Page 160 - Algorithm Collections for Digital Signal Processing Applications using MATLAB

4. Selected Applications 149

________________________________________________________________________
dofilter.m

function [res]=dofilter(x)
load WEIGHTSNOISEFILT
res=simuff(x',W1,B1,'logsig',W2,B2,'logsig');
________________________________________________________________________

3.3 Program Illustration

Note that low frequency information is lost and that is not retrieved in the
filtered signal. This is due to the fact that first 100 samples are used to train
the ANN. If the training sequence is increased, low frequency information
can also be preserved in the filtered signal. Also note that MATLAB Neural
Network toolbox is used to train the network.

Figure 4-7 Noise Filtering using BPNN