Use of Prototypical ARX and ARMAX Prediction Time Series with Practical Application (A Comparative Study)

Abstract

The researchers interested in studying the black box models, which are linking a series of inputs with outputs series athlete models and the two models are: Error equation models and includes a model of self-gradient with external input ARX and model autoregressive and moving averages with external ARMAX inputs, models output error. The model includes directing an error OE and model Box Genghis BJ. This research has singled out only error equation models so research aims to predict the time and chained to determine the rank of the model using some statistical and engineering standards, including the standard Akaike information AIC and the standard error of the Akaike final prediction FPE. This research has included the sequential steps of creating data and Stability chains input and output and determine the rank of the model using some statistical and engineering standards and estimate its parameters. The test model accuracy by testing leftover moral link to the specified model in addition to the moral link test of residuals e (t) and the entrance u (t) . Compared with the prediction models for black box selected using the average absolute error MAE and root mean square error RMSE has been the application on real data, a daily drinking water filter using Ekorh scale water project in the Karkh district of Baghdad for the period 01.01.2015 until 03.31.2015.