Noise Reduction Method for Running TPA Using Significance Probability

In the present study, we investigated a noise reduction method by which to accurately obtain transfer functions in running transfer path analysis (TPA) using the principal component regression method. In the running TPA method, correct extraction of noise components from the principal components, which consist of transfer functions, is important. A statistical verification method for extracting the noise components in the principal components was applied. Rather than using the size of the principal component, as is the case in the conventional method, in the method of the present study, the significance probability of each principal component with respect to the output signal was set as a noise detection criterion. Subsequently, the noise reduction performance was verified through a simple simulation, and the performances of the statistical method and the conventional method were compared. The results of this comparison revealed that the influence of noise was reduced from the calculated transfer function more effectively by applying the statistical method than the conventional method.