Title | ||
---|---|---|
Schnelles Training neuronaler Netze zur Identifikation nichtlinearer dynamischer Systeme (Fast Training of Neural Networks for Nonlinear Dynamic System Identification) |
Abstract | ||
---|---|---|
The time consuming training of dynamic neural networks often prohobits the industrial application. This paper presents a fast method to train neural networks with external dynamics for nonlinear dynamic system identification. The proposed method selects first a suitable set of neurons and trains the resulting network in a second step. This proposed procedure leads to faster convergence compared to standard training algorithms and comparable accuracy. |
Year | DOI | Venue |
---|---|---|
2011 | 10.1524/auto.2011.0898 | AT-AUTOMATISIERUNGSTECHNIK |
Keywords | Field | DocType |
Recurrent neural networks,autocorrelated error,NARX,NOE | Nonlinear system,Computer science,Control engineering,Artificial intelligence,Artificial neural network,System identification | Journal |
Volume | Issue | ISSN |
59 | 2 | 0178-2312 |
Citations | PageRank | References |
0 | 0.34 | 4 |
Authors | ||
2 |
Name | Order | Citations | PageRank |
---|---|---|---|
Michael Deflorian | 1 | 1 | 1.04 |
Florian Klöpper | 2 | 0 | 0.68 |