Stability of discrete-time feed-forward neural networks in NARX configuration

被引：17

作者：

Bonassi, Fabio ^{[1
]}

Farina, Marcello ^{[1
]}

Scattolini, Riccardo ^{[1
]}

机构：

[1] Politecn Milan, Dipartimento Elettron Informaz & Bioingn, Via Ponzio 34-5, I-20133 Milan, Italy

来源：

IFAC PAPERSONLINE | 2021年 / 54卷 / 07期

关键词：

Neural networks; Nonlinear System Identification; Identification for Control; Input-to-State Stability; Incremental Input-to-State Stability; PREDICTIVE CONTROL; IDENTIFICATION; SYSTEMS;

D O I：

10.1016/j.ifacol.2021.08.417

中图分类号：

TP [自动化技术、计算机技术];

学科分类号：

0812 ;

摘要：

The idea of using Feed-Forward Neural Networks (FFNNs) as regression functions for Nonlinear AutoRegressive eXogenous (NARX) models, leading to models herein named Neural NARXs (NNARXs), has been quite popular in the early days of machine learning applied to nonlinear system identification, owing to their simple structure and ease of application to control design. Nonetheless, few theoretical results are available concerning the stability properties of these models. In this paper we address this problem, providing a sufficient condition under which NNARX models are guaranteed to enjoy the Input-to-State Stability (ISS) and the Incremental Input -to-State Stability (6ISS) properties. This condition, which is an inequality on the weights of the underlying FFNN, can be enforced during the training procedure to ensure the stability of the model. The proposed model, along with this stability condition, are tested on the pH neutralization process benchmark, showing satisfactory results. Copyright (C) 2021 The Authors.

引用

页码：547 / 552

页数：6

共 28 条

[1] Moving-horizon state estimation for nonlinear discrete-time systems: New stability results and approximation schemes [J].

Alessandri, Angelo ;

Baglietto, Marco ;

Battistelli, Giorgio .

AUTOMATICA, 2008, 44 (07) :1753-1765

[2] Artificial Intelligence techniques applied as estimator in chemical process systems - A literature survey [J].

Ali, Jarinah Mohd ;

Hussain, M. A. ;

Tade, Moses O. ;

Zhang, Jie .

EXPERT SYSTEMS WITH APPLICATIONS, 2015, 42 (14) :5915-5931

[3]

[Anonymous], 2017, An overview and comparative analysis of recurrent neural networks for short term load forecasting

[4] Model Predictive Control Design for Dynamical Systems Learned by Echo State Networks [J].

Armenio, Luca Bugliari ;

Terzi, Enrico ;

Farina, Marcello ;

Scattolini, Riccardo .

IEEE CONTROL SYSTEMS LETTERS, 2019, 3 (04) :1044-1049

[5] Identification and predictive control of a multistage evaporator [J].

Atuonwu, J. C. ;

Cao, Y. ;

Rangaiah, G. P. ;

Tade, M. O. .

CONTROL ENGINEERING PRACTICE, 2010, 18 (12) :1418-1428

[6]

Bayer F, 2013, 2013 EUROPEAN CONTROL CONFERENCE (ECC), P2068

[7]

Bonassi F., 2020, ARXIV PREPRINT ARXIV

[8] Nonlinear MPC for Offset-Free Tracking of systems learned by GRU Neural Networks [J].

Bonassi, Fabio ;

da Silva, Caio Fabio Oliveira ;

Scattolini, Riccardo .

IFAC PAPERSONLINE, 2021, 54 (14) :54-59

[9]

Bonassi F, 2020, PR MACH LEARN RES, V120, P85

[10] On the discrete-time normal form [J].

Califano, C ;

Monaco, S ;

Normand-Cyrot, D .

IEEE TRANSACTIONS ON AUTOMATIC CONTROL, 1998, 43 (11) :1654-1658

← 1 2 3 →