Strong Tracking State Estimation And Swarm Identification

State estimation and system identification are two important research areas in modern control theory,hi this dissertation,some problems of state estimation and system identification based on strong tracking filter and swarm intelligence are intensively studied and discussed. Several effective strong tracking state estimation and swarm identification methods are proposed. The main results can be summarized as follows:(1) A modified strong tracking Kalman filter (MSTKF) is proposed. By changing the multiple time-varying fading factors of the strong tracking filter,MSTKF switches between Kalman filtering and strong tracking filtering. When Kalman filtering cannot track efficiently the state with abrupt changes,MSTKF switches to strong tracking filtering with varying softening factor. The MSTKF has high estimation accuracy and requires less computing time.(2) The strong tracking extended Kalman filter is extended to a class of nonlinear time-varying stochastic systems with additive combined colored noise. By augmenting the state vector,linearizing the nonlinear augmented state space model and adopting the equivalent measurement equation,the problem of strong tracking extended Kalman filtering of nonlinear systems with additive combined colored noise can be converted into the problem of strong tracking Kalman filtering of linear systems with correlated process and measurement noise. A numerical simulation example is given to show the effectiveness of the method.(3) An adaptive functional observer (AFO) for nonlinear discrete-time systems is proposed. By introducing state transformation,a reduced-order state estimation form is achieved. A slightly modified strong tracking filteringalgorithm is used to estimate the reduced-order state vector,which is then used to estimate the nonlinear state function. Sufficient conditions to ensure local asymptotic convergence of the AFO are established. The AFO is an adaptive observer with strong tracking properties,which can accurately estimate the unknown time-varying parameters besides nonlinear state function.(4) A hybrid particle swarm optimization (HPSO) algorithm is proposed. The HPSO employs local version constriction factor method and global version inertia weight method simultaneously to achieve relatively high performance. To avoid the possible occurring of stagnation phenomenon in the particle swarm optimization algorithm,the re-initialization mechanism based on the global information feedback is introduced in the HPSO. Numerical examples show the effectiveness of the HPSO algorithm.(5) The particle swarm optimization is utilized to identify Hammerstein and MISO Wiener-Hammerstein nonlinear system models. The problems of nonlinear system identification are cast as optimization problems in parameter space,and then the particle swarm optimization algorithm is used to search the parameter space parallel and efficiently in order to find the optimal estimation of the system parameters. Numerical examples show the effectiveness of the method.(6) A method for the identification of time-varying delay systems using hybrid particle swarm optimization is proposed. The basic idea behind the proposed method is that the identification is converted to an on-line optimization of nonlinear functions,and then the hybrid particle swarm optimization algorithm is used to find the optimal estimation of the time-varying parameters. By introducing forgetting factor and using previous information effectively,the suggested identification scheme can accurately estimate the unknown time-delay and possesses a good tracking ability to the variations of the parameter.(7) A moving horizon observer for nonlinear systems based on particle swarm optimization is proposed. The observer is not sensitive to the initial conditions. It is an effective method for estimating the state of the nonlinearsystems.(8) A two-space particle swarm optimization method for solving continuous minimax problem is proposed. Simulation results show the effectiveness of the method. Moreo...