Coupled gradient algorithm for multivariable nonlinear systems

This article deals with the identification problems of multivariable nonlinear systems. A coupled gradient algorithm is developed based on the decomposition technique. Using the Kronecker product, the system is transformed a linear regression model, which is decomposed into several sub-models containing the common parameters. Then the system estimates are estimated by using the coupling identification concept. The coupled gradient algorithm avoids handing with the product terms between the parameters of the linear block and the nonlinear block. A simulated numerical example is employed to validate the effectiveness of the proposed algorithm.