Font Size: a A A

Time Domain Identification Algorithm For Structural Dynamic Load And Research On High Accuracy And Stability

Posted on:2019-02-07Degree:DoctorType:Dissertation
Country:ChinaCandidate:J XuFull Text:PDF
GTID:1360330590966590Subject:Engineering Mechanics
Abstract/Summary:PDF Full Text Request
With the rapid development of science and technology and the increasing requirements in engineering design,the problems of the dynamic loads acting on engineering structures attracts more and more concerns.In many practical situations,due to the restrictions of techniques and economics,it is difficult to perform direct measurements or calculations of the dynamic loads.In such cases,estimating those dynamic loads indirectly by using measured vibration responses and system dynamic characteristics is sometimes necessary,which is known as load identification technology.As the second type inverse problem in structural dynamics,dynamic load identification uses response information and system model to reconstruct exciting force.By the effort of the experts and researchers domestic and overseas,great development of dynamic load identification has been obtained.However,a lot of problems of this technique still remain.Such as low identification accuracy,insufficient research in distributed loads and time-varying systems.In this thesis,identification accuracy problem of the dynamic load identification is studied.A new method based on numerical correction algorithm has been proposed.Using this method studies of different systems and different loads are carried out.And the specific research works are as follows.Based on the principle of function-root searching of the numerical algorithm,the theory of load identification method of the numerical correction algorithm is derived by combining the dichotomy method and the golden section method.In this paper,two methods of the Wilson ? inverse method and the quasi-static method are derived to obtain the initial value of the load.And the theoretical analysis is carried out to show the reason of inaccuracy of the identified load.The numerical correction algorithm is used to iteratively modify the initial load and the basic theoretical calculation system is obtained.For load identification problems of single input,theoretical derivation of the numerical correction algorithm for discrete systems and continuous systems is carried out respectively.The systems mainly include the spring mass system of single-degree of freedom,the spring mass system of multi-degree of freedom and one dimensional Euler-Bernoulli beam model.A large number of simulations are performed to verify the feasibility and the anti-noise performance of the proposed method.The experimental test on the simply supported beam using sinusoidal exciting force is used to verify the load identification capability for the actual structure.For load identification problems of multiple-inputs,using the principle of Gaussian elimination method to modify the numerical correction algorithm.A function relationship is established between one exciting force and acceleration response in order to adapt to the multiple-inputs situation.A large number of simulations are performed to verify the feasibility and the anti-noise performance of the proposed method under multiple input situation.The experimental test on the free beam using two sinusoidal exciting forces is used to verify the load identification capability for the actual structure.Change identified objects from centralized load to distributed load.First the initial value of the distribution load of the beam is reconstructed by using the generalized orthogonal polynomial method.Then one-dimensional orthogonal polynomial method is used to fit the distributed load.Combined with the modal superposition calculation method,the numerical correction method is modified.A function relationship is established between one series coefficient of orthogonal polynomials and acceleration response in order to adapt to get the series coefficient after correction.simulations and experimental tests are performed to verify the feasibility of the proposed method.For time-varying systems,the numerical correction algorithm for the time-invariant system is improved.Different time varying conditions are set up for the simple continuous beam system to verify the recognition results of the algorithm,including the overall and local mass,stiffness and damping in both linear and nonlinear time-varying conditions.The finite element model of GARTEUR plane is established,and the model parameters of the plane is adjusted using the modal test.Then two different time-varying circumstances are set up,the simulation results indicate that the numerical correction algorithm for complex time-varying load identification problem has good applicability.This paper also carried out some corresponding researches of the key problems for load identification.First of all for special structure with close modes,the problems of frequency omission and mode of vibration confusing are easy to happen.The paper using GARTEUR model as an example,and the identification capability of the numerical correction algorithm is verified.Then,the simulations is carried out to analyze the recognition performance of the algorithm in the special case in which the excitation frequency is close to the system's resonance point and anti-resonance point.In addition,for the problem of inaccuracy model during load identification process,quality,rigidity,damping and some other parameters are varied to check the adaptation of the numerical correction algorithm.As for numerical correction algorithm,its theoretical system contains three important parameters,they are ‘the influence of interval amplification coefficient',‘the interval partition coefficient' and ‘the accuracy index'.In this paper,deep analysis of these parameters on the calculation efficiency and accuracy of the results is carried out from the point of view of mathematics.
Keywords/Search Tags:Load identification, time domain, numerical correction, distributed load, time-varying system, identification accuracy
PDF Full Text Request
Related items