Machine learning for sensor-based manufacturing processes

Moldovan, Daniel; Cioara, Tudor; Anghel, Ionut; Salomie, Ioan

doi:10.1109/iccp.2017.8116997

Cited by 40 publications

(16 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…While examining several related works that perform classification on the SECOM data set via a pipeline of preprocessing stages, it can be seen that many choose to apply some or all the preprocessing on the data before splitting the data into training and test data sets. In [3][4][5], the task of feature selection was applied before cross-validation, and in [1,2,6,11] this task was applied before isolating one third of the data instances into the test set.…”

Section: Cross-validation and The Proper Way To Do Itmentioning

confidence: 99%

“…This balance can be created either by oversampling the minority distribution-in our case the failures-or by under-sampling the majority distribution-i.e., the successes. Since oversampling has been shown to outperform under-sampling [3], we use oversampling in this work. The approach used in this work is Synthetic Minority Oversampling Technique (SMOTE) [19].…”

Section: Synthetic Data Generationmentioning

confidence: 99%

See 1 more Smart Citation

An Experimental Evaluation of Fault Diagnosis from Imbalanced and Incomplete Data for Smart Semiconductor Manufacturing

Salem

Taheri

Yuan

2018

BDCC

View full text Add to dashboard Cite

The SECOM dataset contains information about a semiconductor production line, entailing the products that failed the in-house test line and their attributes. This dataset, similar to most semiconductor manufacturing data, contains missing values, imbalanced classes, and noisy features. In this work, the challenges of this dataset are met and many different approaches for classification are evaluated to perform fault diagnosis. We present an experimental evaluation that examines 288 combinations of different approaches involving data pruning, data imputation, feature selection, and classification methods, to find the suitable approaches for this task. Furthermore, a novel data imputation approach, namely "In-painting KNN-Imputation" is introduced and is shown to outperform the common data imputation technique. The results show the capability of each classifier, feature selection method, data generation method, and data imputation technique, with a full analysis of their respective parameter optimizations.Big Data Cogn. Comput. 2018, 2, 30 2 of 20 of fault detection and diagnosis. Previous work has been done to classify this dataset to handle the imbalanced data and irrelevant features. In [1,2], the challenge of imbalanced data was evaluated and approaches for oversampling the minority distribution to create balance between the classes was introduced. In [3], the challenge of imbalanced data was evaluated from an under-sampling perspective as well, showing that oversampling performs better on this dataset. In [2,4-6] different approaches for feature selection were proposed to rise to the challenge of noisy features. The challenge of missing data remains unexplored in the SECOM dataset. To the authors' knowledge, no literature thoroughly classifies the SECOM dataset after performing specialized data imputation.In this work the SECOM dataset is classified using a plethora of combination of approaches. The three challenges of data imbalance, missing data, and noisy data are handled via synthetic data generation, data imputation, and feature selection, respectively. Moreover, different classifiers are evaluated, and their performance is analyzed based on the task at hand. To face the challenge of missing data, a novel data imputation technique called "In-painting KNN-Imputation" is introduced which is inspired by image in-painting. In the end, leveraging the feature importance delivered by the classification model, fault diagnosis is performed, that demonstrates which features and measured parameters during the manufacturing process have high effect on the failure of the device.The rest of the paper is organized as follows. In Section 2 the background knowledge is provided. In this section, the semiconductor manufacturing, the manufacturing processes in this industry are conversed. The methodology of our work including the classification stages, the processes of data preparation, and procedures of constructing, evaluating, and interpreting the model for the data are discussed in Section 3. The designed and executed ex...

show abstract

Section: Cross-validation and The Proper Way To Do Itmentioning

confidence: 99%

Section: Synthetic Data Generationmentioning

confidence: 99%

An Experimental Evaluation of Fault Diagnosis from Imbalanced and Incomplete Data for Smart Semiconductor Manufacturing

Salem

Taheri

Yuan

2018

BDCC

View full text Add to dashboard Cite

show abstract

“…There have been many studies to build the failure prediction model using feature selection, most of which have selected features considering the importance of each feature [20][21][22]. Moldovan et al built a failure prediction model using the selected features to improve prediction accuracy and performed feature selection using three algorithms (i.e., random forest, regression analysis, and orthogonal linear transformation) to compare the prediction accuracy of each for the comparative study [20]. Mahadevan and Shah used a support vector machine recursive feature elimination (SVM-RFE) algorithm to rank the features by their importance [21].…”

Section: Introductionmentioning

confidence: 99%

Failure Prediction Model Using Iterative Feature Selection for Industrial Internet of Things

Kwon

Kim

2020

Symmetry

View full text Add to dashboard Cite

This paper presents a failure prediction model using iterative feature selection, which aims to accurately predict the failure occurrences in industrial Internet of Things (IIoT) environments. In general, vast amounts of data are collected from various sensors in an IIoT environment, and they are analyzed to prevent failures by predicting their occurrence. However, the collected data may include data irrelevant to failures and thereby decrease the prediction accuracy. To address this problem, we propose a failure prediction model using iterative feature selection. To build the model, the relevancy between each feature (i.e., each sensor) and the failure was analyzed using the random forest algorithm, to obtain the importance of the features. Then, feature selection and model building were conducted iteratively. In each iteration, a new feature was selected considering the importance and added to the selected feature set. The failure prediction model was built for each iteration via the support vector machine (SVM). Finally, the failure prediction model having the highest prediction accuracy was selected. The experimental implementation was conducted using open-source R. The results showed that the proposed failure prediction model achieved high prediction accuracy.

show abstract

“…Review of relevant literature: The data in modern manufacturing challenges can suffer from high dimensionality, complexity, non-linearity and inconsistencies [1][2][3][4]. To address these challenges, machine learning and data analytics methods have been employed [2][3][4][5], which concentrate on predictive maintenance and rare event prediction [6].…”

Section: Introductionmentioning

confidence: 99%

“…Susto et al [3] presented a new multiple classifier model for predictive maintenance along with a simulation study and benchmark dataset; the data needed to be pre-processed in order to allow for a suitable classifier such as k-NN and support vector machines (SVM), to be trained. In [5], different ML methods were compared for a semiconductor manufacturing dataset and highlighted the benefits of reducing the data through feature selection. Work in [7] emphasised the benefits of AI and ML for manufacturing, but there has been little investigation into the statistical basis of data preprocessing to improve model performance and learning procedures.…”

Section: Introductionmentioning

confidence: 99%

A New Data Analytics Framework Emphasising Pre-processing in Learning AI Models for Complex Manufacturing Systems

Carbery

Woods

Marshall

2018

Intelligent Computing and Internet of Things

View full text Add to dashboard Cite

Recent emphasis has been placed on improving the processes in manufacturing by employing early detection or fault prediction within production lines. Whilst companies are increasingly including sensors to record observations and measurements, this brings challenges in interpretation as standard approaches for artificial intelligence (AI) do not highlight the presence of unknown relationships. To address this, we propose a new data analytics framework for predicting faults in a large-scale manufacturing system and validate it using a publicly available Bosch manufacturing dataset with a focus on pre-processing of the data.

show abstract

Machine learning for sensor-based manufacturing processes

Cited by 40 publications

References 15 publications

An Experimental Evaluation of Fault Diagnosis from Imbalanced and Incomplete Data for Smart Semiconductor Manufacturing

An Experimental Evaluation of Fault Diagnosis from Imbalanced and Incomplete Data for Smart Semiconductor Manufacturing

Failure Prediction Model Using Iterative Feature Selection for Industrial Internet of Things

A New Data Analytics Framework Emphasising Pre-processing in Learning AI Models for Complex Manufacturing Systems

Contact Info

Product

Resources

About