网络控制系统的自适应预测控制

Adaptive Predictive Control of Networked Control Systems

研究了具有网络诱导时延和数据包丢失的网络控制系统中的自适应预测控制问题。考虑传感器与控制器、控制器与执行器间皆有时延和数据包丢失,充分利用网络传输数据包的容量,提出一种方案,在控制器节点获得被控对象的输入输出数据以在线辨识对象模型参数,并利用辨识模型设计预测控制器,给出了算法的推导过程和时延及数据包丢失的补偿方法。校园网环境下的实验验证了所用方法的有效性。

Adaptive predictive control problems of networked control systems with network induced delay and data packet dropouts. Considering time-delay and data packet loss exist in sensor-controller channel and controller-actuator channel, a scheme is proposed in controller node to obtain the input-output data by using the capacity of the transmitted data packet sufficiently, and furthermore on-line identify the object model parameters, the identified model is used to design predictive controller. The derivation of the algorithm and the compensation method of time-delay and data packet drop are presented. Experiment study over campus-wide network shows the validity of the proposed control scheme.