Zhu Yancheng, Wu Huaiyu, Chen Zhihuan, Zhu Zhenhua, Chen Yang, Zheng Xiujuan
Engineering Research Center for Metallurgical Automation and Measurement Technology of Ministry of Education, Wuhan University of Science and Technology, Wuhan, 430081, China.
College of Science, Wuhan University of Science and Technology, Wuhan, 430081, China.
Sci Rep. 2024 Dec 30;14(1):32013. doi: 10.1038/s41598-024-83654-3.
This paper mainly studies the issue of fractional parameter identification of generalized bilinear-in-parameter system(GBIP) with colored noise. Hierarchical fractional least mean square algorithm based on the key term separation principle(K-HFLMS) and multi-innovation hierarchical fractional least mean square algorithm based on the key term separation principle (K-MHFLMS) are presented for the effective parameter estimation of GBIP system. The K-MHFLMS expands the scalar innovation into the vector innovation by making full use of the system input and output data information at each recursive step. The detailed performance analyses of the K-MHFLMS strategy are compared with the K-HFLMS algorithm for GBIP identification model based on the Fitness metrics, the mean square error metrics and the average predicted output error. The effectiveness and reliability of K-HFLMS and K-MHFLMS algorithms are further verified through the simulation experimentation under different noise variances, fractional orders and innovation lengths, and the K-MHFLMS yields faster convergence speed than the K-HFLMS by increasing the innovation length.
本文主要研究具有有色噪声的广义参数双线性系统（GBIP）的分数阶参数辨识问题。提出了基于关键项分离原理的分层分数阶最小均方算法（K-HFLMS）和基于关键项分离原理的多新息分层分数阶最小均方算法（K-MHFLMS），用于GBIP系统的有效参数估计。K-MHFLMS通过在每个递归步骤中充分利用系统输入和输出数据信息，将标量新息扩展为向量新息。基于适应度指标、均方误差指标和平均预测输出误差，将K-MHFLMS策略的详细性能分析与用于GBIP辨识模型的K-HFLMS算法进行了比较。通过在不同噪声方差、分数阶和新息长度下的仿真实验，进一步验证了K-HFLMS和K-MHFLMS算法的有效性和可靠性，并且通过增加新息长度，K-MHFLMS比K-HFLMS具有更快的收敛速度。
