Research On The Load Identification Of Flexible Bodies

Due to the trend toward large-scale and precision of engineering structures, together withincreasing complicate load environments, it is gradually in need to develop dynamic loadidentification techniques. That’s because dynamic load identification techniques can help peopleto acquire external load condition in the environment, and assist in the structure design work andstrength analysis, which finally ensures safety and stability of targeted structures.Dynamic load can be obtained by the direct measurement methods and indirect identificationmethods. Direct measurement methods use sensors to obtain the loading condition by measuringthe load itself or load-related parameters. It is an easy and practical method to obtain the load forsimple structures. However, it is difficult to get load by direct measurement for the most ofengineering structures. Therefore, it is necessary to identify external loads through an indirectmethod, and that is dynamic load identification method.Derived from the aviation industry, dynamic load identification methods have developed quitea few theories and algorithms over recent forty years. In general, these methods could be dividedinto two categories: frequency domain methods and time domain methods. Frequency domainmethods imply that an inverse frequency response model is established and then be used to identifydynamic loads in frequency domain. The main idea of time domain methods is to set up a reversemodel of targeted system’s inputs and outputs for the dynamic load identification in the timedomain. The time domain methods are quite convenient for engineering applications, as well assave the possibility of online identification. Besides frequency domain methods and time domainmethods mentioned above, many new methods provide new tools for the load identification issue,with the development of signal processing and analysis, modal recognition and otherinterdisciplinary subjects.This dissertation studies on dynamic load identification of flexible beams, and is funded byNational Natural Science Foundation of China (No.11132001,11272202), Aeronautical ScienceFoundation of China (No.20120157002) and Innovation Program of Shanghai Municipal EducationCommission (No.14ZZ021). The main research contents and results are summarized as follows:(1) Based on the Euler-Bernoulli theory, a specific form of distribution load identificationtechnique is studied. This form of distribution load can be separated in space and time. Thisidentification method is based on a variable structure controller and an external load observer, inwhich the controller is used to restrain elastic vibration of targeted beam and the observer is used toobtain the external load in real-time. Numerical simulation results indicate that the proposedidentification method can exactly identify the special form of distribution load. (2) Based on the generalized Chebyshev orthogonal polynomials, the distributed loadidentification technique for the flexible beam is studied. Dynamic response of the system is givenby the Duhamel integral, and distributed load is expanded in series form using generalizedChebyshev orthogonal basis. A functional relationship between system response and load seriescoefficients is established to deduce dynamic loads through a reverse operation processing.Simulation results demonstrate the load identification method based on orthogonal polynomial iseffective.(3) The transverse vibration properties of the axially moving beam and identification of theexcitation power spectrum density (PSD) issues are studied. With the simple boundary conditions,the differential equation of axially moving beam is discretized by the complex analysis procedure,Galerkin method and Ritz method respectively. With the complex boundary conditions, finiteelement method can overcome difficulties brought by boundary conditions in the dynamicsmodeling. The finite element model of axially moving beam is deduced by Lagrange equations.Based on these discrete models, the vibration properties and identification of excitation PSD arestudied. Simulation results demonstrate the mentioned methods above can accurately identify theexcitation PSD with different boundaries.