Research On Event-triggered Fusion Estimation

In the wireless sensor network,the energy of each sensor is limited,and the energy consumed by measurement,conversion or calculation in the communication process is usually much higher than the consumption of the sensor itself.Therefore,in the case of limited energy,it is necessary to study how to reduce the communication frequency and energy consumption with the basic measurement accuracy.However,event-triggered is an effective way to reduce the frequency of communication.Compared with the traditional time-triggered scheme,the event-triggered scheme has the advantages of reducing the computation cost and transmission consumption.Therefore,based on the event triggering principle,the fusion estimation problem is studied in this paper.The main work is as follows:(1)An event-triggered distributed multi-sensor network framework is proposed for discrete linear and nonlinear systems.The sensor network framework using the event-triggered based on state estimation.In contrast to the measurement judgment scheme,the judgment conditions of the event-triggered sensor network based on innovation contain corresponding parameters,which makes the basis for judgment further improved.And the estimator selects Kalman filter algorithm,CKF algorithm and UKF algorithm respectively.(2)The method of hypothesis testing was adopted,a meaning is assigned to the threshold by using standard normal distribution,t-distribution,and the event-triggered condition is given by the rejection field.(3)Based on the event-triggered non-linear sensor,the issue of event-triggered multi-sensor system fusion is further studied.In the fusion center,The CI fusion method is used to receive the data that triggered the transmission is received,and the untriggered data is forecasted or maintained.The proposed filter can be widely used,such as navigation and guidance,big data analysis,intelligent manufacturing systems and other fields.