Evolutionary Algorithm is a stochastic search methods that try to mimic the properties of the biological evolution of living things. Evolutionary Algorithm works by the principle that a good individual, will be able to survive and produce individuals who are getting better with each generation. The elements that exist in this method is taken from natural processes such as selection, combination, mutation, migration, and displacement locally. Evolutionary Algorithm to work more on a number of individuals in a population rather than only on a single individual.

Work processes on Evolutionary Algorithm selection process includes the selection of individuals who will be combined to produce a new individual in the process afterwards. After selection of the next process in the form of combination. This process is a process of combination of individuals who have been selected on the previous selection process. This process will produce a new individual which is different from its parent. After a combination in a small proportion of individuals will be the mutation that is changing the nature of an individual because the individual exchange process. In addition to the above processes are also carried out the calculation process the objective function value which is also called the fitness value.

This algorithm is different from other classical optimization methods in several respects, among others:

1. This method is a nondeterministic method which will produce different settlements although early models had not been changed, due to the use of random sampling in this algorithm.

2. This algorithm has a population containing completion candidates.

3. In its application, Evolutionary Algorithm tries to combine elements of the solutions that already exist to create new solutions with inherited traits possessed by each parent.

In practice, Evolutionary Algorithm has several drawbacks, among others:

1. Sometimes a problem with using Evolutionary Algorithm requires a longer time than other alternatives such as method of settlement that GRG (Generalized Reduced Gradient).

2. These algorithms are usually hampered by the dimensions of the problem. Sometimes the problems with large dimensions is difficult to get near-optimal solutions.

3. This algorithm does not have a concept for an optimal solution or even another way to test whether the solution found was already an optimal solution.

—————***—————

Arifinfo is an Industrial Engineering’s blog.

Contributing Thought Leader Blognotions, for Manufacturers and Engineering.

—————***—————

• «Previous Post
• Next Post »

---