The paper presents the application of a genetic algorithm to determine strategies for disassembly of products that have reached the end of their life. First, a general outline of the proposed methodology is provided and the features and specific properties of the genetic algorithm are described. Then an analysis of the algorithm's behaviour is carried out based on different problems. Once product structure is acquired, feasible disassembly alternatives may be determined; the domain of solutions may then be analysed through the genetic algorithm. First of all, a 'population' of acceptable solutions is randomly generated; then these solutions are estimated based on the criteria of the highest recovery value and the minimisation of discharged parts: genetic mutation and crossover operators are applied to the current population in order to generate a new population as a substitute to the previous one. Some cycles are made estimating, each time, the goodness of each individual solution and its probabilityt o 'reproduce' itself. At the end, the best-rated alternative becomes the solution of the algorithm. The solution of the algorithm is compared to the one provided by a 'best-first' algorithm (providing the optimal) solution), for different types fo products. In the paper, the efficacy of the proposed methodology is analysed, in terms of type of solution and computation time.
展开▼
