Wavelet packet-based modulating identification and its application in TE process

Wavelet-based modulation identification is commonly used to build models. Since orthogonal wavelet transform operates only on the coarse part of the data series while keeping the detail part unchanged, it limits analyzing delicately. For this cause, we put forward the matrix expression of orthogonal wavelet packet transform, and discuss the modulating identification principle based on wavelet packet. It shows that this improved method is the equivalence of least squares method without weighting, and formulation with weighting is put forward based on the established rules. At last this method is applied to TE (Tennessee Eastman) process successfully.