Research Article
Improved Sliding Window Kernel RLS Algorithm for Identification of Time-Varying Nonlinear Systems
@INPROCEEDINGS{10.1007/978-3-030-94554-1_18, author={Xinyu Guo and Menghua Jiang and Ying Gao and Shifeng Ou and Jindong Xu and Zhuoran Cai}, title={Improved Sliding Window Kernel RLS Algorithm for Identification of Time-Varying Nonlinear Systems}, proceedings={Advanced Hybrid Information Processing. 5th EAI International Conference, ADHIP 2021, Virtual Event, October 22-24, 2021, Proceedings, Part II}, proceedings_a={ADHIP PART 2}, year={2022}, month={1}, keywords={Kernel recursive least squares Nonlinear system Variable sliding window System identification}, doi={10.1007/978-3-030-94554-1_18} }- Xinyu Guo
Menghua Jiang
Ying Gao
Shifeng Ou
Jindong Xu
Zhuoran Cai
Year: 2022
Improved Sliding Window Kernel RLS Algorithm for Identification of Time-Varying Nonlinear Systems
ADHIP PART 2
Springer
DOI: 10.1007/978-3-030-94554-1_18
Abstract
The sliding window kernel recursive least squares (SW-KRLS) algorithm is one of the most widely used approach in dealing with nonlinear problems because of its simple structure, low computational complexity and high predictive accuracy. However, as data size increases, the computational efficiency of the SW-KRLS algorithm will be affected by the redundant data and the size of sliding window. In order to solve these problems, this paper proposes a variable sliding window sparse kernel recursive least squares (VSWS-KRLS) algorithm. It first uses the sliding window to constrain the size of the novelty criterion dictionary. After that, the algorithm combines the improved novelty criterion with the SW-KRLS to remove the less relevant data. In addition, mechanisms for window size adjustment are added to adjust the size of sliding window adaptively according to the system changes. The experimental results show that the proposed algorithm has better performance in identification of nonlinear system.
