Finite Element Analysis And Model Updating Of Infrared Earth Sensors

Finite Element (FE) modeling and analysis has increasingly become an importantmeans to solve practical engineering problems, but its accuracy is often difficult to beguaranteed. So FE model updating and verification with the experimental data plays animportant part in FE analysis. This paper studies the two main methods for the FEmodel updating-direct matrix methods and indirect physical property adjustmentmethods, analyzes every feasible method and mainly discusses using the ResponseSurface Method (RSM) for FE model updating.This paper uses a specific model of infrared earth sensors as the research object,builds the corresponding FE model and implements modal analysis and harmonicresponse analysis. Based on the results of modal tests and sine sweep tests, the FEmodel of the infrared earth sensor is updating by using RSM. The updated results showthat the traditional RSM is feasible, but not quite effective, for a complex FE model,and it needs human intervention during the updating process.In order to solve the inefficiency of FE model updating based on RSM, aStep-by-Step Point Selection Method (SSPSM), a Maximum Weighted RegressionMethod (MWRM) and an Adaptive Design of Experiment (ADE) are proposed in thispaper. SSPSM rewrites a linear response surface function into a number of linearequations so as to adjust the design of experiment after every FE calculation. SSPSMaims at better interpreting the implicit information of data provided by FE analysis,faster and more accurately locating the experimental points, and thereby alleviating thecomputational burden. MWRM uses maximum weighting function to weight all theexperimental points. The experimental points weight greater if they are closer to thetarget point, which can effectively improve the fitting accuracy of the response surfacenear the target point. ADE determines a new center point and a new sampling radiusaccording to the average relative error of the characteristic parameters in order to ensurethe iterative calculation convergence and improve the fitting accuracy and convergencerate. A15-element simply supported beam is used as a numerical example, the results ofwhich verify the advantages compared with the traditional RSM and the feasibility ofthe above three improved methods. For practical engineering problems, SSPSM, the most effective method, is applied to update the FE model of the infrared earth sensor,during the process of which, a great progress in both the convergence speed and theupdating accuracy is observed.Using the Response Surface Method (RSM) for FE model updating is to obtain theexplicit relationship between characteristic parameters (such as natural frequencies andmode shapes) and design parameters (such as density and Young’s modulus) by theregression analysis method. The function of RSM uses the characteristic parameters asdependent variables and the design parameters as independent variables. This paperproposes an Inverse Response Surface Method (IRSM), which is constructed by usingthe design parameters as dependent variables and the characteristic parameters asindependent variables. The simply supported beam shows that using IRSM for FEmodel updating is feasible, that it is able to obtain the updated values directly, instead ofby iteration, based on the target value of the characteristic parameters, that it canefficiently improve the calculation speed and accuracy, and that it is suitable for thecomplex engineering structures.