人工神经网络控制的实时仿真系统

为了观测使用人工神经网络作为控制器的实际的自动控制系统的控制效果,我们开发了一种实时闭环仿真系统.该仿真系统具有两个独立的部分:一部分是被控对象部分,由PC计算机计算被控对象的数学模型来模拟;另一部分为控制器部分,由真正的神经网络硬件实现.两部分由硬件接口电路连接在一起.此仿真系统工作于真正的时间轴中,即数学模型中的时间常数不再仅仅是计算中的参数而是反映真正的时间长度,它满足检验用于实际系统的神经网络控制器性能的需要.实验结果表明,此实时仿真系统对于设计基于人工神经网络的控制系统是一种有用的工具.

The Real Time Emulation System of Artificial Neural Networks Control

In order to observe the results of practical automatic control system using artificial neural networks as the controller,we developed a real time closed loop emulation system with two independent parts.One is the PC based simulating model of controlled plant and the other is the real neural networks hardware served as the controller.They are connected together by interface circuits.The emulation system works under real control time sequences.It meets the requirements of checking the results of neural networks controller in practical system.Experiments show that the real time emulation system is a useful tool in designing control system with the artificial neural networks.