ORGANIZATION OF A LOGISTIC NETWORK MODEL USING SOFTWARE AND NEURAL NETWORK ALGORITHMS

Abstract

The work explores the principles of organizing a logistics network model using software and neural network
algorithms. The structure of the modern logistics network is disclosed, which includes: supply chain, sales chain and storage.
It is emphasized that the supply chain side has the most comprehensive information on product analysis and also covers the
sequence of warehouse updates. It is emphasized that the information management efficiency of the logistics network
determines the update of the product at the end of the supply chain, and only with the help of complete product information of
the end supply chain can the efficiency of logistics in the field of operation be effectively controlled. It is noted that the
introduction of intelligent network technology allows for the implementation of intelligent logistics management, and the
gradual creation of nodes of the Internet of Things integrates logistics resources, which allows for the optimization of the
management system in the information system, solving the problem of route planning, the turnover of materials for storage, the
correspondence of information about the product and the client and undifferentiated proximity distribution, can reach a new
level. A diagram of a graph convolutional neural network with a detailed description of the functioning mechanism is proposed.
As a modification, the use of a tensor is proposed and a detailed structure of a network with a tensor is provided. Emphasis is
placed on the diagram of the sparse structure of the wrapping layer with a description of the functional component. It is proved
that the proposed structure differs from the previous one by an increased level of efficiency due to the fact that the real-time
logistics data of each logistics node is collected, and the logistics data is pre-processed to eliminate non-standard data in the
analysis process. In turn, the difference of logistics data at different time nodes increases, and there is a division of logistics
data into a peak period and a smooth period according to the time level, but the set of logistics data at different two stages obeys
random distribution.