Identification of a Box-Jenkins model for rotary drilling laboratory prototype

Conventionally, analytical modelling is used to analyse the dynamics of complex non-linear processes. This paper presents identification of mathematical models by the black box modelling method for non-linear systems. The non-linear system concerned in this work is a laboratory prototype of a rotary drilling rig. The system concerned is distinguished for its additive non-linearity at the output end. The step by step analysis of the procedures and criteria used to select an accurate model for a non-linear process by the black box identification method is explained. The model identified in the paper is a Box-Jenkins model. The model selection procedure uses least squares method, pole zero plots and residual analysis. Accurate simulation results with less than 0.05% error are obtained. The identified Box-Jenkins model is validated by a twofold validation procedure.