Multilayer feedforward networks are universal approximators
Neural Networks
Approximation theory and feedforward networks
Neural Networks
| Hi-index | 0.00 |
Quantitative determination of dopamine in urine and plasma is very important for curing of diseases like ganglioneuroma, schizophrenia, manic-depressive psychosis, stress, burn-out syndrome. The measurements are made mainly with radioimmunoassay and chromatographically. These procedures are time consuming (over 4 hours) and need expensive laboratory technique. Electrochemical biosensors are very simple, easy to use and cheap devices for measuring of substrates. Dopamine is one of them. In this paper a biosensor with active membrane from banana tissue is used for experiments. As far to an enzyme reaction is used for determination of dopamine, temperature has strong influence on the output signal from biosensor. The paper treats CMAC-based modelling the influences of temperature and substrate concentration on the output current of biosensor. The high accuracy of the CMAC-based approximation (at the expense of more experimental data) is confirmed by simulations in MATLAB environment. The relative errors calculated over the new experimental data are used for validation of the proposed method. Additionally, the results are compared with those obtained by backpropagation-based feed-forward neural network.