摘要
针对现有的钢坯热轧过程智能建模方法——模糊神经网络建模存在的收敛速度慢、建模精度不高、易陷入局部极小值、系统输入输出向量维数和空间划分增加使网络结构趋于复杂等问题,提出了一种基于人工免疫系统(AIS)聚类的自适应神经模糊推理系统的建模方法。该方法采用人工免疫聚类学习算法来确定模糊集合的划分,并确定模糊神经网络的结构和初始参数,能以较少的模糊规则达到理想的建模精度,仿真结果表明了该方法的有效性。
In light of the problems with FNN modeling of billet hot-rolling, which include slow convergence speed, low modeling accuracy, strong possibility of obtaining Local minimum value, more complex network structure resulting from the increase in input and output dimensions as well as division space, this paper proposes an ANFIS modeling method based on artificial immune system （AIS） cluster algorithm. The method can determine the division of the fuzzy set, the structure of FNN network and the initial parameters. It can reach the ideal modeling precision with few fuzzy rules. The simula- tion points to the effectiveness of the proposed method.
出处
《武汉科技大学学报》
CAS
2009年第2期146-148,共3页
Journal of Wuhan University of Science and Technology
