Studies Of NCSs Based On Data Feature Driven

Networked control systems (NCSs) has broken through the constraints of the tra-ditional control systems in which point to point control is usually adopted, kept the pace with the development of Network Age. Due to the appealing strengthens of low cost, easy installation and maintenance and remote control etc., NCSs have been ex-tensively applied to the Internet of things, aviation, navigation, guidance, remote edu-cation, remote health and many other areas. While, with the introduction of commu-nication network, it will inevitably encounter the challenges of network induced delay, data dropout and fault sequence. The research plants of NCSs are often have complex structures, multivariable which are hard to and even cannot directly obtain the specific object model information by mechanism analysis, how to utilize the I/O information of the plants to implement control has become a increasing demand of solving the problems. Data feature driven idea derives from computer science, firstly, data feature driven is able to utilize the I/O data of the controlled system to identification the com-plex models of the system and then use the online and offline of system I/O data, and get knowledge via the processing of data to design the controller, no need for contain-ing the mathematical model of the controlled process information, has become a new method to solve these problems.For complex multivariable control plants of NCSs, predictive control provides a better solution, based on the model of network predictive control method is able to overcome the influence of the network induced delay and data dropouts and subspace prediction method based on data feature driven can overcome the difficulty of establish-ing the system mathematical model.How to take the advantages of methods mentioned above a new predictive control algorithm is proposed. This paper considers, under the influence of network induced delay and data dropouts, the predictive networked control algorithm based on data feature driven subspace approach, and the controller design and stability analysis based on the online experiment platform to verify the ef-fectiveness of the algorithm is proposed. ? Combining the networked predictive control theory with the data feature driven subspace predictive control theory, a novel algorithm is proposed to deal with the plants whose models are hard or impossible to be established and overcome the effects of the networked-induced delay and data dropout.? The control designs about both network-induced and data dropout are consid-ered and discussed, by the way an example is made.? Based on the flexible online NCSlab, a experiment about the data feature driven predictive control is designed and the effectiveness of the algorithm is proved here.In the last, we have a review on this research and some open problems are dis-cussed together.