一种新型无约束优化方法的研究

Research on New Unconstrained Optimization Method

提出了一种新型无约束全局优化方法。主要是从当前常用算法的优化过程中总结出共性—反馈特性,然后利用反馈控制系统来实现这种优化特性。在这种智能计算系统中,将无约束优化问题的目标函数作为控制对象,将PID控制器作为一种优化算法。由于传统PID方法容易达到局部极小点,提出了多系统并行PID优化方法。最后采用基准函数对此优化算法进行测试,并与几种不同类型的改进优化算法进行对比,数值实验结果验证了此种算法的有效性。

In this paper,a new kind of global optimization methods is presented.The feedback characteristic is implemented by feedback control system,which is abstracted from the current optimization algorithms.The objective function of unconstrained optimization problem is regarded as the plant and the proportion-integration-differentiation（PID） controller is used as an optimization algorithm in the intelligent computing system.Multi-system Parallel PID（MPPID） algorithms is proposed to conquer the shortcoming of the PID method for getting into the local minimum.Compared with some other optimization algorithms for the benchmark functions in the experiment,the results indicate that the proposed method is feasible and effective.