Applied Research Of Rotary Kiln Temperature Modeling With Subspace Method For System Identification

Accurate system models are foundation for the application of advanced controltechnology. It’s a difficult problem to identify accurately for the complicatedmulti-variable industrial processes. As a new method of system identification,subspace methods have attracted many attentions in the theory and engineering fieldssince1990s. Contrast to tradition system identification methods, subspace method hasmany advantages and more suit for the complicated multi-variable systems. Rotarykiln is a kind of complicated multi-variable industrial production equipment, peopleare lack of recognize of its model, so advanced control technology can not be wellapply to it, which led to a low level automation. In this paper, an applied research ofrotary kiln modeling with subspace identification theory is considered.The main contents of this dissertation are outlined as follows:The basic framework and characteristics of subspace identification method isanalyzed systematically, including the representation of system state-space, thecomposition of subspace data matrix, the fundamental theory and algorithms. spaceprojection theory of subspace identification, three kinds classical algorithms areintroduced and the main steps of subspace identification.Most of current subspace identification methods are based on open-loop systems,and point at the closed-loop systems are few.With the consideration of modeling therotary kiln, study a subspace identification method based on principle componentanalysis, and AIC criterion is introduced to judge the model order, which has betterperformance than SVD methods.In the last it proved to be valid with an example.Using classical subspace identification algorithms and subspace identificationmethod based on principle component analysis to establish the temperature model ofthe rotary kiln. By simulations demonstrate the effect of every algorithm, the resultsshows the validity of subspace identification method based on principle componentanalysis which is used in thermal closed-loop systems.