PROBLEM TO BE SOLVED: To satisfactorily separate overlapping peaks. SOLUTION: This is a peak analysis method in which three-dimensional data is processed and a peak in which a plurality of peaks overlap is separated into individual peaks, and singular value decomposition is performed on an input matrix representing the three-dimensional data to be processed. The step of finding a first matrix representing a plurality of base vectors for the first parameter and a second matrix representing a plurality of weighted vectors for the second parameter, which have a lower number of ranks than the input matrix. In the SVD projection space with the reduced number of ranks as the dimension, geometric analysis is performed on the trajectories defined by multiple weighted vectors, and the feature orientations in the space spanned by multiple base vectors are estimated and related. The step of finding the conversion matrix containing the information to be performed, and the conversion matrix is used to decompose the signal waveform in the first matrix of the dimension of the first parameter, and the conversion matrix is used to obtain the peak in the second signal waveform in the second matrix. Perform the steps to separate and. [Selection diagram] Fig. 1
展开▼