Research On Regularization Method Of Dynamic Load Identification Under High Level Noise

In order to meet the safety criteria of engineering,dynamic load identification has been concerned as a practical subject,and the regularization method is widely used at present.When the noise level is low,the result of the regularization method is ideal,otherwise it is not ideal when the noise level is high.This paper studies the weaknesses of the regularization method,and an improved method can be applied to dynamic load identification problems under high-level noise conditions is proposed.This method solves the problem of dynamic load identification in actual engineering.The specific research is as follows:First,the development and current situation of dynamic load identification problems are introduced,and in view of the problems existing in the theory of dynamic load identification,the main research content of this paper is the dynamic load identification based on regularization method under high level noise.Secondly,the Tikhonov regularization method and the integrated moving average method with good noise resistance are introduced.The advantages and disadvantages of several commonly used optimal regularization parameter selection methods are discussed.Numerical simulation examples verify that the regularization method based on the integral moving average method can effectively overcome the impact of high-level noise.Thirdly,Based on the improvement of the quotient function method in terms of algorithm,an adaptive quotient function method for selecting the optimal regularization parameters is proposed.Compared with the quotient function method,the adaptive quotient function method has the advantage of simple algorithm and can be obtained more than the traditional Dynamic load identification results with better method accuracy.Finally,for the non-Gaussian noise that is uniformly distributed,a L_? norm regularization method for uniform noise is introduced.Based on the sinusoidal series curve fitting technique,a curve model that can effectively solve the problem of dynamic load recognition under high-level uniform noise is proposed.Synthetic norm regularization method.The curve fitting regularization method has high guiding significance for the problem of dynamic load identification based on telemetry data.