Research On The Theory Of Observable Degree Of Nonlinear Systems And Its Application

Air sensor network refers to the tracking target collaborative localization and tracking through the multisensory.The rise of development of information fusion and Kalman filter research make many functions of air cooperation target network(such as target tracking and positioning system,decision-making and evaluation)become possible.Sensor network management is one of the main tasks of the aerial mission network,and it also appeared in the research field of information fusion.Observability is whether the state can be estimated by observation.Observable degree is the possibility of obtaining an estimated state in the field of information fusion.Applying the observable degree to the distribution of air sensor networks can carry out the management and allocation of sensor resources according to different state components.Therefore,the sensor network can be optimized.There are many methods for the analysis of observable degree,but mainly for the linear system.The method of the observability analysis for the nonlinear system is mainly based on the derivative of the Lie.It is a large amount of computation.At present,there is no perfect air sensor allocation algorithm based on nonlinear observable degree.The estimation error covariance is used to design the cost function of sensor allocation management,but the estimation error covariance of the sensor distribution method based on the fusion filtering can get the result only after the sensor allocation adjustment,can not be treated as a preliminary judgment before filtering:(1)Combining the UIF and linear SVD observable degree,the observable degree method of nonlinear system is studied.and the method is compared with that of the original linear system.(2)The nonlinear observable degree analysis method based on SVD decomposition is studied,which dose not depend on the filtering result.By adding the scale adjustment factor,it has the scale transform invariance.(3)The method of sensor network allocation based on the observability of nonlinear systems is studied.The sensor network allocation method dose not depend on the filter results.Compared with the sensor network allocation method based on estimation error covariance,the method is more applicable.