Abstract
For a kind of coupled parameters multivariable system, the coupled multi-innovation stochastic gradient (CMISG) identification method is proposed to estimate the parameters, and its performance analysis is made. The basic idea of this method is utilizing the historical information to extend the scalar innovation item to an innovation vector to enhance the identified effect of each subsystem. Simulation results show that increasing the innovation length can enhance the convergence rate and accuracy of the identified results.
| Original language | English |
|---|---|
| Pages (from-to) | 1527-1530 |
| Number of pages | 4 |
| Journal | Kongzhi yu Juece/Control and Decision |
| Volume | 30 |
| Issue number | 8 |
| DOIs | |
| State | Published - 1 Aug 2015 |
Keywords
- Convergence analysis
- Coupled stochastic gradient
- Multi-innovation vector
- Multivariable system
- Parameter estimation