鲁西坤, 焦龙, 卢超.Hammerstein系统的子迭代更新多新息梯度估计方案[J].河南理工大学学报（自然科学版）,2023,42(4):130-139.

LU X K, JIAO L, LU C.Sub-iteration updating multi-innovation gradient estimation algorithm for Hammerstein system[J].Journal of Henan Polytechnic University(Natural Science) ,2023,42(4):130-139.

Hammerstein系统的子迭代更新多新息梯度估计方案

鲁西坤, 焦龙, 卢超

安阳工学院 电子信息与电气工程学院，河南 安阳 455000

摘要:针对Hammerstein系统阶次未知和多新息长度限制辨识性能问题，提出行列式比确定阶次信息和基于子迭代更新多新息梯度辨识算法。结合半分解技术，构建线性和非线性参数相互分离的辨识模型。为便于后续参数辨识，基于参考模型原理和原始系统设计行列式比算法确定系统阶次信息。根据多次迭代更新理论，将给定多新息更新分解为子新息多次迭代更新，改善辨识性能，解决多新息长度限制辨识性能问题。基于殃差收敛定理，从理论上严格分析提出方案的收敛性能。结果表明，与已有的辨识方案相比，本文方案能有效辨识系统的参数信息，具有较高的辨识性能。

关键词:参数辨识;多新息思想;参考模型;Hammerstein系统;子迭代更新

doi:10.16186/j.cnki.1673-9787.2021080086

基金项目:国家自然科学基金资助项目（62006213）； 河南省科技计划项目（222102240076）

收稿日期:2021/08/24

修回日期:2021/12/12

出版日期:2023/07/25

Sub-iteration updating multi-innovation gradient estimation algorithm for Hammerstein system

LU Xikun, JIAO Long, LU Chao

School of Electronic Information and Electrical Engineering，Anyang Institute of Technology，Anyang 455000，Henan，China

Abstract:To solve the unknown order of Hammerstein and multi innovation length constraint，a sub-iteration updating multi innovation gradient identification algorithm was proposed.By using the sub-decomposition technology，the identification model with parameter separation was constructed.To facilitate the parameter identification，a determinant ratio algorithm was designed to obtain the system degree information.Based on the multiple iterative updating theory，the given multi innovation was decomposed into sub innovation，in which the identification performance was improved.The convergence performance of the proposed scheme was strictly analyzed theoretically by using the martingale difference convergence theorem.Finally，the advantage of the proposed algorithm was validated by comparing existing identification schemes.

Key words:parameter identification;multi-innovation idea;reference model;Hammerstein system;sub-iteration updating

Hammerstein系统...迭代更新多新息梯度估计方案_鲁西坤.pdf