Siamese Neural Network-Based Supervised Slow Feature Extraction for Soft Sensor Application

Chiplunkar, Ranjith; Huang, Biao

doi:10.1109/tie.2020.3014574

Cited by 34 publications

(9 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Besides, slow and multi-dimensional feature extractions are also crucial in industrial processes. A Siamese network, called SSFAN (Supervised Slow Feature Analysis Siamese Network) [18], extracts the latent features from time series based on temporal slowness aspect, while SS-PdeepFM [11] extracts low-and high-dimensional features in a single time step. However, the SAE-based models may trivially copy their inputs to outputs without finding useful patterns in the data.…”

Section: B Deep Feature Extractionmentioning

confidence: 99%

“…Besides, GSTAE has three hidden layers of (256, 128, 64) units and the label is normalized to [0, 1] to meet the output requirements of GSTAE. The design of the model structure and hyperparameters were taken from [17,18], and the number of neurons was scaled up in order to enhance the model performance. 3) Evaluation: Some other experiments are conducted to verify the effectiveness of the model in soft sensor modeling.…”

Section: B Experimentsmentioning

confidence: 99%

See 1 more Smart Citation

A Data-Driven Self-Supervised LSTM-DeepFM Model for Industrial Soft Sensor

Ren

Wang

Laili

et al. 2022

IEEE Trans. Ind. Inf.

View full text Add to dashboard Cite

Soft sensor, as an important paradigm for industrial intelligence, is widely used in industrial production to achieve efficient monitoring and prediction of production status including product quality. Data-driven soft sensor methods have attracted attention, which still have challenges because of complex industrial data with diverse characteristics, nonlinear relationships and massive unlabeled samples. In this paper, a data-driven self-supervised long short term memory-deep factorization machine (LSTM-DeepFM) model is proposed for industrial soft sensor, in which a framework mainly including pretraining and finetuning stages is proposed to explore diverse industrial data characteristics. In the pretraining stage, LSTM-Autoencoder is first unsupervised pretrained. Then, based on two self-supervised mask strategies, LSTM-Deep can explore the interdependencies between features as well as the dynamic fluctuation in time series. In the finetuning stage, relying on pretrained representation, the temporal, high-dimensional and low-dimensional features can be extracted from the LSTM, Deep and FM components respectively. Finally, experiments on the real-world mining dataset demonstrate that the proposed method achieves state-of-the-art comparing with stacked autoencoder (SAE) based models, variational autoencoder (VAE) based models, and semisupervised parallel DeepFM (SS-PdeepFM), etc.

show abstract

Section: B Deep Feature Extractionmentioning

confidence: 99%

Section: B Experimentsmentioning

confidence: 99%

A Data-Driven Self-Supervised LSTM-DeepFM Model for Industrial Soft Sensor

Ren

Wang

Laili

et al. 2022

IEEE Trans. Ind. Inf.

View full text Add to dashboard Cite

show abstract

“…A Siamese neural network is a coupling architecture based on two neural networks that is used to discriminate the similarity of two data sets. The basic idea of a Siamese neural network is to use the same neural network to extract features of the two data and then a selected distance metric to determine if the two data sets are similar. , The specific architecture is shown in Figure . A Siamese neural network contains two identical subnetworks.…”

Section: Preliminariesmentioning

confidence: 99%

A Novel Spatiotemporal Process Feature Learning Method Based On the Pseudo-Siamese Network for Complex Chemical Process Concurrent Condition Monitoring

Jia

Mao

2022

ACS Omega

View full text Add to dashboard Cite

The deep learning-based process monitoring method has attracted great attention due to its ability to deal with nonlinear correlation. However, the further modeling of learned deep features from process data to better depict typical process features to obtain more precise monitoring results remains a challenge. In this paper, a novel nonlinear spatiotemporal process feature learning method is proposed to extract high-value slow-varying spatiotemporal process features, with an explicit temporal relationship model for the concurrent monitoring of the static deviation and the dynamic anomaly of complex chemical processes. Different from directly mixed spatiotemporal information methods, the pseudo-Siamese autoencoder network is designed with two different subencoders to separately describe the nonlinear spatial and temporal relationships of the nonlinear dynamic input data. Correspondingly, a cost function including three losses and one orthogonal constraint is proposed to make sure that the extracted spatiotemporal process features change as slowly as possible and contain the most nonlinear dynamic information on the input data. With the explicit spatial and temporal relationship submodel, predictions are utilized to shrink the variability of the nonlinear temporal correlated data and focus on the unpredictable variabilities to improve process monitoring performance. Meanwhile, the linear dynamic information is further extracted in the reconstructed residual space by the general slow feature analysis (SFA) method to provide a more detailed analysis of the process characteristics and improve the monitoring results. The case study monitoring results demonstrate the effectiveness and superiority of the proposed method over other compared methods for concurrent process monitoring.

show abstract

“…Encoder-decoder networks have achieved superior performance in learning representation from data [27]- [29]. Recent frameworks have considered variational learning of sequential data using recurrent neural networks.…”

Section: Introductionmentioning

confidence: 99%

Nonlinear Semi-supervised Inference Networks for the Extraction of Slow Oscillating Features

Puli¹

2022

Preprint

Self Cite

View full text Add to dashboard Cite

<p>Due to physical and monetary constraints, quality-related variables in the process industries are generally difficult to measure using hardware sensors. Therefore, they are often not available as frequently as the other variables since obtaining them typically via laboratory analysis are time-consuming. On the other hand, the dynamic nature of easy-to-measure process data contains valuable predictive information about the quality variables. Consequently, modelling sequential data is beneficial in building a soft sensor that can predict the quality variable. Several techniques were proposed to extract dynamic features as measured data often suffers from noise and collinearity. This paper presents a semi-supervised oscillating slow feature inference network to address certain shortcomings of the existing solutions. The proposed learning algorithm uses a gated recurrent unit to learn a combined inference network for missing quality variable imputation and oscillating slow feature extraction from nonlinear process data. We evaluate the efficacy of the proposed methodology on a simulated and an industrial process.</p>

show abstract

Siamese Neural Network-Based Supervised Slow Feature Extraction for Soft Sensor Application

Cited by 34 publications

References 37 publications

A Data-Driven Self-Supervised LSTM-DeepFM Model for Industrial Soft Sensor

A Data-Driven Self-Supervised LSTM-DeepFM Model for Industrial Soft Sensor

A Novel Spatiotemporal Process Feature Learning Method Based On the Pseudo-Siamese Network for Complex Chemical Process Concurrent Condition Monitoring

Nonlinear Semi-supervised Inference Networks for the Extraction of Slow Oscillating Features

Contact Info

Product

Resources

About