Font Size: a A A

Fault-tolerance And Disturbance Rejection Of Distributed Model Predictive Control

Posted on:2022-02-28Degree:DoctorType:Dissertation
Country:ChinaCandidate:G N XiaoFull Text:PDF
GTID:1488306527982419Subject:Control Science and Engineering
Abstract/Summary:PDF Full Text Request
Model predictive control(MPC)has been widely used in industrial processes because of its ability to deal with multi-variables and constraints.However,with the increasing complexity of modern industrial processes,the modeling difficulties and computational burden increases,and the systems are geographically fragmented,challenging the centralized control.Therefore,the distributed model predictive control structure is adopted to divide the large-scale system into several subsystems to control separately,The control performance of the whole system is optimized.But then,the complex industrial processes are affected by more and more internal or external uncertain factors.For internal problems,some system devices,such as actuators and sensors,may fail,and also there will be problems in the network communication.For external problems,the unknown noises,disturbances and unknown dynamics from the external systems may affect the system dynamics,especially for the nonlinear systems.In distributed model predictive control,due to the realization of separate control and the coupling between subsystems,these internal and external factors will affect the controlled system reliability,and also provide new potential processing methods because of the special distributed structure.This paper studies the fault-tolerance and disturbance rejection of distributed MPC / economic MPC from the following points.1.The system distributed fault-tolerant model predictive control for intermittent sensor/actuator faults with unknown inputs is considered.Firstly,according to the linear distributed system model decomposition,a distributed state observer with unknown input and the coupling states of distributed system is designed.On this basis,with an existing fault set,combined with fault mode identification,a distributed model predictive control method in the form of fault concealment is proposed.The proposed observer of the has a faster convergence speed,and the controller design does not need the fault random distribution information.2.Aiming at the distributed fault-tolerant model predictive control problem without prior fault set,a distributed cooperative fault-tolerant method is proposed.This method adopts the fault actuator ”plug and play” strategy,and cooperates with the neighbor subsystem to complete the fault tolerance of the fault occurs.For the communication failure,the prediction data is used to make up for the data loss caused by the communication faults.For sensor faults,a large robust extended state observer is designed to estimate the state and sensor faults simultaneously.3.The economic model predictive control for distributed nonlinear systems with unknown bounded disturbances is considered.The differential dissipative constraint condition for the global robust stability of distributed nonlinear systems is presented.The differential dynamic energy storage function is constructed by introducing a robust control contraction metric.This method not only provides a reference point independent energy storage function for nonlinear system global differential dissipativity,but also gives a differential L2 for disturbance rejection.Finally,the global system optimization is realized by distributed optimization algorithm.4.A distributed economic model predictive control method is proposed for nonlinear systems with measurable or predictable disturbances from external systems.To satisfy the differential dissipativity in this situation,this method constrains the average dwell time of piecewise constant disturbances,and constructs a control contraction metric of the switched dynamic system.Finally,the global system dissipativity is guaranteed in the presence of piecewise constant predictable disturbances and prediction errors.5.When the distributed system is running online,due to the change of production requirements,the interconnection topology or the number of subsystems may change,resulting in the change of the overall system equilibrium point and overall dynamics.In this case,the incremental dissipative theory which is independent of the equilibrium point of nonlinear system is proposed.It can not only ignore the change of the equilibrium point,but also has a strong tolerance to the changeable interconnections.Then,from the perspective of the whole system,it analyzes global system dissipativity when the number of subsystems increases or decreases.
Keywords/Search Tags:Distributed model predictive control, Distributed observer, Fault-tolerant control, Disturbance rejection, Dissipativity
PDF Full Text Request
Related items