Zusammenfassung
Evolutionary computation (EC) is a general term
applied to a group of global optimisation techniques
whose main characteristics are inspired by biological
evolution. Instead of working with one possible
solution at a time, they usually start with a
population of random solutions. The initial population
evolves into a better set of solutions through three
main processes: selection, recombination, and mutation.
Those solutions having greater fitness are
preferentially selected for recombination to produce a
new set of possible solutions. Mutation is also used to
maintain diversity within the newly created solutions.
Through these processes, the fittest solutions transfer
their characteristics to a new generation of solutions.
By iterating over many generations, EC can find
solutions to many complex problems.
Models and simulations, especially those working as a
federation, can serve as the fitness function to
determine the value or adequacy of particular
solutions. In such federations, genetic programming
(GP) or other EC techniques can quickly find optimal or
near-optimal solutions for particular problems or
situations. The user is not required to systematically
search for the optimal solution; the computer
accomplishes that task. The tradeoff for accepting this
advantage is the requirement for the use of high
performance computing resources.
In this paper we briefly describe the fundamental
characteristics of EC. We also show some of the results
obtained during our research and development efforts on
different problems like image noise reduction and
discrimination of buried unexploded ordnance. We also
provide examples of how EC can be used with models and
simulations to find optimum solutions to many
complicated problems. This technique has great
potential for use with models and simulations in a
federated environment. The modelling and simulation
community needs to become more aware of these powerful
EC techniques so they may be applied in a wide range of
fields to quickly provide solutions to the war
fighter.
Distribution A. Approved for public release;
distribution unlimited.
Nutzer