Product Information
- Author
- Herausgeber FKM
- EAN
- 4250697510450
- Edition
- 2002
- Delivery time
- next business day
Lebensdauerabschätzung schwingend beanspruchter Bauteile mittels Künstlicher Neuronaler Netze
120.00 EUR *
Gesamtpreis: 120.00 EUR *
Prices incl. VAT
112.15 EUR excl. VAT
available
Description
Lebensdauerabschätzung schwingend beanspruchter Bauteile mittels Künstlicher Neuronaler Netze
FKM 2002
Booklet No. 274
Project No. 246
Abstract:
In the technical regulations, the service life is preferably calculated using the nominal stress concept. In this form of service life estimation, the load collective and the component stress line are compared. The service life is usually calculated using linear damage accumulation (Palmgren-Miner), whereby various modifications can be used. The comparison between calculation and test shows a large scattering of the damage sums. Through a targeted evaluation of a large number of fatigue strength horizons, the prerequisites for an improved relative miner rule were developed. The improvement potential of a service life estimation when using the method lies in the use of artificial neural networks (ANN), which are able to solve complex, multi-dimensional, non-linear c in the trained state through an iterative learning process. The successful applicability of artificial neural networks in the determination of synthetic Wöhler curves has been demonstrated in previous research projects. Artificial neural networks were developed that are capable of estimating the service life based on characteristic values for the description of the Wöhler curve and the amplitude-transformed or the derived range pair collective. In comparison with the results from a linear analysis, the ANNs proved to be superior in terms of prediction accuracy. It is not possible to estimate the service life in the area of high numbers of cycles based on experimental results in the literature. It is not possible to map the influence of the different input values. It was demonstrated that the ANN is also capable of correctly reproducing the general values. trained ANNs were implemented in the spreadsheet .NeuroLeben of the Excel 2000@ spreadsheet software and can therefore be used on any standard PC. The software required for training the ANN is not necessary.
Chairman of the working group:
Head of the FK Chairman of the advisory board:
90 p., 68 ill. and tab., 87 lit. 01.03.2001 28.02.2003 Funding body:
BMWI/AiF-Nr. 12750 N Institute for Mechanical Plant Engineering and Structural Durability at ClausthallMAB Technical University, Clausthal Prof. Dr.-lng. H. Zenner Project management:
Dipl.-lng. C. Marquardt Dipl.-lng. C. Marquardt Dr.-lng. M. Brune, Bayerische Motorenwerke AG Dr. C. Gerdes, Alstom Schweiz AG Dr.-lng. E.h. J. Rabe, Höchstadt
Booklet No. 274
Project No. 246
Abstract:
In the technical regulations, the service life is preferably calculated using the nominal stress concept. In this form of service life estimation, the load collective and the component stress line are compared. The service life is usually calculated using linear damage accumulation (Palmgren-Miner), whereby various modifications can be used. The comparison between calculation and test shows a large scattering of the damage sums. Through a targeted evaluation of a large number of fatigue strength horizons, the prerequisites for an improved relative miner rule were developed. The improvement potential of a service life estimation when using the method lies in the use of artificial neural networks (ANN), which are able to solve complex, multi-dimensional, non-linear c in the trained state through an iterative learning process. The successful applicability of artificial neural networks in the determination of synthetic Wöhler curves has been demonstrated in previous research projects. Artificial neural networks were developed that are capable of estimating the service life based on characteristic values for the description of the Wöhler curve and the amplitude-transformed or the derived range pair collective. In comparison with the results from a linear analysis, the ANNs proved to be superior in terms of prediction accuracy. It is not possible to estimate the service life in the area of high numbers of cycles based on experimental results in the literature. It is not possible to map the influence of the different input values. It was demonstrated that the ANN is also capable of correctly reproducing the general values. trained ANNs were implemented in the spreadsheet .NeuroLeben of the Excel 2000@ spreadsheet software and can therefore be used on any standard PC. The software required for training the ANN is not necessary.
Chairman of the working group:
Head of the FK Chairman of the advisory board:
90 p., 68 ill. and tab., 87 lit. 01.03.2001 28.02.2003 Funding body:
BMWI/AiF-Nr. 12750 N Institute for Mechanical Plant Engineering and Structural Durability at ClausthallMAB Technical University, Clausthal Prof. Dr.-lng. H. Zenner Project management:
Dipl.-lng. C. Marquardt Dipl.-lng. C. Marquardt Dr.-lng. M. Brune, Bayerische Motorenwerke AG Dr. C. Gerdes, Alstom Schweiz AG Dr.-lng. E.h. J. Rabe, Höchstadt
Product link copied to the clipboard.
