论文部分内容阅读
针对自适应神经模糊推理系统(ANFIS)在锌钡白干燥煅烧生产过程建模中出现的模糊结构辨识问题,采用了基于人工免疫系统(AIS)的聚类算法。该算法通过免疫网络对抗体及记忆数据集逐代克隆、变异及抑制操作,提取有用的模糊规则数目,避免ANFIS训练陷入局部极小点。本文详细探讨了AIS随机特性对聚类规模稳定性造成的影响以及AIS的聚类速度问题,对Castro算法做了必要修改。通过与减法聚类算法、模糊C-Means聚类算法(FCM)特性上的对比分析,得出AIS在复杂过程辨识中的实际应用价值。
A fuzzy clustering algorithm based on artificial immune system (AIS) was proposed to solve the fuzzy structure identification problem of adaptive neuro-fuzzy inference system (ANFIS) in the modeling of drying process of lithopone. This algorithm extracts the useful number of fuzzy rules by cloning, mutating and suppressing antibodies and memory data sets one by one by immune network, and avoids the ANFIS training falling into a local minimum. This paper discusses in detail the impact of random characteristics of AIS on the stability of clustering scale and the clustering speed of AIS, and makes necessary modifications to the Castro algorithm. Through comparative analysis with subtractive clustering algorithm and fuzzy C-Means clustering algorithm (FCM), the practical application value of AIS in complex process identification is obtained.