State estimation in a hydraulically actuated log crane using Unscented Kalman Filter
|Author:||Khadim, Qasim1; Yashar Shabbouei, Hagh2; Pyrhönen, Lauri1;|
1Department of Mechanical Engineering, Lappeenranta University of Technology, 53850 Lappeenranta, Finland
2Laboratory of Intelligent Machines, Department of Mechanical Engineering, Lappeenranta University of Technology, 53850 Lappeenranta, Finland
3Materials and Mechanical Engineering research unit, University of Oulu, 90570 Oulu, Finland
|Online Access:||PDF Full Text (PDF, 6.7 MB)|
|Persistent link:|| http://urn.fi/urn:nbn:fi-fe2022061446331
Institute of Electrical and Electronics Engineers,
|Publish Date:|| 2022-06-14
Multibody system dynamics approaches together with state estimation methods can reduce the need for a large number of sensors, especially in the digital twin of working mobile machinery. To demonstrate this, a hydraulically actuated machine was modeled using the double-step semi-recursive multibody formulation and lumped fluid theory in terms of system independent states. Next, because of the high non-linearity of the modeled system and with respect to the reported performance degradation of the Extended Kalman Filters (EKF), which are mostly related to the linearization procedure of this filter, the Unscented Kalman Filter (UKF) was implemented to achieve high accuracy and performance. The methodology of the proposed approaches was applied to a mobile log crane model PATU 655. The implementation of the proposed estimation algorithms is demonstrated with three different multibody based simulation models: the synthetic real system producing the artificial measurements, the simulation model, and the estimation model. Encoders and pressure sensors, installed on the synthetic real system, provided synthetic sensor measurement data. To mimic real-world conditions, the estimation and simulation models used in the processing of the state estimation algorithm were assumed to have errors in the initial conditions and force model. The proposed UKF was applied to the estimation model with the synthetic sensor measurement data. The minimum percent normalized root mean square errors in the associated measured and unmeasured states of case example were 0.11% and 1.86%, respectively. The UKF using the multibody system dynamics formulations is able to estimate the case example states despite 15% and 60% errors in mass and inertial properties of bodies and Payload, respectively, confirming the accuracy and performance of the algorithm.
|Type of Publication:||
A1 Journal article – refereed
|Field of Science:||
214 Mechanical engineering
This work was supported in part by Business Finland (project: Service Business from Physics Based Digital Twins—DigiBuzz-LUT (4329/31/2019)) and in part by the Academy of Finland under Grant #316106.
© The Author(s) 2022. This work is licensed under a Creative Commons Attribution 4.0 License. For more information, see https://creativecommons.org/licenses/by/4.0/.