Study On Limited Information Based Nonlinear System And Multivariable System Identification

The classical control methods and the advanced control methods havebeen made great progress after many years development, and those methodshave been applied in industrial production widely. However, an accuratemathematical model is the foundation to design a controller. So the key to asuccessful application of a control strategy is to find an appropriateidentification method. In this paper, the identification process is discussed indepth for the data de-noising, the identification principle, numerical simulation,etc. The main contributions of this paper are as follows.1、The testing signals in identification experiment and the selectedprinciple are introduced. In data pretreatment, emphatically discuss theapplication of data de-noising by wavelet analysis. Introduce the classicalidentification algorithms: the least squares, particle swarm optimization, etc.Moreover, the advantages and the disadvantages are obtained of eachalgorithm by numerical simulation.2、to solve the problem caused by moving average noises in the multivariable systems identification, a hierarchical identification principle anditerative identification principle based method is used to identify theparameters and combining with the accelerated convergence technique, aneffective estimation can be obtained. According to the hierarchicalidentification principle, a multivariable system can be decomposed into twosubsystems, one containing a parameter vector and the other containing aparameter matrix. Then, we can gain the iterative solutions by the iterativeidentification principle. The simulation results indicate that the algorithmworks quite well and the accelerated convergence technique improves theconvergence rate greatly.3、The identification of nonlinear systems is very difficult because of thelack of the nonlinearity. In this paper, a new approach to identify the linearpart and the nonlinear part of a nonlinear system in sequence is proposed byusing the limited information on the nonlinearity. The linear part of the systemis identified firstly based on the limited information of the nonlinearity, suchas symbolic information, monotonic information and then, we construct theinternal signals to identify the structure and the parameters of the nonlinearsystems. Combining with an improved Particle Swarm optimization algorithm,the estimation of the parameters is accurate, and the numerical simulationshows that the identification procedure proposed in this paper is effective.