In practice,the model structure,parameters and time-delay of the actual process may vary simultaneously.However,the general identification methods of the 3 items are performed with separate procedures which is very inconvenient in practical application.In view of the fact that variable selection procedure can ensure a compact model with robust input-output,relation and in order to explore the feasibility of variable selection algorithm for the simultaneous identification of process structure,parameters and time-delay,non-negative garrote(NNG)algorithm is introduced and applied to system identification and the corresponding procedures are presented.The application of NNG variable selection algorithm to the identification of single input single output(SISO)system,multiple input multiple output(MIN1O)system and Wood-Berry tower industry are investigated.The identification accuracy and the time-series variable selection results are analyzed and compared between NNG and ordinary least square(OLS)algorithms.The derived excellent results show that the proposed NNG-based modeling algorithm can be utilized for simultaneous identification of the model structure,parameters and time-delay with high precision.
Over the past few years,nonlinear manifold learning has been widely exploited in data analysis and machine learning.This paper presents a novel manifold learning algorithm,named atlas compatibility transformation(ACT),It solves two problems which correspond to two key points in the manifold definition:how to chart a given manifold and how to align the patches to a global coordinate space based on compatibility.For the first problem,we divide the manifold into maximal linear patch(MLP) based on normal vector field of the manifold.For the second problem,we align patches into an optimal global system by solving a generalized eigenvalue problem.Compared with the traditional method,the ACT could deal with noise datasets and fragment datasets.Moreover,the mappings between high dimensional space and low dimensional space are given.Experiments on both synthetic data and real-world data indicate the effection of the proposed algorithm.
