Research On Sequential Fusion Estimation Algorithm Of Networked System Based On Event Triggering

With the development and progress of science and technology,networked system has been used in social production and people's daily life.As engineering requirements are getting higher and higher,the scale of the system has become huge and complex,which inevitably leads to a large consumption of network resources.In actual systems,communication bandwidth and sensor energy are limited,so saving resources in system state estimation has become a research hotspot.The event-triggered mechanism effectively saves resources by judging whether the event-triggered condition is met for data transmission.Sequential fusion estimation method has attracted wide attention in multi-sensor information fusion estimation because its small calculation amount and the global optimization.This paper aim at the multi-sensor networked system,considering the noise correlation and random uncertainty,combined with event-triggered mechanism,the problem of sequential fusion estimation is studied based on the projection theory.The main research contents are as follows:For linear discrete systems with uncorrelated noises,the distributed sequential state fusion estimation algorithms are proposed based on the measurement event-triggered mechanism and variance event-triggered mechanism.The measurement event-triggered mechanism is formed by the square of the difference between currently observation and last transmitted observation.When the square of the difference is greater than the given threshold,data transmission is performed,otherwise it is not transmitted.Based on this event-triggered mechanism,a local optimal state filter is proposed.Then,transmit the local estimate to the fusion center for sequential fusion,further propose the distributed sequential state fusion estimation algorithm,and prove the optimality of the algorithms.Moreover,one propose a variance-based event-triggered mechanism,which reduces data transmission by judging whether the fusion estimation accuracy meets the set accuracy requirements in real time,and achieves the purpose of saving communication resources while ensuring the estimation accuracy.Furthermore,a corresponding distributed sequential state fusion estimation algorithm is proposed based on the variance event-triggered,and prove the optimality of the algorithms.For a linear discrete random system where the noises between different sensors are simultaneously correlated,and the observation noises are correlated with the system noise in one step,sequential measurement fusion estimation algorithms based on innovation event-triggered and measurement event-triggered are proposed.The innovation event-triggered mechanism requires the sensor to perform local filtering,and standards the innovation data.If the value is greater than the given threshold,the measurement is transmitted to the fusion center through the network for sequential fusion,otherwise the measurement is not transmitted.Under this event-triggered mechanism,we propose the optimal sequential fusion algorithm,and prove the optimality of the algorithm.In order to reduce the computational burden,we have propose a sub-optimal sequential fusion estimation algorithm,and the stability also be analyzed.For a linear discrete system with one-step auto-correlation,one-step crosscorrelation between different observation noises,and one-step cross-correlation between the observation noises and the system noise,an optimal sequential measurement fusion estimation algorithm based on the measurement event-triggered is proposed.The measurement event-triggered mechanism is formed by the acquired sensor observations,and the datas that meet the trigger condition are fused in the fusion center,and then the sequential fusion estimation algorithm is proposed.Due to the correlation between noises,the fusion estimation algorithm involves the design of sequential fusion filters and predictors of observation noise and system noise.And finally the optimality of the sequential fusion estimation algorithm is proved,that is the event-triggered sequential fusion and the event-triggered centralized fusion have the same estimation accuracy.For networked random uncertain systems,considering random parameter disturbances,fading observations and correlated noises,a sequential fusion estimation algorithm based on variance event-triggered is proposed.First,the system model is converted to the standard model,and then based on the standard model,the variance event-triggered optimal sequential fusion filter in the sense of linear minimum variance is proposed.The variance event-triggered mechanism determines whether the observation datas are transmitted by judging whether the fusion accuracy meets the set accuracy requirements.If the fusion accuracy meets the requirements,then data will no longer transmitted to fusion center,otherwise the data will be transmitted to the fusion center for fusion processing until meets requirement.Finally,the optimality proof of the algorithm is completed,that is the event-triggered sequential fusion and the eventtriggered centralized fusion have the same estimation accuracy.Through simulation research,the validity of the proposed algorithms are verified.