Kotenko V. Improving of the grain crops delivery process by machine learning methods

The object of the dissertation research is the transport process of grain crop delivery from agricultural enterprises to the elevator. The subject is the patterns of changes in the parameters of the transport process of the grain crops delivery from agricultural enterprises to the elevator, depending on the peculiarities of its organization. The purpose of the theses is to improve the organization of the transport process of grain delivery from agricultural enterprises to the elevator using machine learning methods, taking into account functional and cost indicators of order fulfillment. During the work on the dissertation, the following research methods were used: the theory of transport processes, mathematical statistics, mathematical modeling, machine learning methods, the method of hierarchical cluster analysis and computer experiment. The research was carried out using modern software, as well as using application packages of STATISTICA, MATLAB, and Mito spreadsheets. To perform simulations and optimization calculations, an application software was developed in Python 3.9 using the Scikit-Learn library.In the theses, for the first time, a model for predicting fuel consumption by vehicles during the execution of orders for the grain crops delivery from agricultural enterprises to the elevator is developed, which is based on machine learning and the use of the formed database formed on the results of previously performed transport processes, taking into account production conditions.The method of selecting rational vehicles for the performing orders for the grain crops delivery from agricultural enterprises to the elevator has been improved. This method systematically takes into account the peculiarities of the organization of the transport process and the specifics of individual orders, and unlike existing methods involves the use of experience of previously implemented transport processes, includes features of production conditions and characteristics of available vehicles, which ensures the selection of rational vehicles.The practical value of the obtained results lies in the development of an algorithm and a computer model based on the proposed method of selecting rational vehicles for the performing orders for the grain crops delivery from agricultural enterprises to the elevator, which involves the use of the developed model for predicting fuel consumption by vehicles and ensures improved decision-making and efficiency of the relevant transport processes. The obtained research results are intended for use by managers of transport enterprises that organize the grain crops delivery from agricultural enterprises to the elevator. The use of the developed model, the improved method and on their basis the proposed computer model provide the acceleration of decision-making and increase the efficiency of the relevant transport processes.The results of the research were tested in the practical activities of LLC "Volyn-Zerno-Produkt", which made it possible to reduce the cost of grain crops transportation from agricultural enterprises to the elevator. The materials of the dissertation are implemented in the educational process of Vinnytsia National Technical University and Lutsk National Technical University.