detectNARMAX: A GRAPHICAL USER INTERFACE FOR STRUCTURE DETECTION OF NARMAX MODELS USING THE BOOTSTRAP METHOD

摘要

This paper describes a MATLAB-based program called detectNARMAX that utilizes the bootstrap method, a numerical procedure for estimating the distribution of statistical parameters, to find the best NARMAX model structure representing the nonlinear behavior of a system using its noisy input-output data. The performance of the bootstrap-based structure detection techniques is demonstrated by using synthetic data generated by simulation as well as real data measured to determine a model structure for the thermal impact on a sensor system.