The main achievement of this research is the development of a genetic algorithm model as a solution approach to the single model assembly line balancing problem (SMALBP), considered a difficult combinatorial optimisation problem. This is accomplished by developing a genetic algorithm with a new fitness function and genetic operators. The novel fitness function is based on a new front-loading concept capable of yielding substantially improved and sometimes optimum solutions for the SMALBP. The new genetic operators include a modified selection technique, moving crossover point technique, rank positional weight based repair method and dynamic mutation technique. The moving crossover point technique addressed the issue of propagating best attributes from parents to offspring and also supports the forward loading process. The new selection technique was developed by modifying the original rank-based selection scheme. This eliminates the high selective pressure associate with the original rank-based technique. Furthermore, the modified selection technique allows the algorithm to run long enough, if required, without premature convergence and this feature is very useful for balancing more complex real world problems. The repair technique included in this model repairs a higher proportion of distorted chromosomes after crossover than previous methods. Moreover, a third innovative feature, a moving adjacent mutation technique, strengthens the forward loading procedure and accelerates convergence. The performance of the front-loading fitness function currently outperforms the published fitness functions and fifty-four published test cases generated from sixteen precedence networks are used to assess the overall performance of the model. Encompassing the new genetic algorithm concepts, forty-four test problems (81%) achieved the best solutions obtained by published techniques and twenty-four problems (44%) produced better results than the benchmark Hoffmann precedence procedure, the closest non-genetic algorithm method. The superiority of the genetic model over other heuristics is identified in this research and future developments of this genetic algorithm application for assembly line balancing problems is evident.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:273223 |
| Date | January 2002 |
| Creators | Thilakawardana, Duminda |
| Publisher | University of Surrey |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://epubs.surrey.ac.uk/844283/ |
Page generated in 0.0017 seconds