Identification of Countercurrent Rare Earth Extraction Process based on Nonlinear State-space Models

摘要

The system identification problem of rare earth extraction process, which is based on local linear weighted statespace models, is considered in this paper. The parameter is determined by the minimization of output error cost function. In addition, the nonuniqueness of the fully parameterized statespace systems is considered and the parameter update is restricted to the direction which is perpendicular to the tangent space to the manifold of observationally input-output equivalent local linear weighted state-space systems. Moreover, the convergence speed of the proposed algorithm is also given. Finally, real industrial experiment of rare earth extraction process is implemented and the results show the efficiency of the proposed method.