Abstract: The wavelet transform is developed to identify the different phases in a fermentation
process. In this method, the wavelet transform modulus maxima are used to estimate the
local maximum points of the second derivative of the growth curve in order to classify the
different phases of fermentation process. Moreover, the method can effectively get rid of
noise from the signal, making use of the different characters showed by signal and noise in
the wavelet transform modulus maxima. Compared with neural network modeling, the presented
method needs less quantity of information and calculation. The results of experiments show
that this method is effective.

Key words: wavelet transform, fermentation, classification

摘要： The wavelet transform is developed to identify the different phases in a fermentation
process. In this method, the wavelet transform modulus maxima are used to estimate the
local maximum points of the second derivative of the growth curve in order to classify the
different phases of fermentation process. Moreover, the method can effectively get rid of
noise from the signal, making use of the different characters showed by signal and noise in
the wavelet transform modulus maxima. Compared with neural network modeling, the presented
method needs less quantity of information and calculation. The results of experiments show
that this method is effective.

关键词: wavelet transform;fermentation;classification