Biomedical Engineering Reference
In-Depth Information
NN results training (lead)
0.5
Output NN
Target
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
0
200
400
600
800
1000
1200
Samples
NN results training (Copper)
0.7
Output NN
Target
0.6
0.5
0.4
0.3
0.2
0.1
0
0
200
400
600
800
1000
1200
Samples
FIGURE 9.3
Neural Network response after the training phase. These plots report the expected and calculated ion concen-
trations for different samples on a test set for either lead or copper.
potentiogram, as peaks whose area is a linear function of the concentration of the metal in
solution. The software automatically computes the inverse function and integrates the peaks
within appropriate, user-specified intervals, relative to specific metal ions. In particular, for
lead the integration limits are 0.42-0.62 V, for copper -0.1- 0.1 V, for cadmium 0.7-0.9 V, and
for zinc 1-1.2 V. The software then recovers the concentration of an unknown sample by
means of preacquired calibration curves. The neural network processing software is an MLP
with one hidden and one output layer. The learning is achieved with a back propagation
algorithm. The input data are the coefficients of the polynomial curve fitting the acquired
data, i.e., the potential-time plot.
