Research On Vibration Signal Inversion Based Turnout Structure Damage Identification

Turnout is an important part of the railway track equipment,which can ensure the high-density operation of the trains and also improve the passing capacity of railway.Due to the particularity of its work environment,the turnout is prone to damage.The damages above make turnout's structure performance decline,and seriously affect the security of the train operation.Therefore,the study on damage detection theory method of turnout has great value and significance.Once the structure is damaged,it will inevitably lead to changes in its vibration characteristics.By observing the changes of the vibration characteristic parameters,the structural working state can be evaluated.The structural vibration characteristic parameters include the natural frequency,mode shape and frequency response function.In view of this,this thesis studies the damage identification method of turnout based on vibration.In addition,due to the unpredictable excitation signal when trains pass through the turnout,in order to ensure the authenticity and reliability of frequency response function,a vibration signal inversion method is also studied in this thesis.The main works completed in this thesis include the following three aspects:(1)Under the condition that the load is not known,by comparing the effect of various indicators for crack damage recognition,an evaluation index for the crack damage which called the relevant strain modal change rate is proposed based on the modal strain.Simulate and the full-scale model tests show that this index has a good recognition effect on cracks.(2)Combined with dynamic programming theory,a time domain load solution method based on inversion of vibration signal is proposed.Bellman's principle was applied to the optimization of error function so that each step solution can guarantee the minimum of error function and effectively reduce the error accumulation in the inversion process.At the same time,the ill-posed problem of inversion of the coefficient matrix in the process of inversion is suppressed.The optimal regular parameter is determined by the L-curve method,which makes the inversion load closer to the actual value.Examples and experiments show that the inversion result is better,and the method has good anti-noise performance.(3)Based on the measured frequency response function of the structure,a damage identification method of turnout structure based on multivariate control map and Mahalanobis distance is designed.Firstly,in order to avoid the error by the estimation of the frequency response function for damage identification of turnout,the real frequency response function is calculated by inverting the load signal and then its quality is evaluated.Secondly,the principal component analysis method is used to solve the matrix problem that the frequency response function matrix has a higher dimension.Finally,combined with multivariate control map and Mahalanobis distance theory,the damage identification process of turnout structure was studied.Measurement data and test results show that this method can effectively locate the damage and preliminary determine the degree of damage.Finally,the main research content of this thesis is summarized,and the future research direction is expected.