Synthesis of structural diagrams for drum drives of hoisting and transport machines based on a feed forward neural network architecture

Synthesizing drive structural diagrams is typically a one-dimensional (narrowly focused) problem, traditionally solved for specific designs of lifting-and-transport machine mechanisms. This article proposes an algorithm for synthesizing drum mechanism drive structural diagrams based on a feed forward neural network architecture. This algorithm enables the synthesis of a generalized mechanism drive under conditions of design object differentiation, i.e., the use of a single synthesis algorithm for drum mechanisms of various lifting-and-transport machines (elevators, hoists, boom cranes, overhead cranes, belt conveyors, etc.). The architecture of the proposed neural network assumes supervised learning based on adjusting the weighting coefficients of connections between corresponding neurons, with data divided into sets for training, validation, and testing. A dedicated genetic algorithm for synthesizing the drum mechanism drive structural diagram for a specific type of lifting-and-transport machine is embedded in the structure of each neuron.