The study of dynamic response using ARX model in extraction process
This work presents a model using system identification approach namely as ARX to represent
the dynamic response for essential oil extraction process. A fresh set of data under feed in
disturbance was collected using MATLAB Simulink. The 3000 samples of data was collected
by using PRBS as an input and temperature in oC as an output. The collected data was
separated into two groups; training data and estimation data by using interlacing technique. The model estimation was done by using linear regression method. The robustness of the model was evaluated by using best fit (R2), OSA, root mean square error (RMSE), correlation analysis and residual analysis (histogram). Based on validation results, the ARX model was successfully capturing the dynamic response of extraction process by provide the high best fit, low RMSE error and normally distributed by producing small mean and variance.
Keywords: auto-regressive with exogenous input (ARX); pseudo-random binary (PRBS); interlacing technique; prediction method; one-step ahead prediction (OSA).