Development of digital filter based on fuzzy rules and its application to biological signal processing

摘要

In this paper, a new type of digital filter with fuzzy weighting function based on fuzzy rules is represented. This filter can be applied to the analysis of Acoustically Evoked Potentials (AEPs) and Evoked OtoAcoustic Emissions (EOAEs). AEPs andEOAEs have been analyzed with conventional filtering and ensemble averaging techniques, but there are some problems remaining unsolved. The frequency characteristics of biological signal change depending on the stimulus conditions, and those of noise also change. It is difficult to determine the cutoff frequency of the filter before filtering operation. The proposed fuzzy filter presented in this paper is designed by using power spectrum of ensemble and alternative ensemble averaging data of biologicalmeasurement. An ambiguity of frequency characteristics of signal and noise can be treated effectively in this method.The signal to noise ratio of fuzzy-filtered data is improved significantly compared with unfiltered one. Result of the comparison between fuzzy filter and Wiener filter shows that the fuzzy filter is much more adequate for biological signal processing.