Application Of Real Code And Combined Operators Genetic Algorithm To Nonlinear Problem

In reality very many models all are the nonlinear programming problems. To solve the non-linear problem, the traditional methods all are easy to get local optimization. Then appeare the intelligent optimization algorithm, such as genetic algorithm, simulation annealing algorithm, nerve network and so on. The genetic algorithm is the stochastic reconnaissance method which a kind of model biosphere's evolution rule evolves comes. In the standard genetic algorithm, speed is slow. Then combine genetic algorithm with other method to solve the non-linear problem and obtain the quite good result. But these researches is short to mix the operator to solve, the combined operator genetic algorithm need the less operation numbers than the blending genetic algorithm which uses together with other algorithm. If the mixed operator genetic algorithm can obtain the quite good result, it is better than blending genetic algorithm. At first the genetic algorithm use binary code, but afterwards proved the real coded genetic algorithms can obtains a better result compared to a binary code and moreover the computation must be few.This article uses the real code method and mix algorithm operator to enhance the genetic algorithm performance. The selection operation Roulette Wheel Selection , Elitist Selection, three kinds of crossover operators and four kinds of mutation operator mix. Taking advantage of mutually characteristic can optimize the non-linear programming problems. Thought some non-constraint non-linear examples have carried on the algorithm, specifically discussed each operator combination to the influence of the problem. Finally indicate the combined operator genetic algorithm was allowed to obtain a better result. To the constraint non-linear problems, based on the combined operation genetic algorithm, define two kinds of functions describing distance from the feasible zone to develop a new penalty function to carry on the penalty to break through the feasible zone. Based on the feasible zone distance new penalty function method, it has confirmed that this improvement method cen get a better result under the new penalty function combined genetic algorithm compared with the traditional algorithm and other genetic algorithm numerical example.Finally according to the constraint non-linear programming model of the supply chain coordinates. Based on the oriential model, this paper conside that the retailer demand is elastic demand and view the supplier price as a decision variable. To the oriential model and development model,the paper use completely enumeration the binary code genetic algorithm, single operator real code genetic algorithm and real code combined operator and penalty function of improvement genetic algorithm to solve . Compared to the solution, the real code mix operator penalty function of improvement genetic algorithm obtains a better result, and the running time is less.