Authors:
Bruno Alessi Castro
;
Pablo Gustavo Cogo Pochmann
and
Eduardo Borba Neves
Affiliation:
Officers’ Improvement School (Escola de Aperfeiçoamento de Oficiais – EsAO), Duque de Caxias Avenue, 2071, Rio de Janeiro-RJ, Brazil
Keyword(s):
Machine Learning, Logistics, Amazon, Multiple Linear Regression, Resource Optimization.
Abstract:
The present study is an analysis of the use of Machine Learning tools in favor of river logistics transport in an Amazon jungle area and the impacts on the efficiency of the Logistics Commander's planning, due to a research gap identified through imprecise methods for estimating fuel consumption in logistics trips. In this way, a quantitative mathematical model was developed, using Multiple Linear Regression algorithms (due to its simplicity for operators not specialized in the area) to predict fuel consumption on logistical trips carried out by Vessel’s Center of Amazon Military Command (CECMA) vessels, using statistical data found in travel reports. After this, a comparison was made of the model found with the current modus operandi of the complement calculation completed by CECMA. applying a back test to validate the proposed model. The results obtained generated research with an R of 0.935, explaining 87% of the proposed trips. In this context, a software proposal was presented t
o be developed with an online interface and with the interaction of the two algorithms. Thus, the use of machine learning tools such as MLR, integrated with an AI system with feedback on predictive variables and fuel consumption of logistics missions brings an increase in the efficiency of military logistics planning and reduces costs related to fuel management after missions, contributing to the constant evolution and improvement of Military Doctrine.
(More)