Szczegóły publikacji
Opis bibliograficzny
Hybrid AI system based on ART neural network and Mixture of Gaussians modules with application to intelligent monitoring of the wind turbine / Andrzej BIELECKI, Mateusz Wójcik // Applied Soft Computing ; ISSN 1568-4946. — 2021 — vol. 108 art. no. 107400, s. 1–12. — Bibliogr. s. 11–12, Abstr. — Publikacja dostępna online od: 2021-04-18
Autorzy (2)
- AGHBielecki Andrzej
- Wójcik Mateusz
Słowa kluczowe
Dane bibliometryczne
| ID BaDAP | 134089 |
|---|---|
| Data dodania do BaDAP | 2021-06-02 |
| Tekst źródłowy | URL |
| DOI | 10.1016/j.asoc.2021.107400 |
| Rok publikacji | 2021 |
| Typ publikacji | artykuł w czasopiśmie |
| Otwarty dostęp |  |
| Creative Commons | |
| Czasopismo/seria | Applied Soft Computing |
Abstract
Lack of effective intelligent monitoring systems of big wind turbines is a crucial problem in their exploitation. The used systems are based on direct monitoring of the system parameters by using programmable electronics or on the simple artificial intelligent systems, such as a single neural network. The efficiency of such systems is very limited. In the paper an innovative system for intelligent monitoring of wind turbines is proposed. The system is fully automated. It is based on Adaptive Resonance neural network (ART network) combined with two distinct modules based on Mixture of Gaussians. The proposed system is applied to detection of fault and pre-fault states of wind turbines. The turbine operational state data and data from vibration channels are the system input. First, the system specifies clusters that correspond to the healthy states of the turbine. Then, failure states are detected on-line as soon as they occur because the creation of a new cluster is signalized immediately by the ART module. The new cluster, that represents a failure or a pre-failure state is encoded in the Gaussian module which plays the role of the long-term memory. The system effectiveness was verified by using real data from a wind turbine farm.
