Multi-objective evolutionary method for cargo arrangement in a loading space

Abstract

The paper characterises a class of problems for packing boxes in the container. It presents the current state of knowledge in this area and distinguishes evolutionary algorithms, as the most promising in the search for quasi-optimal loading conditions. The method proposed in the paper focuses on certain criteria important from a practical point of view, which in a formalised manner have not been included so far in the solving-problems models. Apart from the traditional consideration of the problem of three-dimensional space loading maximisation, the proposed method considers the deviation of the loaded container weight centre from its symmetry planes and the availability of packages during unloading. New elements in the described method are: the applied criteria, penalty function, way of coding the container loading state in the evolutionary algorithm and fast crossover and mutations operators dedicated to the adopted coding. It was observed that the evolution in the developed algorithm occurs properly, that is seeking to minimise the criteria. The article also includes a calculation example showing the effect of the method with the discussion of the results indicating the advantages and disadvantages of the proposed solution. The performance of the algorithm has been considered in the context of time necessary to obtain the acceptable solution and quality of the obtained solution. It was found that the algorithm in its current form is a strong base for its further improvement.