Modeling Behaviors for a New Compliant Mechanism by Recursive Hierarchy of Random Forest Models
Compliant mechanisms have been widely employed for precision engineering. Due to a kinematic coupling between rigid kinematics and flexible kinematics of compliant mechanisms, modeling behavior of two degrees of freedom (2-DOF) compliant mechanism is a challenging task. Therefore, the goal of this p...
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IEEE
2025-01-01
|
| Series: | IEEE Access |
| Subjects: | |
| Online Access: | https://ieeexplore.ieee.org/document/10819341/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Compliant mechanisms have been widely employed for precision engineering. Due to a kinematic coupling between rigid kinematics and flexible kinematics of compliant mechanisms, modeling behavior of two degrees of freedom (2-DOF) compliant mechanism is a challenging task. Therefore, the goal of this paper is to model complicated behaviors of the 2-DOF compliant mechanism. Random Forest is one of the popular and best machine learning models for tabular data. However, most of them have aggregated the component predictions based on the average or the voting process. The objective of this paper is to explore a more reasonable rule for aggregate. In the new technique, several Random Forest regressors are utilized as the weak learners, and their predictions are considered predictors again. The aggregated rule is then effectively learned by a primary Random Forest. The new method is subsequently applied to modeling the behaviors of a two-degrees of freedom compliant mechanism, a research area to which applying machine learning methods is limited. The numerical results demonstrate that the proposed hierarchical manner can improve the performance of component Random Forest models tuned with different values of number of trees and max-depth. Particularly, the results on the test sets of the two case studies have shown that the proposed method can reduce the MAE by 0.02 to 0.04 (about 10%) compared to the best method among the component RFs. Those results verify the applicability of the proposed method to compliant mechanism data. |
|---|---|
| ISSN: | 2169-3536 |