求解边值固定动态优化问题的一种混合算法

A hybrid algorithm for dynamic optimization with fixed boundary

针对边值固定动态优化问题的数值求解,提出了一种集成随机性方法与确定性方法的一种新的混合算法。迭代遗传算法(IGA)把初始种群及繁衍产生的后代不断供给两点梯度法,两点梯度法以其为初值搜索满足边值固定约束的可行控制策略并回送给迭代遗传算法,迭代遗传算法则根据可行控制策略对应的目标函数值进行选择与进化操作。该混合算法简便易行。实例研究显示了该混合算法的可行性与稳健性,能以足够的精度满足边值约束。

For solving dynamic optimization with fixed boundary by numerical method, a new hybrid algorithm was developed, which integrated Iterative Genetic Algorithm （IGA） and two-point step size gradient method. The initial group of iterative genetic algorithm and their children were supplied to two-point step size gradient method as its initial values. Two-point step size gradient method was then run to satisfy fixed boundary requirements, whose result, that is the feasible control profiles, was returned to IGA as the criterion of selections and evolutions. The hybrid algorithm is easy to implement. Case studies show it is feasible and robust. It has the abilities to satisfy fixed boundary requirements with enough accuracy.