer problem when solving the open set problem, resulting in the aliasing of samples in the feature space and the inability to separate the unknown classes. To solve this problem, we propose a Weighted Domain Adaptation with Double Classifiers (WDADC) method. Specifically, WDADC designs the weighting module based on Jensen–Shannon divergence, which can evaluate the similarity between each sample in the target domain and each class in the source domain. Based on this similarity, a weighted loss is constructed to promote the positive transfer between shared classes in the two domains to realize the recognition of shared classes and the separation of unknown classes. In addition, the structure of double classifiers in WDADC can mitigate the overfitting of the model by maximizing the discrepancy, which helps extract the domain-invariant and class-separable features of the samples when the discrepancy between the two domains is large. The model’s performance is verified in several fault datasets of rotating machinery. The results show that the method is effective in open set fault diagnosis and superior to the common domain adaptation methods.

show abstract

“…[7][8][9] Recently, transfer learning has been applied to product quality prediction, operating performance assessment, and optimization. [10][11][12] Due to the changeable requirements of the market and customer, optimization is critical to improve product quality, as well as to assist the process in adapting to the increasingly competitive environment. Necessary condition of optimality (NCO) matching is especially important since it guarantees the feasibility of the solution earned from the model when it is performed in practice, [13] and it is closely relevant to the product competitiveness.…”

Section: Introductionmentioning

confidence: 99%

“…[3] Further, by depth analysis of the mismatch existing in the process transfer model (PTM), the main cause of the mismatch in different stages of transfer was clarified and a system method was proposed for optimization of transfer model. [12,18,19] Although the traditional MA strategy can achieve better optimization results for the plant-model mismatch by pursuing consistency of first-order approximate terms to meet the NCO. However, approximating only to the first-order may miss some critical information, while higher order methods can effectively speed up the convergence.…”

Section: Introductionmentioning

confidence: 99%

Efficient cobalt oxalate synthesis process optimization via second‐order modifier adaptation with transfer learning

Chu,

Chai,

Wang

et al. 2023

Can J Chem Eng

View full text Add to dashboard Cite

This paper proposes a second‐order modifier adaptation optimization based on the transfer model, which aims to improve the optimization efficiency. Although the traditional modifier adaptation strategy adds a bias to the model to meet the necessary condition of optimality (NCO), approximating only to the first‐order may neglect some key higher‐order information. Making use of as much higher‐order information as possible is an effective way to improve the efficiency of optimization. For this issue, by introducing the second‐order information into the modifier adaptation method to compensate the mismatch of higher‐order information during the optimization, the compensation of the mismatch between model and plant can get accelerated, thus improving the optimization efficiency. To overcome the difficulty of insufficient data for new process modelling, a process transfer model is also used to fit the relationship between manipulated variables and final product quality. The simulation study of the cobalt oxalate synthesis process shows that this strategy has better optimization efficiency.

show abstract

Class-Imbalance Adversarial Transfer Learning Network for Cross-Domain Fault Diagnosis With Imbalanced Data

Cited by 66 publications

References 22 publications

Tool wear assessment and life prediction model based on image processing and deep learning

Tool wear assessment and life prediction model based on image processing and deep learning

Cross-Domain Open Set Fault Diagnosis Based on Weighted Domain Adaptation with Double Classifiers

Efficient cobalt oxalate synthesis process optimization via second‐order modifier adaptation with transfer learning

Contact Info

Product

Resources

About