Research On Networked Systems Based On Predictive Control Algorithms

In recent years,with the rapid development of network technology and computer technology,the research on networked control systems(NCSs)has attracted extensive attention of experts and scholars in the field of control at home and abroad.In NCSs,control signals obtained by a controller and measurement signals collected by a sensor can be transmitted through networks.With the rapid development of networked control technology,multi-agent systems(MASs)are gradually connected by the networks,and networked multi-agent systems(NMASs)come into being.The NCSs studied at present are particular NMASs.The NCSs and the NMASs are called networked system in this paper.Compared with traditional control methods,networked control methods has lots of advantages such as low cost,low power consumption,strong flexibility,simple installation and easy maintenance.However,the introduction of networks brings new problems to the control system,including network-induced delays,packet dropouts,packet disorder,quantification and network security,which brings great challenges to the design and analysis of networked system controller.Aiming at the above network problems,this paper studies some control problems in networked systems based on predictive control method.Main research results are as follows:Firstly,a predictive control problem on NCSs with actuator saturation constraints is studied when both forward and feedback channels have time-varying network-induced delays and packet dropouts.Considering that actuator saturation often exists in actual control systems,a networked predictive control(NPC)algorithm with saturation control signals is proposed to actively compensate for network-induced delays and packet dropouts in forward and feedback channels.Based on the NPC algorithm,stability criterion of the control system is obtained and the estimation of the attraction region under the actuator saturation constraint is given.The effectiveness of the proposed algorithm is verified by a numerical simulation.Then,a predictive control problem on NMASs with quantized signals under denial-of-service(DoS)attacks is studied.An arbitrary region quantizer is studied to obtain effective values of control signals.Considering that the forward channel is attacked by a DoS attacker,the probability of packet dropouts caused by the attacker is analyzed by a game theory.An NPC algorithm is proposed for NMASs to compensate for the network-induced delays and attack-induced packet dropouts actively.Based on a “Zoom” strategy,sufficient conditions are given for the NMASs to achieve consensus.The effectiveness of the proposed algorithm is verified by a numerical simulation example.Finally,a model predictive control(MPC)algorithm is studied based on a multi-rate method for nonlinear NMASs.Considering that agents in the NMASs often have many constraints,the MPC algorithm is used to obtain an optimal control sequence.Considering the large computational complexity of the MPC algorithm in practical applications,a networked multi-agent system control architecture is proposed based on a dual-mode controller.Aiming at the problem of time delays in forward and backward channels,the MPC algorithm based on multi-rate method is proposed to compensate for the time delays actively.Feasibility of the proposed method is analyzed and conditions for the practical stability of the system are given.An example of spring-cart simulation is given to verify the effectiveness of the proposed theory.