An Evolutionary Genetic Algorithm For Integer Nonlinear Programming

Many problems of engineering and system are integer or mixed integer programming problems.lt is well known that the problem of integer programming is a NP-hard problem, and the method for it is exponential complex. Therefor, the approximate and evolutionary methods for integer programming have been developed quickly in these years. However, how to solve integer nonlinear programming is very difficult. Some research work for integer programming has been done by now, but less for integer nonlinear programming, and little has been done to it by initiatory method. Genetic algorithms are good evolutionary methods, and have been applied to many fields.In this thesis, genetic algorithms and mixed integer nonlinear programming are discussed particularly. At first their prosperities are given, according to these, the definition of searching excellent initial group and technical of changing integer scale are given, basing on this definition a new algorithm genetic algorithms with excellent initial group is described for mixed integer nonlinear programming.By the technic of changing integer scale, real nonlinear programming can be well solved by the algorithm.By theoretical analysis and numerical experimentation, the genetic method for large scaling multi-apex and no smooth mixed integer nonlinear programming can get a good global solution, and it is better than other algorithms used to resolve in the feasibility, stabilization and convergent speed of the solution.