Self-tuning Fusion Estimation Of The System With Unknown Model Parameters And Missing Observation Rate

With the rapid development of information science,computer science,artificial intelligence,multi-sensor networked systems have received extensive attention.However,due to the complexity of the system model and many uncertain factors in the working environment,the system model is usually uncertain or unknown in parameters.In this paper,the self-tuning fusion estimation problem for multi-sensor networked systems with unknown model parameters,missing measurement rates or fading measurement rates is studied.The main research contents are as follows:The information fusion estimation problem for multi-sensor linear discrete-time stochastic systems with unknown model parameters and missing measurement rates is studied.The phenomena of missing measurements are described by a group of Bernoulli distributed random variables.When model parameters and missing measurement rates are unknown,model parameters and missing measurement rates are identified online based on the recursive extend least squares(RELS)algorithm and correlation functions,respectively.A distributed fusion identifier is presented for unknown model parameters.The corresponding self-tuning filtering algorithms are obtained by substituting the identified model parameters and missing measurement rates into local optimal filters,cross-covariance matrices and distribution fusion filter algorithms.The convergence of the algorithms is proved by using a dynamic variance error system analysis(DVESA)method and a dynamic error system analysis(DESA)method.A simulation example shows the effectiveness of the proposed algorithms.The information fusion estimation problem for multi-sensor linear discrete-time stochastic systems with unknown model parameters and fading measurement rates is considered.When model parameters and fading measurement rates are unknown,a distributed weighted fusion identifier of unknown model parameters is presented based on the RELS algorithm and weighted fusion estimation algorithm.Both the mathematical expectations and variances of random variables which describe the phenomena of fading measurements are identified by using correlation functions.The corresponding self-tuning distributed fusion state filtering algorithms are obtained by substituting the identified model parameters,the mathematical expectations and variances into the optimal distributed fusion state filter.The convergence of the proposed algorithms is proved by using a dynamic error system analysis(DESA)method.A simulation example shows the effectiveness of the proposed algorithms.