Bolted spherical joint connection is one of the common joint connections in space structures and it has the advantages of simple installation,fast installation and short construction period.Joints are the places where the stress of space structures concentrated.Once there are failures in these joints,the structure will lose its bearing capacity locally or even collapse.On the other hand,bolted spherical joint connection has some problems in its application.For example: in designing stage,bolted spherical joints are often considered as articulated joints,and it is actually a "semi-rigid joint connection",in which case reduces the reliability of the design results;In installation stage,there might be some problems such as false twisting and insufficient screwing depth of high-strength bolts.In the service stage,it may have some problems,such as the loss of high-strength bolt,bolt loosening,sleeve loosening and so on.In monitoring stage,bolts connect bars and bolt-sphere from inside of the joint,and it is difficult to inspect and monitor them outside.These problems lead to a fact that the probability of space structure accidents caused by joint damage is higher than that of bar member damage caused.However,at present,most research on damage identification of space structures focused on the identification of bar damage,and less research has focusd on damage identification of bolted spherical joint.This paper proposes a damage identification method of bolted spherical joint connection in space structures based on piezoelectric impedance technique and neural network.Firstly,the relationship between impedance index of the piezoelectric impedance technology and structural damage is analyzed theoretically.Then,the correlation between the impedance index and the tightness degrees of bolted spherical joint connect is verified by numerical simulation.Finally,a bolted spherical joint connection damage test is conducted on a triangle-on-triangle offset grid model composed of bolted spherical joints.The proposed method can identify the occurance,the location and the degree of damage successfully.The specific research contents are as follows:(1)Based on the piezoelectric impedance theory,the admittance equation of the piezoelectric ceramic(PZT)patch which coupled the structure system is deduced;the admittance equation of the PZT patch is deduced by analyzing the piezoelectric equation of the PZT which coupled the SMD system,the mechanical impedance of the structure and the mechanical impedance of PZT patch.From the equation,it can beknown that the change of the electrical impedance signal measured by the PZT sensor can directly reflect the change of the mechanical impedance of the structure,that is,the structural damage.(2)A damage identification method of bolted spherical joint connection in space structures based on piezoelectric impedance technique and neural network is proposed.Firstly,the finite element model of bolted spherical joint connection is built,and the impedance data of spherical joint under different tightening degrees are extracted.The correlation between the impedance index(admittance)and the tightness of joints is proved,thus proving the feasibility of the proposed damage identification method.Then the steps of damage identification method are introduced in detail.(1)The occurrence of damage is determined by the change of admittance spectrum;(2)The root mean square deviation(RMSD)of admittance subband is defined as the damage index to determine the damage location;(3)The damage degree is determined based on BP neural network.This method can realize on-line real-time monitoring of structures,and can be used to identify initial or minor structural damage.It does not depend on the accuracy of finite element modeling of structures.It has the advantages of simplicity,high local sensitivity,high identification accuracy and reliable results.It is especially suitable for damage monitoring of bolted spherical joint connection in space structures.(3)Damage experiment is designed for a triangle-on-triangle offset grid model.The results show that the proposed damage identification method can identify the occurance,the location and the degree of damage successfully.The details are as follows:(1)The PZT patch is bonded to the structure and excited in the range of 100kHz-300 kHz and the admittance spectra of joint before and after damage are extracted.The results show that the peak value of the admittance and the frequency of the peak value changed.This indicates the occurrence of damage.(2)Indentify the damage location based on piezoelectric impedance technology.12 damage cases are designed according to the different damage location;Part of the cases are selected and the PZT patch is excited.Base on the sub-band RMSD value of the extracted admittance curve,a range of 200 kHz-220 kHz is selected ultimately as the impedance sensitive frequency band.For 12 damage location cases(GK1-GK12),thePZT is excited in the selcted sensitive band and the admittance spectra are extracted.The RMSD value of each sub-band are analyzed and compared to determine the damage location.The results show that the PZT patch attached to the bar can monitor the occurrence of damage,and the damage at other locations have only a limited effect on the results.That is,the damage location of bolted spherical joint connection can be identified by RMSD.(3)The damage degree of bolted spherical joint is identified based on BP neural network.Eight kinds of tightness degrees are designed for joint connection and each of them is tested 20 times after which the admittance data are extracted.The BP neural network is trained by some of test data which is selected arbitrarily,and the other data is tested by BP neural network.The results show that the BP neural network identified the damage degree of all test samples successfully.The damage identification method of bolted spherical joint in space structures based on BP neural network can accurately identify the tightness degree of bolted spherical joint.This method is simple,feasible,and of high precision and accuracy. |