On using intelligent scheduling for multi-criteria optimization in a PC assembly shop

This thesis deals with the implementation of a simulated annealing based intelligent multi-objective scheduling system for a PC assembly plant with two parallel assembly lines. The scheduling system aims to optimize six performance indices. These are average flow time, maximum tardiness, customer priority, inventory holding cost, production balancing and transportation cost between the two assembly lines.

The scheduling system adopts an a priori approach. Each performance index has a weight. The user chooses the weights (priorities) for the performance indices to prioritize them. The objective function is thus the weighted sum of the performance indices and is called the 'Utility Function'. After an initial schedule has been developed, the value of utility function is calculated and using a local search technique based on simulated annealing algorithm, the schedule is improved. This procedure is repeated on a number of starting schedules developed using known heuristics. This whole procedure has been carried out in a highly interactive environment in Visual Basic.

Various parameters of the scheduling system such as the number of iterations in simulated annealing technique, total number of orders, plant loading, new schedule generation, etc. are evaluated for their effect on the optimization performance. By comparing the performance of the scheduling system with available heuristics such as SPT, EDD, Fry, Blackstone and Armstrong, etc., it is seen that the system consistently develops good schedules. The methodology used is fast, simple, efficient and robust to incorporate any number of objectives in the objective function.