Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning
Soft sensors based on deep learning have been growing in industrial process applications, inferring hard-to-measure but crucial quality-related variables. However, applications may present strong non-linearity, dynamicity, and a lack of labeled data. To deal with the above-cited problems, the extrac...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-05-01
|
| Series: | Sensors |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1424-8220/21/10/3430 |
| _version_ | 1797533993738960896 |
|---|---|
| author | Jean Mário Moreira de Lima Fábio Meneghetti Ugulino de Araújo |
| author_facet | Jean Mário Moreira de Lima Fábio Meneghetti Ugulino de Araújo |
| author_sort | Jean Mário Moreira de Lima |
| collection | DOAJ |
| description | Soft sensors based on deep learning have been growing in industrial process applications, inferring hard-to-measure but crucial quality-related variables. However, applications may present strong non-linearity, dynamicity, and a lack of labeled data. To deal with the above-cited problems, the extraction of relevant features is becoming a field of interest in soft-sensing. A novel deep representative learning soft-sensor modeling approach is proposed based on stacked autoencoder (SAE), mutual information (MI), and long-short term memory (LSTM). SAE is trained layer by layer with MI evaluation performed between extracted features and targeted output to evaluate the relevance of learned representation in each layer. This approach highlights relevant information and eliminates irrelevant information from the current layer. Thus, deep output-related representative features are retrieved. In the supervised fine-tuning stage, an LSTM is coupled to the tail of the SAE to address system inherent dynamic behavior. Also, a k-fold cross-validation ensemble strategy is applied to enhance the soft-sensor reliability. Two real-world industrial non-linear processes are employed to evaluate the proposed method performance. The obtained results show improved prediction performance in comparison to other traditional and state-of-art methods. Compared to the other methods, the proposed model can generate more than 38.6% and 39.4% improvement of RMSE for the two analyzed industrial cases. |
| first_indexed | 2024-03-10T11:24:30Z |
| format | Article |
| id | doaj.art-09a3e97d90f2447ab5d04ace16b57ec2 |
| institution | Directory Open Access Journal |
| issn | 1424-8220 |
| language | English |
| last_indexed | 2024-03-10T11:24:30Z |
| publishDate | 2021-05-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Sensors |
| spelling | doaj.art-09a3e97d90f2447ab5d04ace16b57ec22023-11-21T19:46:37ZengMDPI AGSensors1424-82202021-05-012110343010.3390/s21103430Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation LearningJean Mário Moreira de Lima0Fábio Meneghetti Ugulino de Araújo1Computer Engineering and Automation Department, Federal University of Rio Grande do Norte, 3000 Senador Salgado Filho Avenue, Natal, RN 59078970, BrazilComputer Engineering and Automation Department, Federal University of Rio Grande do Norte, 3000 Senador Salgado Filho Avenue, Natal, RN 59078970, BrazilSoft sensors based on deep learning have been growing in industrial process applications, inferring hard-to-measure but crucial quality-related variables. However, applications may present strong non-linearity, dynamicity, and a lack of labeled data. To deal with the above-cited problems, the extraction of relevant features is becoming a field of interest in soft-sensing. A novel deep representative learning soft-sensor modeling approach is proposed based on stacked autoencoder (SAE), mutual information (MI), and long-short term memory (LSTM). SAE is trained layer by layer with MI evaluation performed between extracted features and targeted output to evaluate the relevance of learned representation in each layer. This approach highlights relevant information and eliminates irrelevant information from the current layer. Thus, deep output-related representative features are retrieved. In the supervised fine-tuning stage, an LSTM is coupled to the tail of the SAE to address system inherent dynamic behavior. Also, a k-fold cross-validation ensemble strategy is applied to enhance the soft-sensor reliability. Two real-world industrial non-linear processes are employed to evaluate the proposed method performance. The obtained results show improved prediction performance in comparison to other traditional and state-of-art methods. Compared to the other methods, the proposed model can generate more than 38.6% and 39.4% improvement of RMSE for the two analyzed industrial cases.https://www.mdpi.com/1424-8220/21/10/3430soft sensorsdeep learningstacked autoencodersmutual informationLSTM |
| spellingShingle | Jean Mário Moreira de Lima Fábio Meneghetti Ugulino de Araújo Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning Sensors soft sensors deep learning stacked autoencoders mutual information LSTM |
| title | Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning |
| title_full | Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning |
| title_fullStr | Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning |
| title_full_unstemmed | Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning |
| title_short | Industrial Semi-Supervised Dynamic Soft-Sensor Modeling Approach Based on Deep Relevant Representation Learning |
| title_sort | industrial semi supervised dynamic soft sensor modeling approach based on deep relevant representation learning |
| topic | soft sensors deep learning stacked autoencoders mutual information LSTM |
| url | https://www.mdpi.com/1424-8220/21/10/3430 |
| work_keys_str_mv | AT jeanmariomoreiradelima industrialsemisuperviseddynamicsoftsensormodelingapproachbasedondeeprelevantrepresentationlearning AT fabiomeneghettiugulinodearaujo industrialsemisuperviseddynamicsoftsensormodelingapproachbasedondeeprelevantrepresentationlearning |