Research On Networked Linear System Model Predictive Control Algorithm

In recent years,more and more attention has been paid to the constraints of model predictive control and the characteristics of solving optimal control on-line.From the initial single system to multiple systems,from simple linear system to complex nonlinear system,the research of model predictive control has been catching up with the needs of the times.Because of the coming of the networked era,the networked system also came into being,so considering the application of the model predictive control in the networked system is a promising research direction,at the same time,there will be many difficulties and challenges.Therefore,the main contents of this paper are network-induced delay,communication data quantization and self-triggering mechanism.In addition,considering the characteristics of the model's predictive control,some researches are done on random disturbance and switching cost function.First of all,because of the time-varying delay of network communication,the model predictive control algorithm itself has the inherent advantage to deal with the delay problem.In addition,the existing quantization strategy is used to process the network information,and then the purpose of saving network resources is mainly to save network bandwidth.Considering the combination of quantization strategy and model predictive control algorithm is also focused on the characteristics of large amount of computation and large amount of transmission information.Secondly,in order to further save the network resources,the self-triggering mechanism is also combined with our model predictive control algorithm.Considering that the self-triggering mechanism can greatly reduce the number of sensor sampling,therefore,it can not only prolong the service life of the sensor,but also the full utilization of limited network resources can be realized.The next trigger time is calculated by the control algorithm designed in this paper,and when the next trigger time comes,the sensor is driven to work and the new state quantity is sampled at the current time.The state quantity of the current moment is transmitted to the self-triggered mechanism,and the next trigger time is calculated by the control algorithm designed in this paper.Finally,considering the influence of complex environmental parameters in practical engineering design,the strategy of switching cost function is also our key research direction.By designing multiple cost function,and then achieve better control effect.In this paper,consider random perturbations that obey the probability constraint,the perturbations here are not determined quantities,but occur with a specific probability.Compared with the perturbations of a given exact quantity,random perturbations that obey certain probability constraints are more objective and can more simulate disturbances in real production life.