基于改进KH-ANFIS 的海洋溶菌酶发酵过程软测量

Soft Sensor Modeling of Marine Lysozyme Fermentation Based on Improved KH-ANFIS

摘要:

针对海洋溶菌酶（Marine Lysozyme,ML）发酵过程菌体浓度在线检测难以实现，离线测量不能反映发酵过程当前变化等问题，提出了一种基于改进磷虾群—自适应模糊神经网络软测量（HLKH-ANFIS）建模方法。首先利用自适应莱维飞行策略对传统KH进行改进，从而提升算法的全局搜索能力；同时利用跳变技术（HOT）对KH算法位置更新公式进行改进，提高算法的局部寻优能力，然后利用改进的KH算法对自适应模糊神经网络反馈进行优化，改善其过度修正和计算量大的问题；最后建立基于HLKH-ANFIS的海洋溶菌酶发酵过程菌体浓度软测量预测模型，仿真分析表明：相较于KH-ANFIS预测模型，HLKH-ANFIS模型的误差较小，具有更好的预测能力，能够满足ML发酵关键参量的在线预测需要。

Abstract:

In view of the fact that the bacterial concentration of the fermentation process in the Marine Lysozyme in difficult to be detected online and the off-line measurement cannot reflect the the change of the parameters in time, a new method of soft sensor modeling which is based on improved Krill Herd algorithm (KH) and improved adaptive fuzzy neural network is proposed. Firstly, the adaptive levy flight strategy is used to improve the global KH search and the HOT is used to improve the local search capabilities; Then the improved KH is used to reduce over correction and large calculation in the adaptive fuzzy neural network; Finally, the soft sensor prediction model based on HLKH-ANFIS of bacterial concentration in ML fermentation process can be established. The simulation result show that the improved model has smaller error and better prediction performance than KH-ANFIS, which can meet the online prediction demand of the key parameters in ML fermentation.