摘要
提出了被控对象模型——扑翼模型,它是对可进化的实际生物模型的简化,具有非线性、多输入多输出等特点。用在复杂的学习系统时,可对其效率及性能进行检验。基于该模型,构造了一个神经网络控制方案。它是一个逐级发展的学习控制系统,可用于分析和研究生物的学习过程。在该系统上,进行了对目标捕捉的计算机模拟,当目标静止、随机运动或以一定规律运动时,均获得了满意的结果。随着学习的深入,该系统还可获得预测能力。由于本系统具有逐层建构的特点,故可完成复杂的控制任务。
This paper presents a new objective control model,flapping wing,which is simplified from the evolutionable biological model. Its motion equations have nonlinear,MIMO (multi-input multi--output) features and can be used in a complex learning system to test its efficency and function. Based on the model,a neural network control scheme is presented. It is a learning control system that can augment its abilities level by level to study and analyze the biological learning process. Objective catching simulation has been realized in the system so that whether the object is fixed or moving, a satisfactory result is always able to obtain with the intensification of the study,the system can even obtain the ability of prediction. Since the system is built level to level,it can undertake complex control tasks.
出处
《清华大学学报(自然科学版)》
EI
CAS
CSCD
北大核心
1996年第5期48-53,共6页
Journal of Tsinghua University(Science and Technology)