Hardware implementation of backpropagation algorithm based on CHEMFET sensor selectivity

In this study, modeling approach for interpretation of data logged from chemically field-effect transistor (CHEMFET) sensor is described. Firstly, backpropagation algorithm is used to train the proposed network by optimizing the parameters of the network. Then, by applying the optimized parameters obtained from the trained network, the feed forward neural network algorithm is implemented using C language for compatibility with 16-bit microcontroller board and the output is compared with the simulation output which has been simulated using MATLAB software. Initial findings showed that the neural the proposed method is able to provide excellence estimation of main ion concentration in mixed solution as well as capable to interpret and estimate the ion concentration in mixed solution.