TY - GEN
T1 - A Smart Predictive Maintenance Scheme for Classifying Diagnostic and Prognostic Statuses
AU - Palembiya, Revi Asprila
AU - Setiawan, Muhammad Nanda
AU - Gultom, Elnora Oktaviyani
AU - Prayitno, Adila Sekarratri Dwi
AU - Kurniati, Nani
AU - Iqbal, Mohammad
N1 - Publisher Copyright:
© 2021, Springer Nature Singapore Pte Ltd.
PY - 2021
Y1 - 2021
N2 - This study attempts to propose a smart predictive maintenance method to classify manufacturing machines’ diagnostic and prognostic statuses. The main goal of this study is to reduce the manual predictive maintenance budgets of manufactures in Indonesia. In the proposed method, we perform feature maps to obtain the binary states of sensor data, which is further clustered into the machine’s error states (diagnostic status) and the machine’ useful life states (prognostic status). Moreover, the proposed method comprises the two states predictions of machines based on Deep Long Short Term Memory. The proposed method is demonstrated on the Rawmill and Kiln machines of a cement factory in Indonesia for evaluation performances. Without labelling manually, we investigated the annotation of both states, which are similar to the ground truth. In addition, the proposed method can achieved high accuracy and outperformed to another baseline method.
AB - This study attempts to propose a smart predictive maintenance method to classify manufacturing machines’ diagnostic and prognostic statuses. The main goal of this study is to reduce the manual predictive maintenance budgets of manufactures in Indonesia. In the proposed method, we perform feature maps to obtain the binary states of sensor data, which is further clustered into the machine’s error states (diagnostic status) and the machine’ useful life states (prognostic status). Moreover, the proposed method comprises the two states predictions of machines based on Deep Long Short Term Memory. The proposed method is demonstrated on the Rawmill and Kiln machines of a cement factory in Indonesia for evaluation performances. Without labelling manually, we investigated the annotation of both states, which are similar to the ground truth. In addition, the proposed method can achieved high accuracy and outperformed to another baseline method.
KW - Cement factory
KW - Deep learning
KW - Diagnostic state prediction
KW - Prognostic state prediction
KW - Smart predictive maintenance
UR - http://www.scopus.com/inward/record.url?scp=85119424740&partnerID=8YFLogxK
U2 - 10.1007/978-981-16-7334-4_8
DO - 10.1007/978-981-16-7334-4_8
M3 - Conference contribution
AN - SCOPUS:85119424740
SN - 9789811673337
T3 - Communications in Computer and Information Science
SP - 104
EP - 117
BT - Soft Computing in Data Science - 6th International Conference, SCDS 2021, Proceedings
A2 - Mohamed, Azlinah
A2 - Yap, Bee Wah
A2 - Zain, Jasni Mohamad
A2 - Berry, Michael W.
PB - Springer Science and Business Media Deutschland GmbH
T2 - 6th International Conference on Soft Computing in Data Science, SCDS 2021
Y2 - 2 November 2021 through 3 November 2021
ER -