Cargo flows distribution over the loading sites of enterprises by using methods of artificial intelligence

Development of information technologies and market requirements in effective control over cargo flows, forces enterprises to look for new ways and methods of automated control over the technological operations. For rail transportation one of the most complicated tasks of automation is the cargo flows distribution over the sites of loading and unloading. In this article the solution with the use of one of the methods of artificial intelligence – a fuzzy inference has been proposed. The analysis of the last publications showed that the fuzzy inference method is effective for the solution of similar tasks, it makes it possible to accumulate experience, it is stable to temporary impacts of the environmental conditions. The existing methods of the cargo flows distribution over the sites of loading and unloading are too simplified and can lead to incorrect decisions. The purpose of the article is to create a distribution model of cargo flows of the enterprises over the sites of loading and unloading, basing on the fuzzy inference method and to automate the control. To achieve the objective a mathematical model of the cargo flows distribution over the sites of loading and unloading has been made using fuzzy logic. The key input parameters of the model are: «number of loading sites», «arrival of the next set of cars», «availability of additional operations». The output parameter is «a variety of set of cars». Application of the fuzzy inference method made it possible to reduce loading time by 15% and to reduce costs for preparatory operations before loading by 20%. Thus this method is an effective means and holds the greatest promise for railway competitiveness increase. Interaction between different types of transportation and their influence on the cargo flows distribution over the sites of loading and unloading hasn't been considered. These sites may be busy transshipping at that very time which is characteristic of large enterprises. Besides, the question of gaining experience in this model isn't tackled yet and it is to be developed in further publications