Juan Pablo Usuga Cadavid
Research Subjects
Juan Pablo Usuga Cadavid has a total of 7 co-authors. Here are the top co-authors by number of shared publications:
Arnaud Fortin
FAKT France SAS
5 shared publicationsBernard Grabot
LGP, INP/ENIT
4 shared publications- Samir Lamouri
LAMIH-UMR CNRS 8201, Arts et Métiers Sciences et Technologies, Institute of Technology, 75013 Paris, France
3 shared publications - Robert Pellerin
Polytechnique Montreal
2 shared publications Samir Lamouri
LAMIH UMR CNRS 8201, Arts et Métiers – Institute of Technology
2 shared publications- Bernard Grabot
Université de Toulouse
1 shared publication
Latest Publications
- 29 July 2021
- Published by Springer Nature
- p. 57-68
Artificial Data Generation with Language Models for Imbalanced Classification in Maintenance
- 29 July 2021
- Published by Springer Nature
- p. 57-68
Artificial Data Generation with Language Models for Imbalanced Classification in Maintenance
- 18 July 2021
- Published by Taylor & Francis in International Journal of Production Research
- Vol. 60 (14) , 4548-4575
Using deep learning to value free-form text data for predictive maintenance
Past maintenance logs may encapsulate meaningful data for predicting the duration of machine breakdowns, the potential causes of a problem, or the necessity to stop production to perform repair activities. These insights may be accessed using machine learning (ML). However, maintenance logs tend to have imbalanced distributions and rely on noisy unstructured text data provided by operators. Additionally, the limited interpretability of ML models results in human reluctance when accepting model predictions. Hence, this study explored the use of two recent deep learning models (CamemBERT and FlauBERT) for natural language processing (NLP) to harness unstructured data from maintenance logs. The class imbalance effect was mitigated using data-level and algorithm-level approaches. To improve interpretability, a technique called LIME was employed to interpret single predictions and to propose a method for insight extraction from several maintenance reports. Results suggest three key points: CamemBERT and FlauBERT can achieve excellent results with minimum text pre-processing and hyperparameter tuning. Second, random oversampling (ROS) generally mitigates the effect of class imbalance. However, ROS was observed to be unnecessary when performing pertinent data pre-processing. Finally, at the maintenance level, the proposed insight extraction method can provide valuable information from a set of poorly structured maintenance reports.
- 18 July 2021
- Published by Taylor & Francis in International Journal of Production Research
- Vol. 60 (14) , 4548-4575
Using deep learning to value free-form text data for predictive maintenance
Past maintenance logs may encapsulate meaningful data for predicting the duration of machine breakdowns, the potential causes of a problem, or the necessity to stop production to perform repair activities. These insights may be accessed using machine learning (ML). However, maintenance logs tend to have imbalanced distributions and rely on noisy unstructured text data provided by operators. Additionally, the limited interpretability of ML models results in human reluctance when accepting model predictions. Hence, this study explored the use of two recent deep learning models (CamemBERT and FlauBERT) for natural language processing (NLP) to harness unstructured data from maintenance logs. The class imbalance effect was mitigated using data-level and algorithm-level approaches. To improve interpretability, a technique called LIME was employed to interpret single predictions and to propose a method for insight extraction from several maintenance reports. Results suggest three key points: CamemBERT and FlauBERT can achieve excellent results with minimum text pre-processing and hyperparameter tuning. Second, random oversampling (ROS) generally mitigates the effect of class imbalance. However, ROS was observed to be unnecessary when performing pertinent data pre-processing. Finally, at the maintenance level, the proposed insight extraction method can provide valuable information from a set of poorly structured maintenance reports.
- 7 July 2020
- Published by Taylor & Francis in Enterprise Information Systems
- Vol. 16 (6)
Valuing free-form text data from maintenance logs through transfer learning with CamemBERT
Coupling a production scheduling process with maintenance logs can provide important advantages. For instance, this enables the adaptation of planning to the reality of the shop floor. Nevertheless, maintenance logs are often highly unstructured, as they mainly rely on free-form text comments from operators, and are imbalanced, as commonplace issues happen more often than critical problems. This hinders the application of machine learning methods to exploit this data. Thus, this study explores the use of a recent model named CamemBERT to tackle these difficulties through transfer learning. More specifically, the purpose is to predict the criticality and duration of a maintenance issue from the description provided. Findings suggest that fine-tuning CamemBERT outperforms other classical and feature-based approaches. Furthermore, the class imbalance problem is addressed from a data pre-processing and training perspective: firstly, k-means with silhouette diagrams allowed the creation of more homogenous classes, and secondly, the use of resampling enabled an improvement in the model’s performance.
- 7 July 2020
- Published by Taylor & Francis in Enterprise Information Systems
- Vol. 16 (6)
Valuing free-form text data from maintenance logs through transfer learning with CamemBERT
Coupling a production scheduling process with maintenance logs can provide important advantages. For instance, this enables the adaptation of planning to the reality of the shop floor. Nevertheless, maintenance logs are often highly unstructured, as they mainly rely on free-form text comments from operators, and are imbalanced, as commonplace issues happen more often than critical problems. This hinders the application of machine learning methods to exploit this data. Thus, this study explores the use of a recent model named CamemBERT to tackle these difficulties through transfer learning. More specifically, the purpose is to predict the criticality and duration of a maintenance issue from the description provided. Findings suggest that fine-tuning CamemBERT outperforms other classical and feature-based approaches. Furthermore, the class imbalance problem is addressed from a data pre-processing and training perspective: firstly, k-means with silhouette diagrams allowed the creation of more homogenous classes, and secondly, the use of resampling enabled an improvement in the model’s performance.
- Top Cited Papers
- 11 January 2020
- Published by Springer Nature in Journal of Intelligent Manufacturing
- Vol. 31 (6) , 1531-1558
Machine learning applied in production planning and control: a state-of-the-art in the era of industry 4.0
- 11 January 2020
- Published by Springer Nature in Journal of Intelligent Manufacturing
- Vol. 31 (6) , 1531-1558
Top Cited PapersMachine learning applied in production planning and control: a state-of-the-art in the era of industry 4.0
- 1 September 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Estimation of Production Inhibition Time Using Data Mining to Improve Production Planning and Control
To be competitive under the paradigm of Industry 4.0 (I4.0), companies must develop a Production Planning and Control (PPC) able to respond to disturbances. Notably, unexpected machine breaks leading to corrective maintenance actions drastically delay operations as they inhibit the machine production, affecting the production schedule. To create an adaptive PPC, data from the shop floor must be collected and analyzed. However, these data are often unstructured, as they come from either sensors or humans, which makes their use difficult. Notably, human based operations such as maintenance activities often produce textual data. Hence, the objective of this paper is twofold: firstly, it aims to propose a framework for a model capable of estimating the production inhibition time based on textual data. Secondly, it settles the basis for a Dynamic Production Schedule (DPS) model considering these estimations. To achieve this, an approach using Text Mining (TM) and Machine Learning (ML) techniques is proposed.
- 1 September 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 1-6
Estimation of Production Inhibition Time Using Data Mining to Improve Production Planning and Control
To be competitive under the paradigm of Industry 4.0 (I4.0), companies must develop a Production Planning and Control (PPC) able to respond to disturbances. Notably, unexpected machine breaks leading to corrective maintenance actions drastically delay operations as they inhibit the machine production, affecting the production schedule. To create an adaptive PPC, data from the shop floor must be collected and analyzed. However, these data are often unstructured, as they come from either sensors or humans, which makes their use difficult. Notably, human based operations such as maintenance activities often produce textual data. Hence, the objective of this paper is twofold: firstly, it aims to propose a framework for a model capable of estimating the production inhibition time based on textual data. Secondly, it settles the basis for a Dynamic Production Schedule (DPS) model considering these estimations. To achieve this, an approach using Text Mining (TM) and Machine Learning (ML) techniques is proposed.