Moving regression artificial neural network framework for reliability evaluation of complex structure/system

For evaluating reliability level of complex structure/system, a moving regression artificial neural network (MRANN) framework is proposed based on artificial neural network (ANN), moving regression technique and intelligent optimization algorithm. Under this framework, four methods of ANN-based movi...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Reliability engineering & system safety Jg. 267; S. 111876
Hauptverfasser: Lu, Peng, Chen, Yong-Jing, Dong, Rui-Yi
Format: Journal Article
Sprache:Englisch
Veröffentlicht: Elsevier Ltd 01.03.2026
Schlagworte:
Artificial neural network
Complex structure/system
Intelligent optimization algorithm
Moving regression technique
Reliability evaluation
Moving regression technique
Complex structure/system
Reliability evaluation
Artificial neural network
Intelligent optimization algorithm
ISSN:0951-8320
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:For evaluating reliability level of complex structure/system, a moving regression artificial neural network (MRANN) framework is proposed based on artificial neural network (ANN), moving regression technique and intelligent optimization algorithm. Under this framework, four methods of ANN-based moving least squares (MLS) (MLS-ANN), enhanced ANN-based MLS (MLS-EANN), ANN-based improved MLS (IMLS-ANN) and enhanced ANN-based improved MLS (IMLS-EANN) models are developed by introducing the Ivy algorithm and MLS technique into the ANN model. Herein, the Ivy algorithm is used to find the optimal radius of compact support region for selecting effective modeling samples or to search the optimal values of weights and thresholds of ANN model, the MLS technique is applied to assign weights for these effective modeling samples and to resolve unknown coefficients of ANN model, and the ANN model is employed to reflect the relationship of structural response and input parameters. Furthermore, three examples, including multivariate function approximation and probabilistic analyses, aeroengine high-pressure turbine blisk deformation reliability assessment and aircraft flap system deflection angle reliability analysis, are performed to validate and verify the effectiveness of these developed approaches. The analytical results show that MRANN framework holds excellent ability in the modeling and simulation features relative to the existing surrogate models.
ISSN:0951-8320
DOI:10.1016/j.ress.2025.111876