Font Size: a A A

Structural Reliability Analysis Based On Fourier Orthogonal Neural Network Improved Response Method

Posted on:2014-02-26Degree:MasterType:Thesis
Country:ChinaCandidate:J LiuFull Text:PDF
GTID:2232330395496470Subject:Engineering Mechanics
Abstract/Summary:PDF Full Text Request
With the development of the modern science and technology, engineering structures are applied more and more widely in the real world.Frequent accidents are caused by the structural failure, however, lots of uncertainty problems are in practical engineering structure. People begin research for the safety of the structure reliability theory.The analysis of engineering structure has very important guiding significance. However, the existing methods of the reliability analysis are not enough perfect, we need more precise and effective reliability analysis methods be applied to the engineering practice.Based on the structural reliability theory, combining the orthogonal basis、the basic theory of neural network and the response surface method, proposed the Fourier orthogonal neural network improved response method(shortened form FONN-IRSM).I use the method for the reliability analysis of automobile engine’s connecting rod, then compared with the existing reliability analysis method. The correctness and practicability of the proposed method is verified,so is its good approximation performance.The method of the reliability theory has certain reference value and practical significance.Firstly,This paper made full use of the good approximation properties of the Fourier orthogonal, as well as the neural network’s powerful nonlinear mapping ability and its good fault tolerance, combined with Fourier orthogonal basis theory and neural network theory proposed the Fourier orthogonal basis neural network model.The convergence of the model and its excellent approximation performance are verified by the numerical examples,provided a new algorithm for the reliability theory.Secondly, This paper combined the technology of gradient projection sampling point, improved the response surface method. Then this paper used the respective advantages of the Fourier orthogonal neural network and the improved response surface method,proposed FONN-IRSM.By the numerical examples and comparison with other methods of reliability analysis, verified the validity and its advantage of the proposed FONN-IRSM;Finally, as an example of the automobile engine’s connecting rod,combined with the basic theory of the fatigue reliability, on the one hand, we used the softwares such as Ug、 Ansys and Adams for the finite element analysis of connecting rod,found the biggest danger point of the connecting rod;On the other hand, used the FONN-IRSM do the fatigue reliability analysis of the dangerous point. used the Matlab language to program and calculate fatigue safety index of its dangerous point, verified the correctness and practicable of the FONN-IRSM. This method has a good reference value for the structure reliability analysis, and provides a relatively reliable method of the reliability analysis.
Keywords/Search Tags:Fourier orthogonal neural network improved response method, structuralreliability, reliable indicator, fatigue
PDF Full Text Request
Related items