Neuro-fuzzy modelling and control of multistage dynamic processes that depend on inputs with uncertainty elements
| Parent link: | Journal of Theoretical and Applied Information Technology: Scientific Journal.— , 2005- Vol. 80, № 1.— 2015.— [P. 1-12] |
|---|---|
| Korporativní autor: | |
| Další autoři: | , , , |
| Shrnutí: | Title screen In practice, we meet with a frequent necessity for modelling dynamic multistage processes that depend onseveral time-varying factors, which can be measured with accuracy. This requires using a model thatcombines properties of the difference, switched, and neuro-fuzzy models with a large number of inputs anda big memory depth of each input. Such model will take into account all uncertainties, converting inputactions into fuzzy processes by means of fuzzification, and will precisely reflect the multistage nature anddynamics of the object being studied. However, a large number of parameters can significantly complicateits configuration. The purpose of this work is to improve efficiency of structural and parametric modellingof multistage dynamic processes due to the development of a class of difference neuro-fuzzy switchedmodels, as well as to the research and development of approaches to fuzzification of discrete processes onmodel inputs in order to simplify the process of its configuration. The authors have introduced a new classof difference neuro-fuzzy switched models that are characterized by a combination of structures ofdifference fuzzy models, neural networks, and systems with switchings enabling to model complexmultistage processes, which are characterized by abrupt changes in structure or parameters. The authorshave proposed a mechanism for fuzzification of input actions of a difference neuro-fuzzy switched model,which is characterized by the ability to convert input actions into discrete fuzzy processes using twodimensionalfuzzy sets, and allows reducing the number of configurable parameters of the model. |
| Vydáno: |
2015
|
| Témata: | |
| On-line přístup: | http://www.jatit.org/volumes/Vol80No1/1Vol80No1.pdf |
| Médium: | Elektronický zdroj Kapitola |
| KOHA link: | https://koha.lib.tpu.ru/cgi-bin/koha/opac-detail.pl?biblionumber=645933 |
| Shrnutí: | Title screen In practice, we meet with a frequent necessity for modelling dynamic multistage processes that depend onseveral time-varying factors, which can be measured with accuracy. This requires using a model thatcombines properties of the difference, switched, and neuro-fuzzy models with a large number of inputs anda big memory depth of each input. Such model will take into account all uncertainties, converting inputactions into fuzzy processes by means of fuzzification, and will precisely reflect the multistage nature anddynamics of the object being studied. However, a large number of parameters can significantly complicateits configuration. The purpose of this work is to improve efficiency of structural and parametric modellingof multistage dynamic processes due to the development of a class of difference neuro-fuzzy switchedmodels, as well as to the research and development of approaches to fuzzification of discrete processes onmodel inputs in order to simplify the process of its configuration. The authors have introduced a new classof difference neuro-fuzzy switched models that are characterized by a combination of structures ofdifference fuzzy models, neural networks, and systems with switchings enabling to model complexmultistage processes, which are characterized by abrupt changes in structure or parameters. The authorshave proposed a mechanism for fuzzification of input actions of a difference neuro-fuzzy switched model,which is characterized by the ability to convert input actions into discrete fuzzy processes using twodimensionalfuzzy sets, and allows reducing the number of configurable parameters of the model. |
|---|