Design The Initial Population Of The GENOCOP Algorithm

Genetic Algorithm(GA) is a kind of self-adaptive,stochastic,global optimization method. To solve problem,GA deals with structural object directly without any requirements of differentiability and continuation of function;it's global, robust, and parallel in working mode.This dissertation intruduces GA and GENOCOP algorithm in brief, then gives a example to show that randomly generated initial population which is biased in local constrained space is responsible for premature convergence and increasing iterative times.If the initial population can be distributed in the whole feasible region uniformly,the GENOCOP algorithm can avoid the disadvantage metioned above effectively.We propose two methods to design the initial population of GENOCOP algorithm which solve the linear constrained numerical optimization problem:1. Uniform Initalization - takes initial population which scatter roughout the whole feasible region with upper and lower range constrains.uniformly .Makes the initial population of GENOCOP algorithm corresponds to uniform design with multi-factor and multi-level.So we use less test times to find initial points which scatter throughout the whole feasible region uniformly .By comparing the results of Rondom Initalization results,we can find that Uniform design is better.2. Boundary Initalization -takes boundary points as initial populationPractically and empirically,most of optimal solutions of real-world constrainted problems are on the boundary of constraints.The method initializes the population randomly on the boundary of constrained space,and makes the search of solutions more efficient.This dissertation proposes two method of Boundary Initalization.i. Takes all vertexes as initial population.ii. Gains about 2n vertexes and takes them as initial population (n is the numbers of equations).At last, the convergence properties of improved GENOCOP algorithm and improved GENOCOP algorithm are proved, and examines the improved GENOCOP algorithem by a set of examples, the numerical results illustrate the feasible and effecctive of the improved GENOCOP algorithem.