Research On System Identification Using Tabu Search And Development Of Identification Software

This paper gives a review of system identification. It introduces the concept of system identification, its history, and the research actuality.A software package for system identification is introduced, which is the main part of "Advanced Control and System Identification Platform". This software presents some regular identification algorithms; data preprocess tools, model translations, model comparisons, residue calculation methods and so on. It has been encapsulated to COM component, which makes it easy to maintenance and reuse.This paper introduces a new type of optimize algorithm - Tabu search. Its principle and main characters are presented and the algorithm flow chart is also presented. An enhanced one that is used in a continuous space is presented. And we provide a rule to set the parameter of the algorithm.A new method to solve Hammerstein model using tabu search is also introduced. The conception of searching area is defined and the strategy of choosing neighboring points is given. And many simulation tests are done. Using this method to identify the noise including system is also studied, and the simulation results have shown its validity.Finally, the thesis points out the advantage of using tabu search to solve system identification problem and some problems waiting to be solved in this area.