About control of a neuron as an elementary unit of multilayer artificial neural networks with variable signal conductivity

Abstract

The report is devoted to a possibility of a control of each Hopfield’s neuron as an elementary unit of the neural network with variable signal conductivity used for transport scheduling tasks. It is set that direct control of the neuron is impossible in terms of initial data available for initializing the neural network. Also the result of some control algorithms based on statistical computations and signals dynamic is described below.

Development of control systems and safety devices for solving the problems of digital transformation of the transport complex

“…Events that are considered safe under the current binary approach should be re-evaluated accordingly in the new MCS. An unscheduled stop of the train on the stage is not dangerous, and is even considered a safe condition [14]. However, such a train stop can create the prerequisites for a chain of potentially dangerous events.…”