A Novel Low-dimensional Modeling Method for Control of Unknown Nonlinear Distributed Spatial Processes 

Xiangping CAO, Yan CHEN Hunan 

Technical College of Railway High-Speed, Hengyang, CHINA 

Abstract: Data-based low-dimensional modeling for control design of nonlinear distributed spatial processes is necessary because there are usually some unknown uncertainties in first-principles modeling. In this paper, a novel low-dimensional modeling method is proposed for nonlinear distributed spatial processes. New discrete basis functions are generated according to linear combination of empirical eigen functions from empirical orthogonal function analysis(EOF). A low-dimensional model is identified by traditional identification techniques for the corresponding temporal dynamics. Thus, the nonlinear spatio-temporal dynamics of unknown distributed spatial processes can be reconstructed by synthesizing new discrete basis functions and the obtained low-dimensional model. The numerical simulations show that the proposed method has evidently better performance than that empirical eigen functions based modeling. 

Keywords: Nonlinear Distributed Spatial Processes;