Structural Health Monitoring Using Mahalanobis-Taguchi System

In order to monitor the structural health condition in the case of damage samples shortage, a new approach is proposed by only using the healthy samples of vibration signals. By taking root mean square (RMS), variance, kurtosis and crest factor as characteristic parameters, an identification model for structural health condition monitoring is built by using Mahalanobis-Taguchi System (MTS). It is shown by experiments that the Mahalanobis distance outputted from the MTS model can effectively describe the degree of abnormality of the structure condition. On the basis of the MTS model, an online monitoring approach is proposed which uses the MD average over a given monitoring cycle as an index to assess the trend of a structure health condition.