Abstract
With eco-awareness and eco-regulation, the disposal of end-of-life (EOL) product has been considered. Disassembly is the operation to apart the component or subassembly from the main product. The disassembly sequence would affect the disassembly efficiency. So, disassembly sequencing problem (DSP) has become increasingly important in the process of recycling, reclamation, or remanufacturing EOL products. However, most of the studies in the disassembly sequencing plan assume that the processing time is deterministic for each component’s disassembling procedure. For a realistic and logical approach to the DSP, this thesis takes account of the learning effect with stochastic concept and job-dependent, that is the condition in which different components may have different learning rate. With the quantity of components increasing, the sequences of DSP would grow too dramatically to find the optimal solution. Considering the NP-complete nature of the stochastic problem, a genetic algorithm (GA) is applied to minimize the total expected disassembling time for this problem. Taking six EOL products as examples, an experiment is executed and the results are compared with another algorithm, particle swarm optimization (PSO). GA is verified with better effectiveness and convergence efficiency than PSO. Another experiment is executed to determine the best setting of iterative generation and chromosome numbers for making the GA more efficiency.