Font Size: a A A

Online Neural Network Algorithm Researching And The Application Of The Hybrid Test

Posted on:2018-01-09Degree:MasterType:Thesis
Country:ChinaCandidate:X H ZhaiFull Text:PDF
GTID:2322330518975358Subject:Civil engineering
Abstract/Summary:PDF Full Text Request
Hybrid test is in recent years new combining numerical simulation and the real load test of structure seismic test method,the related researchers' attention.Hybrid experiment,the key parts in the structure as a general test substructure test,numerical simulation was made for other parts as numerical substructure.To simulate a strongly nonlinear numerical substructure,this paper puts forward a on-line neural network algorithm is used to on-line identification restoring force model,and simulated numerical verification of the structure and the buckling of support components test,finally put forward the hybrid test method based on on-line neural network algorithm.The concrete research content is as follows:(1)In order to be able to predict online nonlinear restoring force structure,an online algorithm for BP neural network is put forward.For two degrees of freedom nonlinear structures,this paper has carried on the Bouc-Wen model of restoring force,to verify the effectiveness of online neural network algorithm.Analyses the sample size and target error of the prediction accuracy and computational time consuming.Research shows that: compared with the traditional off-line learning algorithm,the online BP neural network improves the computing efficiency and prediction accuracy.(2)Based on the recursive form of online Elman neural network,the buckling support component test,research shows that: the Elman neural network algorithm based on recursive form has good real-time and dynamic,can effectively improve the prediction accuracy of restoring force.(3)Hybrid test method based on on-line neural network algorithm,through the two degrees of freedom nonlinear structure is obtained by using hybrid simulation verified the effectiveness of the method.Experimental results show that,compared with BP network algorithm online,online Elman algorithm has better adaptability,and can effectively improve the predictive accuracy and computational efficiency restoring force;Online Elman algorithm based on the structure of the hybrid test method can improve the accuracy of test results.
Keywords/Search Tags:hybridtesting method, neural network, online prediction, restoring force
PDF Full Text Request
Related items