Identification and Control of Nonlinear System using Kernel-RVFLN
Rakesh Kumar Pattanaik, Mohan Debarchan Mohanty, Mihir Narayan Mohanty · 2022
The contribution of this research work is to propose a new controller model for nonlinear plants. The controller is based on a non-iterative network-based model. The polynomial kernel for the Random vector functional link neural network is varied with the advantage of generalisation performance. The concept of Random weight is implemented in the network between the input layer and the enhancement layer. Accordingly, the performance is verified through the application of two appropriately weighted activation functions, along with a weighted direct link. The direct link forms a connection between the input and output layers through the kernel layer. The hidden mapping function used in RVFLN for selecting the number of hidden nodes can be done more efficiently by implementing the concept of the kernel function. To test how well the proposed model works, a nonlinear benchmark system is used to represent the Single Input, Single Output (SISO) system, and the results are compared to well-known approaches like the original RVFLN, WRFNN, RSONFIN, NFIS-DN, ANFIS, CFLANN-DRLS, and Takagi-Sugano fuzzy with unscented Kalman filter in the result section.