Uncertainty of the first type in industrial ecology

Abstract

Abstract The effect of low-frequency electromagnetic fields on the human body of those working in the North of the Russian Federation in many cases remains difficult to identify. The reaction of the cardiovascular system to this factor in many cases is rather difficult to register within the framework of traditional statistics. However, it is possible to use artificial neural networks in two new modes. In this case, for each feature separation procedure xi(t), we set the initial values of the weights wio of the features xi(t) chaotically, i.e. wio ∈ (0, 1). We repeat the neural network setup many times (at least 1000 iterations) and the resulting samples xi(t) are ranked. If the weights wi after 1000 iterations show values above the statistical mean <wj> ≥ 0.05, then such j-th features can be considered the main (or order parameters). As a result, we have now formalized the procedure for finding the order parameters, that is, we have formalized the system synthesis (finding the main features).