适应型神经网络系统的控制

控制对象系统根据控制领域的不同，或多或少具有非线形特性。尤其汽车发动机系统中更具有很强的非线性。目前，三层构造的神经网络对于非线性系统的控制非常有效。

An Adaptable Neural Network System for Control

According to the difference of its control fields the control object systembears,more or less,a nonlinear characteristic. Especially the automobile engine system hasa much stronger nonlinearity. The tri-layer-structured neural network is very efficient incontrolling the nonlinear system. But, generally speaking, this non-linearity mostly varieswith time. If the time change in non-linear field can be controlled by neural network system the function of neural network will undoubtedly double. This paper presents a methodto build a two-unit neural network syStem, one unit to control the other to supervise theactions of controlled objects and join computers to carry on BP study.As time divisionchanges, the two-unit neural network system will alternate from one to another,so that thewhole system will adapt itself to the time change.