An Evaluation of a Modified Simulated Annealing Algorithm for Various Formulations
AIR FORCE INST OF TECH WRIGHT-PATTERSON AFB OH
Pagination or Media Count:
Many combinatorial optimization problems are too large to allow exact solution methods to give results in a reasonable amount of time. An alternative is the use of heuristics that sacrifice a degree of optimality in return for timelier solutions. One such heuristic, simulated annealing, is a form of iterative improvement that probabilistically accepts less optimal configurations. This procedure allows the algorithm to escape local minima or maxima in its search for a global minimum. This paper presents modified simulated annealing formulations of common industrial engineering problems. The modified algorithm behaves like a biased random walk that can be tailored to suit a users particular bias set. Modifications to the standard application include an expert system front end, a constraints module, a user interrupt capability, and the creation of alternative acceptance functions.
- Numerical Mathematics