卡尔曼滤波器
动态模态分解
控制理论(社会学)
计算机科学
核(代数)
非线性系统
滤波器(信号处理)
扩展卡尔曼滤波器
控制工程
工程类
人工智能
数学
计算机视觉
机器学习
控制(管理)
物理
组合数学
量子力学
作者
Yu Caol,Mengshi Zhang,Bo Yang,Jin Huang
标识
DOI:10.1109/icdl55364.2023.10364545
摘要
This paper presents a data-driven modeling approach for enhancing the understanding of pneumatic muscles (PMs) through Koopman kalman filter (KKF). The proposed method involves the construction of a comprehensive pneumatic muscle motion data-set, obtained by capturing extensive input-state-output data through the application of random inputs. Subsequently, the states of pneumatic muscle are lifted through a kernel-based technique. The nonlinear pneumatic muscle model is then transformed into a linear counterpart using the extended dynamic mode decomposition (EDMD). Then, a kalman filter is further used to improve the model accuracy. The effectiveness of this approach is demonstrated through simulation analysis, highlighting its capacity to yield superior pneumatic muscle modeling and state estimation results when compared to conventional Koopman operator methodologies.
科研通智能强力驱动
Strongly Powered by AbleSci AI