A Research For CPS Spatial-temporal Modeling Method And Collaborative Event Detection

Cyber physical system can make objects exchange information between each other through the Internet by using perception,communication and computing technology,we can use cyber physical system to detect and control physical environment.Cyber physical system spatial-temporal modeling can divide cyber physical system into three layers:physical layer,physical-cyber layer and the cyberlayer;two flows:information gathering flow and decision control flow.The hierarchical finite state machine can easily express the three-layers system in a mathematical expression,and the state transition between each layer and state sets,and the conversion relationship between the two flows.In the information gathering flow,sensor nodes monitor the physical environment and generate sensor events,sensor motes transform sensor events into physical events,sink nodes transform physical events into physical-cyber events;In the decision control flow,deccision Node transform the physical-cyber events into cyber events,controller nodes transform cyber events into control events,and pass them to the actors.The lattice-based event modeling method can devide cyber physical system event into three parts:event attributes,the observer information,the occurence time and location and attribute information of the event.And collaboration event detection algorithm can improve the detection precision.Cyber physical system is drived by event,the system performance can be affected by event detection precision.Coordination event detection algorithm can improve the accuracy of event detection,it includes three parts:Firstly sensor motes make collaborate detection,secondly they ask the surrounding sensor motes after finding the abnormal by calculating the similarity of mean vector sequence,at last the credible degree will be got by the voting results.If credibility is bigger than threshold,sensor motes and sink nodes will do cluster analysis of the multi-dimensional events.After receiving the multi-dimensional event data reported by the sensor motes,sink nodes calculate the similarity of the boundary vector sequence,and classify and merges multi-dimensional event data by using the improved K-means algorithm.At last,sink nodes and decision node will together detecting the multi-dimensional event type.Decision node matches event type by using the probability distribution of data attribute.The experiment prove that compared with the traditional single factor modeling methods,multi-layer spatio-temporal modeling method realizes the joint modeling with spatio and temporal attribute,through coordination event detection,can reduce the error detection,improve the detection accuracy,improve the overall model performance effectively.Multi-layers spatio-temporal cyber physical sysytem modeling used in the facility agriculture,can improve the production and management level,adapt to the modern agriculture requirement of high precision and high automation.