Highly efficient three-stage maximum likelihood recursive least squares identification method for multiple-input multiple-output systems

Abstract

The objective of this paper is to propose novel identification methods with high computational efficiency for multiple-input multiple-output systems. Through decomposing the system into several subsystems for reducing the dimension and the number of parameters, the identification model is derived. To improve the identification efficiency, the identification model is further divided into three virtual sub-models, and then a three-stage auxiliary model-based maximum likelihood recursive least squares (3S-AM-ML-RLS) algorithm is proposed on basis of the hierarchical identification principle. To show the advantages of the proposed algorithm, an existing algorithm is given for comparison. By comparing the total floating point operations (flops), the proposed 3S-AM-ML-RLS algorithm has higher computational efficiency. Furthermore, simulation results test that the proposed 3S-AM-ML-RLS algorithm has better performance and captures the dynamics of the system well.