Система прогнозування потреби запасних частин автомобілів-тягачів на основі гібридних нейронних мереж за допомогою статистичних даних

Abstract

There are various methods for determining the need for spare parts: a method of calculating according to nomenclatural rules, definition of requirements for spare parts to power units according to parameters that define the loading and high-speed operation mode; determination of the rates of consumption of spare parts for the approximate evaluation of the first parts replacement. The conducted research has shown that the existing methods allow to calculate the required number of spare parts, each method has its advantages, but there are also definite drawbacks. Therefore, the development of the universal method that would allow to take into account the advantages and to eliminate the defects of existent methods, and which would be easy to use at the enterprises is expedient. Neural networks are well suited for the tasks of classification, optimization and forecasting. The system developed based on fuzzy logic elements using neural networks, in which as expert opinion incorporated the results of statistical data processing. As a result of the research work a hybrid network was obtained, by the means of wich the expert system that can predict the number of failures was developed.