Optimal Placement Of Sensors And Damage Identification For Space Structure Considering Randomness

Health monitoring for space structure has been deeply studied in recent years.Because of the properties of materials and construction errors, spatial structure hasinitial random defects, which induce the structural geometrical features, mechanism ofdamage evolution and capacity of bearing weight become different from the modeldesigned originally, the monitoring scheme based on the designed model doesn’tapply to practical structure any more, therefore, we should study methods of monitorlayout optimization and damage identification on the basis of taking randomness ofstructure character into account.The effect of different minor damage cases on the total space structure is studiedbased on vulnerability theory in structural analysis. The concept of generalizedequivalent stiffness is introduced and the important factor of component is defined.For numerical simulation, the random character both for structural parameters andloads is considered, and the random samples are established. The fragility failure pathof each sample is calculated and all the important members on the fragility failurepath are listed, therefore the sensor placement scheme is determined according to thestatistic data. This method is extended to dynamic analysis. For every dynamictime-history analysis, time-varying responses of the structure are calculated byselecting appropriate calculating interval and considering the randomness of structuralparameters and load. The time-varying response is analyzed and the important factorof members is sorted, finally the dynamic sensor placement scheme is determined.The effectiveness of the method in this paper is certified by example.Traditional monitor layout optimization methods usually result in non-idealmodel identification consequences due to limitations of optimization principles itself.This text applies Zernike matrix to the expression of vibration modes of spacestructure considering Zernike matrix’s advantages, so that more modal information isretained. As a result, the difference of Zernike matrix made from modals on differentstages is more remarkable. The Iaccard coefficient is introduced to serve asoptimization index, firstly get the Jaccard similarity coefficient matrix with Zernikematrix amplitude vector of each vibration mode, then choose monitoring spotsaccording to the variation of the Jaccard matrix’s shape. Besides the method above, this text also uses the MAC norm to optimize the monitor layout, then compares thetwo methods by experiment to prove the first method’s superiority.Zernike matrix is also applied to the damage identification area of space structure,Zernike matrix amplitude vector serves as damage identification index. A largeamount of Zernike matrix amplitude vectors are obtained by simulating therandomness of the structure to train the random forest, which identifies structuredamages with high accuracy rate.