Research Article

A Novel FRBF-Type Model for Nonlinear Time Series Prediction

Table 1

Pseudocode of the proposed method for FRBF-ARX model.

For the selected state vector and the model order in the FRBF-ARX model (4).
%Initialization
Generate a set of initial values of FRBF centers.
% Calculate based on the LSM
calculate the scaling factor using equation (9)
calculate the vector of equation (8)
stack and to form the vectors and for all samples; calculate using equation (10);
calculate the error in equation (11);
calculate the error in equation (12);
% Update the parameters
for iteration = 1: selected iteration
step 1: calculate the gradient of in equation (13) with respect to ;
update based on the calculated gradient;
update based on ;
calculate ;
if is not reduced then change the search direction and go back to Step 1;
and if the change of desired value, then break;
end (for selected iteration)