NON-LINEAR SYSTEM IDENTIFICATION USING LUMPED PARAMETER MODELS WITH EMBEDDED FEEDFORWARD NEURAL NETWORKS

2002 ◽  
Vol 16 (2-3) ◽  
pp. 357-372 ◽  
Author(s):  
YIMIN FAN ◽  
C. JAMES LI
Keyword(s):  
Neural Networks ◽  
System Identification ◽  
Linear System ◽  
Feedforward Neural Networks ◽  
Lumped Parameter ◽  
Lumped Parameter Models ◽  
Non Linear ◽  
Linear System Identification ◽  
Non Linear System

Sensitivity of NL-RDM to Errors in the Non-Linear Test Process

2009 ◽  
Author(s):  
Mehdi Sarmast ◽  
Saeed Bostan Manesh M. ◽  
Mahmood R. Mehran
Keyword(s):  
System Identification ◽  
Linear System ◽  
Continuous Systems ◽  
Negative Effects ◽  
Practical Applications ◽  
Lumped Parameter ◽  
Non Linear ◽  
Linear System Identification ◽  
Non Linear System ◽  
Test Process

NL-RDM is a non-linear system identification method that combines a number of linear and non-linear system identification methods and offers a practical approach to the identification of lumped parameter and continuous systems using a classical linear modal model with additional non-linear terms. The method was started by identifying the modal parameters of the underlying linear system via the FRF, MMIF and appropriated force vector. The criteria for an ideal method are detailed in the some earlier papers, but the reality creates a limitation. This paper is divided into several sections relating to the “Nonlinear Test Process”. Error which arise from test, environmental and equipment effects, are quantization errors, input (or process) noise and measurement noise. So, the effects of these inaccuracies and possible solutions for decreasing any negative effects are considered. Then, The sensitivities to noise and quantization which could be encountered in practical applications of the NL-RDM, are discussed in concept, generated, applied and analyzed through simulation programme for two degree of freedom uncoupled and coupled examples.

Sign in / Sign up

Export Citation Format

Share Document