Research On CPS-oriented Fuzzy Temporal Reasoning Method For Complex Event Streams

In the application of Cyber-Physical System(CPS),due to the time granularity difference of distributed systems and the data noise in heterogeneous wireless networks,the event flows often has timing uncertainty.For the scheduling of event flow,existing studies us ually assume that the execution time is known,and only a few studies discuss the event scheduling model when the execution time is inaccurate.However,those are lack of the consideration of the impact of the event itself attributes on the timing.This di ssertation focuses on the impact of CPS event multi-attribute constraints on event flow timing,applying fuzzy theory to establish a CPS uncertain temporal event flow timing solution model,and enhances the dynamic scheduling capability of the model by add ing real-time feedback strategies.This dissertation mainly conducts research through theoretical analysis,mathematical modeling,simulation verification and other research methods.The research content and contributions include:(1)A method of uncertain tem poral event flow sequence reasoning based on multiple attribute decision making(MAD-UTES)is proposed.Aiming at the problem of insufficient evidence for the timing decision of the CPS event flows,this paper uses the D-S evidence theory to establish an ev idence solution and fusion model of the fuzzy end time of the event based on considering multiple constraints.Then,based on the correlation model of D-S evidence theory and Intuitionistic Fuzzy Set(IFS),and the theory of IFS negative time inference,th e IFS probability solution model at the start of the fuzzy is established.Finally,according to the moment score,a time series solution of the uncertain temporal event flow is obtained.When the fuzzy interval is limited to 5,and the number of events does not exceed 600,the scheduling accuracy remains above 85% b y simulation experiment analysis.When the fuzzy interval is limited to below 7,the scheduling accuracy does not decrease by more than 15%.(2)An uncertain temporal event flows dynamic scheduling method based on priority coefficient feedback(PFF-UTEDS)is proposed.Aiming at the problem of the lack of dynamic processing capability of the MAD-UTES model,on the one hand,considering the real-time value benefits brought by the execution of events,the concept and model method of "time execution value coefficient" was proposed.On the other hand,using the grey-DEMATEL method to understand the mutual influence relationship among events in the physical world,the concept and model method of "event feedback coefficient" is proposed.Finally,a "comprehensive priority coefficient" calculation model is established based on the former two.On this basis,the dynamic feedback of the uncertain event flow and the preemptive scheduling strategy of the improved MAD-UTES static model are given.Through simulation experiments,when the number of events is less than 600,the miss rate of event execution can be controlled within 35%.Added the ability of model dynamic scheduling.(3)Finally,taking the intelligent t raffic single-point intersection as an application scenario,considering the traffic flow data of the entrance road and the interconnection between the entrance roads,this dissertation uses the MAD-UTES and PFF-UTEDS model methods to design the time serie s inference model of the intersection phase.On this basis,combined with the traditional traffic signal timing strategy,the fuzzy control of the four-phase traffic signal is realized.The simulation results show that,compared with the Fixed Time Plan(FT P)method,the vehicle passing level is increased by 23.76%,and the peak value is increased by 30% compared with FTP.