Application Research Of Optimization Algorithms In Indirect Structural Testing Techniques

Many important mechanical parameters in structural mechanics test cannot be obtained directly,so it is necessary to use indirect testing techniques.Load identification is the common problems of indirect testing.It is an important issue to solve ill-posed problems with indirect testing techniques.However,the most frequently used identified algorithms,like least square method,have very poor effect in highly ill-posed problems.On the contrary,optimization algorithms can deal with this kind of engineering problems better.Optimization algorithms can divide into two categories: regularization algorithms and intelligent optimization algorithms.Applicability of two optimization algorithms for different ill-posed problems is studied in this paper.In engineering application,suitable optimization algorithms can be selected by means of condition number of coefficient matrix.In this situation,it can reduce computation and improve computational efficiency.Advantages and disadvantages of regularization algorithms and intelligent optimization algorithms are discussed through analyzing the results of optimization algorithms and two load identification examples.The main contents of this paper are as follow:Firstly,two regularization algorithms and parameter choice methods are studied.Three new regularization algorithms are built by combining regularization algorithms and parameter.Three regularization algorithms are tested and applicability is discussed under different ill-posed problems by coefficient matrices with different condition numbers.Secondly,two basic intelligent optimization algorithms' theories about genetic algorithm and particle swarm optimization algorithm are studied.Three improved intelligent optimization algorithms are proposed by using chaotic local search algorithm.At last,applicability by coefficient matrices with different condition numbers is compared and discussed.Thirdly,the results of overall comparison of optimization algorithms are analyzed.Suitable optimization algorithms in intelligent optimization algorithms and regularization algorithms are selected,based on the former analysis and applied them to static load identification and dynamic load identification,respectively.Advantages and disadvantages of regularization algorithms are evaluated and intelligent optimization algorithms are used in the practical application after combing the results of overall comparison of optimization algorithms and two load identification examples.