Evolutionary Computing: Application for Design of Technological Processes

Serhiy Shtovba

Vinnitsa State Technical University Khmelnitskoe Shosse, 95, Vinnitsa, 21021, Ukraine

serg@faksu.vstu.vinnica.ua

http://www.vinnitsa.com/shtovba

Poster

Evolutionary computing represents a stochastic method of optimization based on the mechanisms of natural evolution acting in live nature [1]. Evolutionary algorithms allow find out closed to global optimal solution quickly, and besides do not have much mathematical requirements about the optimization problem The notions of chromosome, gene and population constitute the base of evolutionary computing; and classical optimization theory terms of decision variables vector, decision variable and decision set can be brought into correspondence with them [2].

The basic operations of evolutionary computing are crossover, mutation and selection. Crossover represents an operation on two parents-chromosomes yielding two offspring-chromosomes each of which inherits some genes from parents-chromosomes. Mutation is a random gene modification. Selection represents itself as some procedure of population formation from the most adapted chromosomes according to its fitness function values.

An application of evolutionary computing are illustrated on example of solving the following technological process design tasks:

• optimal allocation of checking procedures in technological process;
• optimal choosing of multiplicity of checking-retrofit procedures;
• optimal choosing of variants of work and checking procedures;
• synthesis of optimal scheme of water purification technological process.

Carried computational experiments show the practicality of using evolutionary algorithms in compare with traditional techniques of discrete process optimization [3], is growing with the growth of problem complexity and technological process dimension.