摘要
根据体液免疫中抗体的二官性性质,用CMAC神经网络模拟抗体的恒定区、ADA-LINE神经网络模拟抗体的可变区,用两种神经网络的结合近似模拟一个抗体。一个抗体可识别多个抗原,这与多模型控制中一个控制器可控制某个变化范围内的对象有着极大的相似之处。根据上述的原理,该文用两种神经网络构造出多模型控制器并结合某过热汽温对象验证了该控制器的有效性。
CMAC and ADALIN neural network are used to simulate constant and variable region and they are combined to simulate an antibody. Just like an antibody could identify different antigens, one controller could control many objects. Based on this theory, two 1 aral networks are used to make multi-model controller and its validity are validated.
出处
《自动化与仪器仪表》
2006年第5期8-13,共6页
Automation & Instrumentation
基金
教育部博士基金(20020286001)资助
