ウェーブレット変換を用いた時系列データの解析

概要

論文の詳細を見る
Wavelet transform was conducted by French researchers about 10 years ago. This transform, which mapps time history data to time-frequency domain, has two parameters; scale and position (time). The basis of discretized wavelet expansion is orthonormal for suitable analyzing wavelet, which is the kernel of transform. On this basis, the square of wavelet coefficient of expansion correspond to energy. Therefore, it is possible to discuss the characteristics of the original data in terms of energy based on some scale and position. Moreover, a fast numerical calculation method has been developed for both wavelet expansion and inverse wavelet expansion. Thus, the method is expected to be applicable in various fields. In this paper, the authors describe derivations of wavelet transform and a discretized wavelet expansion. Then, the fundamental properties of the wavelet expansion are studied using two fundamental time history data. Finally, the time history data of an earthquake record is discussed.
一般社団法人日本建築学会の論文
1992-03-20