Research On State Estimation Of Networked Nonlinear Systems Based On Extended State Observer

Since the 21 st century,with the development of embedded technology,camera technology and sensor technology,it has promoted the emergence and application of all kinds of robots,UAVs and unmanned vehicles.Whether it is the joint control of robot,the pose control of UAV,or the path planning of unmanned vehicle,the state of dynamic system needs to be estimated.In this paper,based on extended state observer,for the problems of disturbance,time delay and packet loss in nonlinear systems,state estimation is discussed for two kinds of networked nonlinear systems with disturbance:(1)Aiming at the problems of nonlinear uncertain time-delay,bounded output time-delay and missing input signal in the system,an augmented system and networked input signal transmission mechanism are designed to process the system state,the measured nonlinear time-delay data and the input signal with packet loss.In the state estimation,firstly,the extended state estimator is designed to predict the state of the target system,the augmented error estimation system is designed to obtain the prediction error information of the designed estimator,and the multi-channel network information transmission mechanism is designed to ensure the stability and controllability of the system;Then,based on the prediction error information,the Riccati inequality of the target state optimal one-step update error covariance and its corresponding optimal estimator gain are designed.According to the optimal gain,the target system state data is modified combined with the measurement data;Finally,the stability of the designed estimator is analyzed by using the Shure complement theorem.This kind of state estimation algorithm can not only deal with the time-delay data in the system,but also use the extended state estimator to ensure the accuracy of state estimation and optimize the controllability of the system.(2)Aiming at the problem of missing and bounded nonlinear measurement data in distributed sensor networks,a distributed state estimation system is designed.A multi-sensor local estimator is designed based on extended state Kalman filter,which collects and stores the measurement data of all the sensors around it.When the sensor measures the target state data,the sensor transmits the measurement data to the local estimator,and the local estimator allocates the storage space to store the sensor data;When the sensor does not measure the target state data,the sensor no longer transmits the measurement data to the local estimator,which allocates the storage space and stores the 0 data.Then,the local estimator updates the state estimation based on the extended state Kalman filter according to the collected sensor measurement data set.Finally,the fusion estimator is optimized according to the data of multiple local estimators to obtain the target system state information accurately.Finally,the Matlab tool is used to model and simulate the two kinds of systems,test the state tracking effect of the two state estimation algorithms for the two kinds of target systems,and further analyze the convergence of the error.