In this paper, we propose a new parameter estimation method that can deal with the data of multiple time intervals simultaneously. If there are common modes in the multiple time intervals, it is possible to create a new polynomial by summing the coefficients of the prediction error polynomials of each time interval. By calculating the roots of the new polynomial, it is possible to estimate the common modes that exist in each time interval. The accuracy of the proposed parameter estimation method has been proven by using appropriate test signals.