Thanks again.Identification Algorithm for Parameter-Varying System?

Click For Summary
SUMMARY

The discussion focuses on identifying parameters in a parameter-varying system using differential and algebraic equations. The user has input and output measurements and is attempting to identify 24 parameters derived from 8 expressions, each containing 3 parameters. Previous attempts with algorithms such as least squares, gradient descent, and Newton-Raphson have been unsuccessful. The user seeks advice on effective identification algorithms suitable for their specific engineering problem.

PREREQUISITES
  • Understanding of parameter-varying systems
  • Familiarity with differential and algebraic equations
  • Knowledge of parameter identification algorithms
  • Experience with optimization techniques such as least squares and gradient descent
NEXT STEPS
  • Research advanced parameter identification techniques for time-varying systems
  • Explore the use of Kalman filters for real-time parameter estimation
  • Investigate the application of genetic algorithms in parameter optimization
  • Learn about system sensitivity analysis to improve parameter estimation
USEFUL FOR

Engineers and researchers working on system identification, particularly those dealing with parameter-varying systems and optimization challenges in modeling and simulation.

FaroukSchw
Messages
6
Reaction score
0
Hello everybody
I started recently working on an engineering problem that requires as a first step, the identification of the system.
I already have a model which is a set of differential and algerbraic equations. Unfortunatelly, the system is a parameter varying one, and the aime is to identify these parameters. Taking in consideration that i have input and output measurements, and about 20 parameter to identify.
Does anyone have an idea about an identification algorithm that can be used for such problem?

Regards.
 
Engineering news on Phys.org
It really doesn't make much sense to estimate the parameters of a time-varying system, unless you have the model for the time dependence of the parameters as well (periodicity, explicit functional form?) or you have some specific real-time application that I can't think off right now. Anyway, I'd really appreciate it if you'd share your models and tell us what you're trying to do
 
Hi elibj123

Thanks for taking time and responding to my thread.

In fact, i do have the expressions of the parameters, but since the model has 8 parameters, each parameter's expression contains three parameters, so 8*3=24 paameter to be identified.

The sensitivity of the model to the parameters varries from one parameter to an other.

I tried some algorithms such as different versions of the least squares, Gradient algo, Newton-Raphson, but they don't work!

Unfortunatelly, i can't share the model, because i don't have the right to, at least for the time beying.:frown:

I will be verry glad if you can provide some advices.
 

Similar threads

S11 and Frequency Modeling for Physical Parameter
  • · Replies 8 ·
Replies
8
Views
3K
Maxwell Linear Generator (Input Power < Output Power and losses?)
  • · Replies 2 ·
Replies
2
Views
1K
System Parameter Estimation with Projection Algorithm
  • · Replies 1 ·
Replies
1
Views
1K
Measuring Power output of DC generator
  • · Replies 38 ·
2
Replies
38
Views
5K
Identifying a Non-Linear Parameter Varying System
  • · Replies 1 ·
Replies
1
Views
1K
Python MCMC algorithm -- understanding some paremeters
  • · Replies 4 ·
Replies
4
Views
2K
Undergrad Gradient descent algorithm and optimizers
  • · Replies 5 ·
Replies
5
Views
2K
System Identification: Finding Transfer Function w/ Graphical Methods
  • · Replies 2 ·
Replies
2
Views
3K
Undergrad Algorithmic model for primary decay chains
  • · Replies 1 ·
Replies
1
Views
2K
Automotive Estimation of the damping coefficient of a suspenion
  • · Replies 10 ·
Replies
10
Views
2K