Frequency Domain Identification of Multiple Input Multiple Output Nonlinear Systems

The proposed study introduces a total least squares with structure selection (TLSS) algorithm to identify continuous time differential equation models from generalized frequency response function matrix (GFRFM) of multiple-input multiple-output (MIMO) nonlinear system. The estimation procedure is progressive where the parameters of each degree of nonlinearity of each subsystem is estimated beginning with the estimation of linear terms and then adding higher order nonlinear terms. The algorithm combines the advantages of both the total least squares and orthogonal least squares with structure selection (OLSSS). The error reduction ratio (ERR) feature of OLSSS are exploited to provide an effective way of detecting the correct model structure or which terms to include into the model and the total least squares algorithm provides accurate estimates of the parameters when the data is corrupted with noise. The performance of the algorithm has been compared with the weighted complex orthogonal estimator and has been shown to be superior