Оцінювання достовірності контролю токсичності стічних вод мультиспектральним методом з використанням нейромережі

Abstract

The system of monitoring the index of toxicity of sewage with the help of a multispectral method with the use of bioindications for phytoplankton and neural networks for processing results. The solution of the inverse optical problem for indirect measurement of the parameters of the medium on the basis of the multispectral method was carried out using neural networks based on a multilayer perceptron with one hidden layer, a multilayer perceptron with two hidden layers, a network with a radial basis function, a generalized regression neural network. The reliability of toxicity control and the probability of errors of the first and second kind is estimated.