求解一类非线性规划问题的混合遗传算法

提出了一种求解目标函数和约束条件均二阶可导的非线性规划问题的混合计算智能算法．该算法是把一种浮点数编码遗传算法和约束变尺度法相结合提高求取全局解的速度和概率．在该算法中，选择、交叉和变异等遗传操作算子是以非线性规划问题的一个惩罚函数为求解对象，目的是把解引向全局解附近，为约束变尺度算子提供初值；而约束变尺度算子直接以原非线性规划问题为求解对象，以发挥其局部搜索能力强的优点，数值实验表明，混合算法是一种可靠、高效的全局优化算法．

Hybrid Genetic Algorithm for Solving a Class of Nonlinear Programming Problems

A hybrid genetic algorithm for solving nonlinear programming problems with twice differentiable objective and twice differentiable constraints was presented. In the algorithm, a constrained variable metric method is set in a real-code genetic algorithm to improve the genetic algorithm's global convergence speed and global convergence probability. The constrained variable metric method is taken as an optimization operator paralleling to the selection operator, crossover operator and mutation operator. The selection operator, the crossover operator and the mutation operator are used to optimize the penalty functions of the original nonlinear programming problems to provide good initial values for the constrained variable metric method. However, the constrained variable metric method directly solves the original nonlinear programming problem to take advantage of its powerful local searching ability. The numerical experiments illustrate that the present hybrid algorithm is an efficient and reliable global optimization approach.