State Prediction of Chaotic Time-Series Systems Using Autoregressive Integrated with Adaptive Network-Fuzzy

Abstract

In this research, advanced technology is used to monitoring chaotic time-series signals. The combination of autoregressive with adaptive networkfuzzy algorithms is suggested for chaotic signal prediction. The autoregressive prediction algorithm is recommended for chaotic time-series prediction. This technique is linear, and the modeling prediction accuracy has a limitation. To reduce the root means square (RMS) error of prediction, the order of autoregressive prediction should be increased which is caused to increase the number of parameters and nonlinearity as well. Thus, the combination of autoregressive prediction with an adaptive network-fuzzy algorithm is suggested to reduce the prediction error in chaotic time-series signals. To test the power of the proposed prediction algorithm, the 2nd order proposed method is compared with the 2nd order and 6th order of ARtechnique, and the RMSerror in these three algorithms are 0.0967, 0.4953, and 0.3159, respectively. So far and compared to the classical autoregressive method, the proposed prediction model is efficient for chaotic time-series signals.