混合优化人工免疫网络用于过程动态优化

Hybrid optimal artificial immune network and it's application to dynamic process optimization

常见的用于求解过程动态优化的方法局部寻优能力强,易陷入局部点;而优化人工免疫网络虽局部寻优能力弱,但不易陷入局部点.针对这些方法的不足,提出了一种新的算法——混合优化人工免疫网络,将优化人工免疫网络植入局部寻优操作和二次响应机制,应用于Park-Ramirez和Lee-Ramirez生物反应器,此算法能以较少的计算代价搜索到最佳控制策略.将其用于模型参数发生变化的Lee-Ramirez生物反应器,实验结果表明,此算法的二次响应机制可以节省85%的评价次数.

The commonly used methods for the dynamic process optimization strongly converge to and easily trap in the local point, but the optimal artificial immune network （Opt-AiNet） difficultly converge to and trap in the local point. By introducing the operator of local exploiting and the second response to the opt-AiNet, a new algorithm, hybrid optimal artificial immune network （HOpt-AiNet） was formed. This new algorithm was tested by problems of dynamic optimization of Park-Ramirez bioreactor and Lee- Ramirez bioreactor. Experimental results showed that this algorithm can find the optimal control strategy, just needing less effort. For the dynamic optimization problems of Lee-Ramirez bioreaetor, whose parame- ters changing, the strategy of second response can reduce the evaluation number of the objective function by 85%.