Research On The Stiffness Of Frame Structure Damage Diagnosis Based On BP Neural Networks Of Theory Study

It is crucial that the health status of structural detect and today it is the foundation and centre of civil engineering in structure health detective, vibrate modal analysis technology and neural network technology become a effective instrument to solve this problem. On the basis of collection and analysis of papers about structural damage identification and artificial neural networks, with the help of software Structural Mechanics Solver 2.0, through a plane second floor frame structure stiffness single-damage and two-damage example, get 33120 sample elements of frequencies and the first frequency corresponding of vibration shape, after dressing by screening them , get 13980 sample elements, the data is trained by artificial neural networks as the relation of displacement of damage and degree of damage with the compartment data , after converging one settled. With the help of software Matlab7. 0 simulate, analysis outcome data and explanation clearly neural network technology could do some work at frame construction in the damage of position and the extent of damage, some conclusion is drawedPrime conclusion:1) it was feasible that use low order frequency and certain mode of vibration or their derived quantity incorporation came by way of structural damage position, extent of damage, as well as could orient the damage of position and the extent of damage, and proceed extent of damage identify , and the difficult problem of symmetrical structure position discriminating puzzle was also solved. The error precision can be controlled within 10%.2) The best number of neural network element in implication is2n^1/2+1( n for the number of input dimension vectors).3) some conclusion and method are drawed to solve how to choose vibration vector, just as follows:...