Power prediction of a 4-CRU parallel mechanism based on extra gradient boosting regressor

Mochammad Solichin*, Latifah Nurahmi, Bimo Jati Putro

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

Abstract

Power analysis of a robot is the last step of the dynamic modelling after calculating its kinematics. The power calculation takes a long time due to the complexity of the process. This paper deals with the power prediction of a 4-CRU parallel mechanism by using Extra Gradient Boosting Regressor (XGBR). XGBR is one of the algorithms in machine learning that can take good learners and leave weak learners from the models built. Then, XGBR is optimized using Random Search to overcome the hyper parameter tuning with the best prediction accuracy. The XGBR tuned hyper parameter was performed a satisfactory prediction model. The prediction results can perfectly show the robot behavior since it is based on the smallest error of model prediction. The error value based on Mean Absolute Percentage Error (MAPE) is 0.05099% and based on Mean Square error (MSE) is 0.0001 which took 11 minutes. The value of accuracy and efficiency is very reasonable to say that the power prediction model of a 4-CRU parallel mechanism has successfully performed.

Original languageEnglish
Title of host publicationInnovative Science and Technology in Mechanical Engineering for Industry 4.0
Subtitle of host publicationProceedings of the 4th International Conference on Mechanical Engineering, ICOME 2019
EditorsVivien Djanali, Fahmi Mubarok, Bambang Pramujati, Suwarno
PublisherAmerican Institute of Physics Inc.
ISBN (Electronic)9780735419346
DOIs
Publication statusPublished - 10 Dec 2019
Event4th International Conference on Mechanical Engineering: Innovative Science and Technology in Mechanical Engineering for Industry 4.0, ICOME 2019 - Yogyakarta, Indonesia
Duration: 28 Aug 201929 Aug 2019

Publication series

NameAIP Conference Proceedings
Volume2187
ISSN (Print)0094-243X
ISSN (Electronic)1551-7616

Conference

Conference4th International Conference on Mechanical Engineering: Innovative Science and Technology in Mechanical Engineering for Industry 4.0, ICOME 2019
Country/TerritoryIndonesia
CityYogyakarta
Period28/08/1929/08/19

Fingerprint

Dive into the research topics of 'Power prediction of a 4-CRU parallel mechanism based on extra gradient boosting regressor'. Together they form a unique fingerprint.

Cite this