Aplicação de algoritmo para recuperação de casos no problema de condução de trens de carga

Abstract

This paper has as main objective to use the Case Based Reasoning technique in order to help in freight trains conduction. CBR is composed of steps, like similar cases retrieval, reuse of retrieved cases in order to generate a candidate solution to the applied one, review and retention of the applicable solution. In this work, the focus will be on the data retrieval step, which uses a case base composed by cases and theirs respective solutions, and algorithms to select the most similar cases to a determined problem. The retrieved cases are used and treated by the other steps of the CBR method. On the focused domain, each case corresponds to a set of traits that refer to a displacement of a train in a railway. With the implementation of the retrieval algorithm improvements are hoped in the train's conduction performance, just like the decrease of fuel expense and consequently a decrease in the pollution caused by it. The improvements can be reached because, once facing the train conduction problem in a part of a railway, the CBR technique will retrieve a set of possible actions from the available case base and the train conductor can apply these actions.