Predictive Control Of Asynchronous Sampling Mass With Multiple Networks-induced Uncertainties

Multi-agent systems have attracted the attention of many researchers due to their distributed sensing,computing,and cooperative control characteristics.They have important application prospects in intelligent robot formation,smart grid optimization control,and multi-satellite system control.MASs only need to obtain the information of their neighbor agents from the host to design a control strategy.Considering the limited communication bandwidth and the multiple network induced uncertainties,a sampling mechanism for efficient data utilization and a reasonable event-triggered conditions are particularly important.This thesis considers the consensus control of multi-agent systems under asynchronous sampling,multi-rate sampling,and adaptive event-triggered mechanisms.In addition,wireless network transmission causes some network-induced uncertainties Occurrence,such as delay,packet loss and network attacks,which makes the system performance worse or even unstable.Therefore,how to design a suitable control mechanism to deal with the asynchronous sampling mechanism and coherent control under the coexistence of multiple network-induced uncertainties becomes an important issue for multi-agent systems.For MASs with eventtriggeed mechanism,asynchronous sampling,and cyber attacks,the following predictive control problems are studied:(1)The problem of predictive control of a agent system under a switching observer based on dual-rate sampling data is studied.Dual-rate sampling refers to the sampling of multiple output signals of a controlled object by sensors with two sampling periods.In order to study The effect of the uncertainty caused by the bilateral network on the system is proposed.A novel predictive control protocol is proposed to compensate the bilateral network induced delay and packet loss.An observer with a fixed switching rate based on double-rate sampling data is designed,where the switching rate is determined by the dualrate sampling moment.Then,the network predictive control system model is modeled as a threshold-dependent delay model,which introduces two related and different event-triggered transmission schemes to determine the output information or control command whether it should be transmitted.Finally,the stability criteria and the design method of the predictive controller are established.(2)The security consensus problem of MASs with cloud predictive control mechanism under compound network attack is studied.The sensor group uses asynchronous sampling,that is,each agent is equipped with its own sensor and the sampling rate is different.Designed a observer based on asynchronously sampled data to deal with the problem of asynchronous time to the cloud network.Assume that the attack occurs in the network transmission channel between the controller and the actuator,and the attack is no longer a simple typical network attack,but a heavy one.A more complex attack generated by the combination of replay and deception attacks.Considering the existence of network-induced delays,packet loss,and network attacks,a predictive controller based on a cloud-based simulated attack mechanism is designed,which not only does the control input inconsistency between the observer and predictive controller be resolved,it also compensates for network-induced delays and compound network attacks in the feedforward and feedback channels.The discrete-time MASs under compound network attacks and asynchronous sampling are modeled as a timedelay system model.The safety consensus criteria of MASs are given.(3)The cloud prediction security control problem of discrete-time MASs based on cyber detector is studied.Since information is transmitted to the cloud through the network,it will inevitably be affected by malicious attack nodes in the process.Such network attacks may lead to The system performance is worse or even unstable.Therefore,the attack detection mechanism based on the 2test is designed to detect the agents after the attack,and the controller design based on the detection makes the multi-agent system stable and eventually achieve consensus.Subsequently,a cloud predictive control protocol was designed to achieve both stability and consensus,and the network delay in the feed forward and feedback channels can be actively compensated.The security consensus criteria of the MASs based on the detector is given.