The Bolt Looseness Of CFRP Lap Plate Recognition Method Research Based On The Modal Analysis And RBF Neural Network

Carbon fiber reinforced polymer?abbreviated as CFRP?is one of the most valuedadvanced composite materials at present by virtue of“stronger than steel and lighter than aluminum”.In practical engineering application,bolted joints have been extensively applied to CFRP connection because of large bearing capacity and convenient disassembly/assembly,and the bolted connection status has a direct bearing on structural operating performance.Therefore,it appears to be particularly important to do health monitoring and early recognition of bolted connection status.Dynamic detection method based on modal analysis is a kind of global detection method which can realize global detection through single-point measurement,which is a remarkable advantage for connection status detection of multi-bolt connecting structures like CFRP.The influence characteristics of bolt pre-tightening force on inherent frequency were investigated in this paper by establishing a coupling stiffness model of bolt joint surface and a high-precision CFRP finite element analytical model.RBF artificial neural network algorithm was combined to realize positioning of loosening bolts and quantitative identification of pre-tightening force,and the main research work completed in this paper included:?1?Dynamic modeling and theoretical analysis of coupling stiffness characteristics at bolt joint surface.Based on the fractal theory of contact stiffness at bolt joint surface,the action laws of bolt pre-tightening force on contact stiffness at bolt joint surface of CFRP material were analytically calculated,followed by a further discussion on virtual material layer model-based modeling method for coupling stiffness at bolt joint surface.?2?Analysis of influence characteristics of bolt pre-tightening force on parameters of virtual material layer model.The action laws of bolt pre-tightening force on geometrical parameters and physical parameters of the virtual material layer model were discussed,the functional relationships of bolt pre-tightening force with elasticity modulus and Poisson's ratio of the virtual material were respectively fitted,so as to provide a coupling stiffness model for the subsequent finite element modal analysis of bolt pre-tightening force based on the virtual material layer.?3?High-precision CFRP finite element modeling based on model correction.In order to guarantee the accuracy of the follow-up finite element analysis result of CFRP lap plate,a CFRP finite element model correction method based on correlation analysis and approximate model was proposed.In comparison with the initial model,the maximum error and average error of the corrected model were 1.47%and 0.45%,respectively,so the CFRP finite element modeling precision was considerably improved.?4?High-precision CFRP finite element modeling based on model correction.In order to guarantee the accuracy of the follow-up finite element analysis result of CFRP lap plate,a CFRP finite element model correction method based on correlation analysis and approximate model was proposed.In comparison with the initial model,the maximum error and average error of the corrected model were 1.47%and 0.45%,respectively,so the CFRP finite element modeling precision was considerably improved.?5?Recognition research of loosening bolt based on RBF neural network.With the RBF artificial neural network,SRCF_i,j and FCR_i were taken as positioning and quantitative recognition indexes of loosening bolt,respectively.It was verified that the RBF neural network could effectively identify the position and degree of loosening bolt with a certain anti-noise jamming capability.