| Assembly line has been widely used in producing complex items, such as automobiles and other transportation equipment, household appliances and electronic goods. Assembly line balancing is to maximize the efficiency of the assembly line so that the optimal production rate or optimal length of the line is obtained. Since the 1950s there has been a plethora of research studies focusing on the methodologies for assembly line balancing. Methods and algorithms were developed to balance an assembly line, which is operated by human workers, in a fast and efficient fashion. However, more and more assembly lines are incorporating automation in the design of the line, and in that case the line balancing problem structure is altered. For these automated assembly lines, novel algorithms are provided in this dissertation to efficiently solve the automated line balancing problem when the assembly line includes learning automata.;Recent studies show that the task time can be improved during production due to machine learning, which gives the opportunities to rebalance the assembly line as the improvements occur and are observed. The concept of assembly line rebalancing or task reassignment are crucial for the assembly which is designed for small volume production because of the demand variation and rapid innovation of new product. In this dissertation, two forms of rebalancing are provided, forward planning and real time adjustment. The first one is to develop a planning schedule before production begins given the task time improvement is deterministic. The second one is to rebalance the line after the improvements are realized given the task time improvement is random. Algorithms address one sided and two sided assembly lines are proposed. Computation experiments are performed in order to test the performance of the novel algorithms and empirically validate the merit of improvement of production statistics. |
