Accelerated iterative identification of nonlinear systems based on the momentum approach

Regarding a Wiener nonlinear output-error system that includes a dynamic linear part and a static nonlinear part, there exists a problem of the parameter coupling in the linear part and the nonlinear part. This paper presents a decomposition-based gradient iterative algorithm for the Wiener nonlinear output-error system with the application of the auxiliary model identification idea and in line with the hierarchical identification principle, a decomposition of the system into two subsystems is carried out. Meanwhile, using the momentum approach, the momentum term is added to the correction of the parameter update estimates to speed up the gradient descent search and improve the convergence rate. In comparison to the gradient-based iterative algorithm, the proposed algorithm has a smaller computational burden. Finally, a numerical simulation example is provided to demonstrate the superiority of the proposed algorithm, and an application example of determination of concentration during the crystallization process proves the practicability of the proposed algorithm.