Analysis And Research Of CPS Real-time Scheduling Based On Discrete Event Model

Energy structure issues concerning human health and sustainable development.To realize the global optimization of energy production and coordination,which is an important way to meet China's increasing energy demands.Cyber physical system can be a new method to solve the above issues.Further,Grid cyber physical system is a new research way that the power physical system and information space are highly merged and integrated in energy power systems.CPS can be viewed as an embedded feedback system which is interactive application of information technology and control technology.It is the foundation of our country's “two integration”.The computing word of CPS takes advantage of physical environment data,spatial calculation data and communicating network data,aimed to perceive,analysis,feedback and control the hybrid system of physical world automatically,and to implement some system's requirements include rapid iteration,dynamic optimization,real-time analysis,timely interaction,safe,efficient and comprehensive decision-making and accurate implementation.Cyber physical system links the physical space and information space.It needs events to drive system states to transfer,which can optimize the system in real time.Perception of CPS's characteristics needs to analyze each particular case one-by-one at different levels.Oriented the five characteristics of CPS what is opened-connection,drove-heterogeneous integration,virtual reality interactive mapping,safe and efficient precision and autonomous optimization execution,and the five challenges what is Reliability and predictability,delay and packet loss,discrete and continuous analysis,global optimization strategy and adaptive scheduling.It is aimed to the research and application of CPS's real-time scheduling that based on the basic theory of discrete event model,and through theoretical analysis,mathematical proof and modeling analysis for the proposed real-time scheduling research methods,which mainly includes these:(1)A real-time label method is proposed.It is defined for the signal events and made events obtaining a value constraint.It is based on superdensetime and through events' constraints that using the theory of partial order to make the continuous expansive states of the system can be ordered.And it is proved that the operation sequence of real-time process in the condition of value constraints can be guaranteed in non-real-time condition,and there are certain events and states.(2)Based on what has been proposed real-time event's label above,a feed forward and feedback-predictive scheduling model is proposed,and it is the combination of a feed forward model and feedback monitoring model,which used for adjusting the sequences of events in the released queues in real time.(3)The self-information of output events,the information entropy of actors and Qo D are proposed for the function of value evaluation index respectively during the time of actors outputting events.Combining what has been propsed methods above,they evaluate the process of executing the tasks,and the control strategy of CPS tasks scheduling is described.Further,the schedulability of CPS is analyzed,and an actor state automata analytical method is proposed based on finite state machine.In the end,it is in order to understand the accuracy and performance of the analytical method that a classified search strategy of state set for ASA based on decision tree is adopted.(4)Based on the above research results,an adaptive scheduling model for GCPS is adopted,and to prove the measurement way is consistent with the presentation method.Finally,there is a simulation analysis by Ptolemey II.