A Dynamic Active Safe Semi-Supervised Learning Framework for Fault Identification in Labeled Expensive Chemical Processes

Jia, Xuqing; Tian, Wende; Li, Chuankun; Xia, Yang; Luo, Zhongjun; Wang, Hui

doi:10.3390/pr8010105

Cited by 10 publications

(5 citation statements)

References 25 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The Tennessee Eastman Process (TEP) is a commonly used benchmark in industrial and chemical engineering research and it has been thoroughly investigated in terms of process dynamics and control [39][40][41][42][43]. Recently, it has been also used to validate active learning or soft sensor modeling approaches [44][45][46][47][48]. It was initially published in 1993 [49] but since then it has been further developed and improved.…”

Section: Tennessee Eastman Processmentioning

confidence: 99%

Stream-based active learning with linear models

Cacciarelli

Külahçı

Tyssedal

2022

Knowledge-Based Systems

View full text Add to dashboard Cite

Section: Tennessee Eastman Processmentioning

confidence: 99%

Stream-based active learning with linear models

Cacciarelli

Külahçı

Tyssedal

2022

Knowledge-Based Systems

View full text Add to dashboard Cite

“…The variables used in this study are shown in Table 1. IDV (15) was not studied because, in previous studies conducted by Zhan et al [20], Wu et al [21], and Jia et al [27], IDV (15) is very difficult to classify. IDV (15) behavior is like IDV (0), so the program will need to diagnose no oscillation.…”

Section: Methodsmentioning

confidence: 99%

Intermittent Oscillation Diagnosis in a Control Loop Using Extreme Gradient Boosting

Rabba

Wardana

Effendy

2022

JNTE

View full text Add to dashboard Cite

The control loop in the industry is a component that must be maintained because it will determine the plant's performance. Most industrial controllers experience oscillations with various causes, such as noise, oscillation, backlash, dead band, hysteresis, random variation, and poor controller tuning. The oscillation diagnosis system, which can understand the oscillation type characteristics, is built based on machine learning because it is dynamic and not based on specific rules. This study developed an online oscillation diagnosis program using the extreme gradient boosting (XGBoost) method. The data was obtained through the simulation of the Tennessee Eastman process. The data is segmented on specific window sizes, and then time series feature extraction is performed. The extraction results are then used to build an XGBoost model capable of performing oscillation diagnosis tasks. There are seven types of oscillations tested in this study. The model that has been made is implemented online with the help of sliding windows. The results show that the XGBoost model performs best when the data window size is 100, with the accuracy performance and the F1 score of the model in classifying the type of oscillation being 0.918 and 0.905, respectively. The model can detect the type of oscillation with an average diagnosis time of 712 seconds on diagnostic tests.

show abstract

“…In addition, the amount of sample data for each fault type may be unbalanced. Therefore, in order to solve the problem that the datasets are usually unlabeled and unbalanced in real situations, some scholars have adopted the semi-supervised learning (SSL) approach to fully utilize the unlabeled fault sample data for the task of fault diagnosis [6][7][8].…”

Section: Introductionmentioning

confidence: 99%

A novel semi-supervised learning rolling bearing fault diagnosis method based on SNNGAN

Qiu,

Fan,

Liang

et al. 2024

Meas. Sci. Technol.

View full text Add to dashboard Cite

n practical industrial environments, rotating machinery typically operates under normal conditions. As a result, the signals collected are primarily normal signals. This imbalance in the sample data diminishes the effectiveness of fault diagnosis. To address this issue, this paper produces a novel semi-supervised fault diagnosis approach based on a Siamese neural network combined with a generative adversarial network (SNNGAN) to enhance classification accuracy. Firstly, vibration signals collected are subjected to continuous wavelet transformation to obtain time-frequency representations, which are utilized for pre-training convolutional encoders in the generator and discriminator. Subsequently, a cosine similarity algorithm is employed to ensure the quality of generated samples. For generated data, set a similarity threshold. Those surpassing the threshold are assigned their corresponding labels and added to the original sample set. Otherwise, those falling below the threshold are transformed back into vibration vectors through an inverse transform and then serve as input to create new samples. Finally, fault diagnosis experiments are conducted on the newly balanced data set. In four imbalanced data experiments, the results demonstrate that SNNGAN outperforms other methods in average accuracy, G-mean, and F1 score, with average accuracy values of 0.919, 0.948, 0.927, and 0.953 for the respective datasets. Therefore, SNNGAN exhibits outstanding fault diagnosis performance under conditions of data imbalance.

show abstract

A Dynamic Active Safe Semi-Supervised Learning Framework for Fault Identification in Labeled Expensive Chemical Processes

Cited by 10 publications

References 25 publications

Stream-based active learning with linear models

Stream-based active learning with linear models

Intermittent Oscillation Diagnosis in a Control Loop Using Extreme Gradient Boosting

A novel semi-supervised learning rolling bearing fault diagnosis method based on SNNGAN

Contact Info

Product

Resources

About